Re: [Gluster-users] Gluter 3.12.12: performance during heal and in general
Hi Pranith, i recently upgraded to version 3.12.14, still no change in load/performance. Have you received any feedback? At the moment i have 3 options: - problem can be fixed within version 3.12 - upgrade to 4.1 and magically/hopefully "fix" the problem (might not help when problem is within brick) - replace glusterfs with $whatever (defeat... :-( ) thx Hubert 2018-09-03 7:55 GMT+02:00 Pranith Kumar Karampuri : > > > On Fri, Aug 31, 2018 at 1:18 PM Hu Bert wrote: >> >> Hi Pranith, >> >> i just wanted to ask if you were able to get any feedback from your >> colleagues :-) > > > Sorry, I didn't get a chance to. I am working on a customer issue which is > taking away cycles from any other work. Let me get back to you once I get > time this week. > >> >> >> btw.: we migrated some stuff (static resources, small files) to a nfs >> server that we actually wanted to replace by glusterfs. Load and cpu >> usage has gone down a bit, but still is asymmetric on the 3 gluster >> servers. >> >> >> 2018-08-28 9:24 GMT+02:00 Hu Bert : >> > Hm, i noticed that in the shared.log (volume log file) on gluster11 >> > and gluster12 (but not on gluster13) i now see these warnings: >> > >> > [2018-08-28 07:18:57.224367] W [MSGID: 109011] >> > [dht-layout.c:186:dht_layout_search] 0-shared-dht: no subvolume for >> > hash (value) = 3054593291 >> > [2018-08-28 07:19:17.733625] W [MSGID: 109011] >> > [dht-layout.c:186:dht_layout_search] 0-shared-dht: no subvolume for >> > hash (value) = 2595205890 >> > [2018-08-28 07:19:27.950355] W [MSGID: 109011] >> > [dht-layout.c:186:dht_layout_search] 0-shared-dht: no subvolume for >> > hash (value) = 3105728076 >> > [2018-08-28 07:19:42.519010] W [MSGID: 109011] >> > [dht-layout.c:186:dht_layout_search] 0-shared-dht: no subvolume for >> > hash (value) = 3740415196 >> > [2018-08-28 07:19:48.194774] W [MSGID: 109011] >> > [dht-layout.c:186:dht_layout_search] 0-shared-dht: no subvolume for >> > hash (value) = 2922795043 >> > [2018-08-28 07:19:52.506135] W [MSGID: 109011] >> > [dht-layout.c:186:dht_layout_search] 0-shared-dht: no subvolume for >> > hash (value) = 2841655539 >> > [2018-08-28 07:19:55.466352] W [MSGID: 109011] >> > [dht-layout.c:186:dht_layout_search] 0-shared-dht: no subvolume for >> > hash (value) = 3049465001 >> > >> > Don't know if that could be related. >> > >> > >> > 2018-08-28 8:54 GMT+02:00 Hu Bert : >> >> a little update after about 2 hours of uptime: still/again high cpu >> >> usage by one brick processes. server load >30. >> >> >> >> gluster11: high cpu; brick /gluster/bricksdd1/; no hdd exchange so far >> >> gluster12: normal cpu; brick /gluster/bricksdd1_new/; hdd change >> >> /dev/sdd >> >> gluster13: high cpu; brick /gluster/bricksdd1_new/; hdd change /dev/sdd >> >> >> >> The process for brick bricksdd1 consumes almost all 12 cores. >> >> Interestingly there are more threads for the bricksdd1 process than >> >> for the other bricks. Counted with "ps huH p | wc >> >> -l" >> >> >> >> gluster11: >> >> bricksda1 59 threads, bricksdb1 65 threads, bricksdc1 68 threads, >> >> bricksdd1 85 threads >> >> gluster12: >> >> bricksda1 65 threads, bricksdb1 60 threads, bricksdc1 61 threads, >> >> bricksdd1_new 58 threads >> >> gluster13: >> >> bricksda1 61 threads, bricksdb1 60 threads, bricksdc1 61 threads, >> >> bricksdd1_new 82 threads >> >> >> >> Don't know if that could be relevant. >> >> >> >> 2018-08-28 7:04 GMT+02:00 Hu Bert : >> >>> Good Morning, >> >>> >> >>> today i update + rebooted all gluster servers, kernel update to >> >>> 4.9.0-8 and gluster to 3.12.13. Reboots went fine, but on one of the >> >>> gluster servers (gluster13) one of the bricks did come up at the >> >>> beginning but then lost connection. >> >>> >> >>> OK: >> >>> >> >>> Status of volume: shared >> >>> Gluster process TCP Port RDMA Port >> >>> Online Pid >> >>> >> >>> -- >> >>> [...] >> >>> Brick gluster11:/gluster/bricksdd1/shared 49155 0 >> >>> Y 2506 >> >>> Brick gluster12:/gluster/bricksdd1_new/shared49155 0 >> >>> Y 2097 >> >>> Brick gluster13:/gluster/bricksdd1_new/shared49155 0 >> >>> Y 2136 >> >>> >> >>> Lost connection: >> >>> >> >>> Brick gluster11:/gluster/bricksdd1/shared 49155 0 >> >>> Y 2506 >> >>> Brick gluster12:/gluster/bricksdd1_new/shared 49155 0 >> >>> Y 2097 >> >>> Brick gluster13:/gluster/bricksdd1_new/shared N/A N/A >> >>> N N/A >> >>> >> >>> gluster volume heal shared info: >> >>> Brick gluster13:/gluster/bricksdd1_new/shared >> >>> Status: Transport endpoint is not connected >> >>> Number of entries: - >> >>> >> >>> reboot was at 06:15:39; brick then worked for a short period, but then >> >>> somehow disconnected. >> >>> >> >>> from gluster13:/var/log/glusterfs/glusterd.log: >> >>> >> >>> [2018-08-28 04:27:36.944608] I [MSGID: 106005] >> >>>
Re: [Gluster-users] Gluter 3.12.12: performance during heal and in general
On Fri, Aug 31, 2018 at 1:18 PM Hu Bert wrote: > Hi Pranith, > > i just wanted to ask if you were able to get any feedback from your > colleagues :-) > Sorry, I didn't get a chance to. I am working on a customer issue which is taking away cycles from any other work. Let me get back to you once I get time this week. > > btw.: we migrated some stuff (static resources, small files) to a nfs > server that we actually wanted to replace by glusterfs. Load and cpu > usage has gone down a bit, but still is asymmetric on the 3 gluster > servers. > > > 2018-08-28 9:24 GMT+02:00 Hu Bert : > > Hm, i noticed that in the shared.log (volume log file) on gluster11 > > and gluster12 (but not on gluster13) i now see these warnings: > > > > [2018-08-28 07:18:57.224367] W [MSGID: 109011] > > [dht-layout.c:186:dht_layout_search] 0-shared-dht: no subvolume for > > hash (value) = 3054593291 > > [2018-08-28 07:19:17.733625] W [MSGID: 109011] > > [dht-layout.c:186:dht_layout_search] 0-shared-dht: no subvolume for > > hash (value) = 2595205890 > > [2018-08-28 07:19:27.950355] W [MSGID: 109011] > > [dht-layout.c:186:dht_layout_search] 0-shared-dht: no subvolume for > > hash (value) = 3105728076 > > [2018-08-28 07:19:42.519010] W [MSGID: 109011] > > [dht-layout.c:186:dht_layout_search] 0-shared-dht: no subvolume for > > hash (value) = 3740415196 > > [2018-08-28 07:19:48.194774] W [MSGID: 109011] > > [dht-layout.c:186:dht_layout_search] 0-shared-dht: no subvolume for > > hash (value) = 2922795043 > > [2018-08-28 07:19:52.506135] W [MSGID: 109011] > > [dht-layout.c:186:dht_layout_search] 0-shared-dht: no subvolume for > > hash (value) = 2841655539 > > [2018-08-28 07:19:55.466352] W [MSGID: 109011] > > [dht-layout.c:186:dht_layout_search] 0-shared-dht: no subvolume for > > hash (value) = 3049465001 > > > > Don't know if that could be related. > > > > > > 2018-08-28 8:54 GMT+02:00 Hu Bert : > >> a little update after about 2 hours of uptime: still/again high cpu > >> usage by one brick processes. server load >30. > >> > >> gluster11: high cpu; brick /gluster/bricksdd1/; no hdd exchange so far > >> gluster12: normal cpu; brick /gluster/bricksdd1_new/; hdd change > /dev/sdd > >> gluster13: high cpu; brick /gluster/bricksdd1_new/; hdd change /dev/sdd > >> > >> The process for brick bricksdd1 consumes almost all 12 cores. > >> Interestingly there are more threads for the bricksdd1 process than > >> for the other bricks. Counted with "ps huH p | wc > >> -l" > >> > >> gluster11: > >> bricksda1 59 threads, bricksdb1 65 threads, bricksdc1 68 threads, > >> bricksdd1 85 threads > >> gluster12: > >> bricksda1 65 threads, bricksdb1 60 threads, bricksdc1 61 threads, > >> bricksdd1_new 58 threads > >> gluster13: > >> bricksda1 61 threads, bricksdb1 60 threads, bricksdc1 61 threads, > >> bricksdd1_new 82 threads > >> > >> Don't know if that could be relevant. > >> > >> 2018-08-28 7:04 GMT+02:00 Hu Bert : > >>> Good Morning, > >>> > >>> today i update + rebooted all gluster servers, kernel update to > >>> 4.9.0-8 and gluster to 3.12.13. Reboots went fine, but on one of the > >>> gluster servers (gluster13) one of the bricks did come up at the > >>> beginning but then lost connection. > >>> > >>> OK: > >>> > >>> Status of volume: shared > >>> Gluster process TCP Port RDMA Port > Online Pid > >>> > -- > >>> [...] > >>> Brick gluster11:/gluster/bricksdd1/shared 49155 0 > >>> Y 2506 > >>> Brick gluster12:/gluster/bricksdd1_new/shared49155 0 > >>> Y 2097 > >>> Brick gluster13:/gluster/bricksdd1_new/shared49155 0 > >>> Y 2136 > >>> > >>> Lost connection: > >>> > >>> Brick gluster11:/gluster/bricksdd1/shared 49155 0 > >>> Y 2506 > >>> Brick gluster12:/gluster/bricksdd1_new/shared 49155 0 > >>> Y 2097 > >>> Brick gluster13:/gluster/bricksdd1_new/shared N/A N/A > >>> N N/A > >>> > >>> gluster volume heal shared info: > >>> Brick gluster13:/gluster/bricksdd1_new/shared > >>> Status: Transport endpoint is not connected > >>> Number of entries: - > >>> > >>> reboot was at 06:15:39; brick then worked for a short period, but then > >>> somehow disconnected. > >>> > >>> from gluster13:/var/log/glusterfs/glusterd.log: > >>> > >>> [2018-08-28 04:27:36.944608] I [MSGID: 106005] > >>> [glusterd-handler.c:6071:__glusterd_brick_rpc_notify] 0-management: > >>> Brick gluster13:/gluster/bricksdd1_new/shared has disconnected from > >>> glusterd. > >>> [2018-08-28 04:28:57.869666] I > >>> [glusterd-utils.c:6056:glusterd_brick_start] 0-management: starting a > >>> fresh brick process for brick /gluster/bricksdd1_new/shared > >>> [2018-08-28 04:35:20.732666] I [MSGID: 106143] > >>> [glusterd-pmap.c:295:pmap_registry_bind] 0-pmap: adding brick > >>> /gluster/bricksdd1_new/shared on port 49157 > >>> > >>> After 'gluster volume start shared force' (then with new
Re: [Gluster-users] Gluter 3.12.12: performance during heal and in general
Hi Pranith, i just wanted to ask if you were able to get any feedback from your colleagues :-) btw.: we migrated some stuff (static resources, small files) to a nfs server that we actually wanted to replace by glusterfs. Load and cpu usage has gone down a bit, but still is asymmetric on the 3 gluster servers. 2018-08-28 9:24 GMT+02:00 Hu Bert : > Hm, i noticed that in the shared.log (volume log file) on gluster11 > and gluster12 (but not on gluster13) i now see these warnings: > > [2018-08-28 07:18:57.224367] W [MSGID: 109011] > [dht-layout.c:186:dht_layout_search] 0-shared-dht: no subvolume for > hash (value) = 3054593291 > [2018-08-28 07:19:17.733625] W [MSGID: 109011] > [dht-layout.c:186:dht_layout_search] 0-shared-dht: no subvolume for > hash (value) = 2595205890 > [2018-08-28 07:19:27.950355] W [MSGID: 109011] > [dht-layout.c:186:dht_layout_search] 0-shared-dht: no subvolume for > hash (value) = 3105728076 > [2018-08-28 07:19:42.519010] W [MSGID: 109011] > [dht-layout.c:186:dht_layout_search] 0-shared-dht: no subvolume for > hash (value) = 3740415196 > [2018-08-28 07:19:48.194774] W [MSGID: 109011] > [dht-layout.c:186:dht_layout_search] 0-shared-dht: no subvolume for > hash (value) = 2922795043 > [2018-08-28 07:19:52.506135] W [MSGID: 109011] > [dht-layout.c:186:dht_layout_search] 0-shared-dht: no subvolume for > hash (value) = 2841655539 > [2018-08-28 07:19:55.466352] W [MSGID: 109011] > [dht-layout.c:186:dht_layout_search] 0-shared-dht: no subvolume for > hash (value) = 3049465001 > > Don't know if that could be related. > > > 2018-08-28 8:54 GMT+02:00 Hu Bert : >> a little update after about 2 hours of uptime: still/again high cpu >> usage by one brick processes. server load >30. >> >> gluster11: high cpu; brick /gluster/bricksdd1/; no hdd exchange so far >> gluster12: normal cpu; brick /gluster/bricksdd1_new/; hdd change /dev/sdd >> gluster13: high cpu; brick /gluster/bricksdd1_new/; hdd change /dev/sdd >> >> The process for brick bricksdd1 consumes almost all 12 cores. >> Interestingly there are more threads for the bricksdd1 process than >> for the other bricks. Counted with "ps huH p | wc >> -l" >> >> gluster11: >> bricksda1 59 threads, bricksdb1 65 threads, bricksdc1 68 threads, >> bricksdd1 85 threads >> gluster12: >> bricksda1 65 threads, bricksdb1 60 threads, bricksdc1 61 threads, >> bricksdd1_new 58 threads >> gluster13: >> bricksda1 61 threads, bricksdb1 60 threads, bricksdc1 61 threads, >> bricksdd1_new 82 threads >> >> Don't know if that could be relevant. >> >> 2018-08-28 7:04 GMT+02:00 Hu Bert : >>> Good Morning, >>> >>> today i update + rebooted all gluster servers, kernel update to >>> 4.9.0-8 and gluster to 3.12.13. Reboots went fine, but on one of the >>> gluster servers (gluster13) one of the bricks did come up at the >>> beginning but then lost connection. >>> >>> OK: >>> >>> Status of volume: shared >>> Gluster process TCP Port RDMA Port Online Pid >>> -- >>> [...] >>> Brick gluster11:/gluster/bricksdd1/shared 49155 0 >>> Y 2506 >>> Brick gluster12:/gluster/bricksdd1_new/shared49155 0 >>> Y 2097 >>> Brick gluster13:/gluster/bricksdd1_new/shared49155 0 >>> Y 2136 >>> >>> Lost connection: >>> >>> Brick gluster11:/gluster/bricksdd1/shared 49155 0 >>> Y 2506 >>> Brick gluster12:/gluster/bricksdd1_new/shared 49155 0 >>> Y 2097 >>> Brick gluster13:/gluster/bricksdd1_new/shared N/A N/A >>> N N/A >>> >>> gluster volume heal shared info: >>> Brick gluster13:/gluster/bricksdd1_new/shared >>> Status: Transport endpoint is not connected >>> Number of entries: - >>> >>> reboot was at 06:15:39; brick then worked for a short period, but then >>> somehow disconnected. >>> >>> from gluster13:/var/log/glusterfs/glusterd.log: >>> >>> [2018-08-28 04:27:36.944608] I [MSGID: 106005] >>> [glusterd-handler.c:6071:__glusterd_brick_rpc_notify] 0-management: >>> Brick gluster13:/gluster/bricksdd1_new/shared has disconnected from >>> glusterd. >>> [2018-08-28 04:28:57.869666] I >>> [glusterd-utils.c:6056:glusterd_brick_start] 0-management: starting a >>> fresh brick process for brick /gluster/bricksdd1_new/shared >>> [2018-08-28 04:35:20.732666] I [MSGID: 106143] >>> [glusterd-pmap.c:295:pmap_registry_bind] 0-pmap: adding brick >>> /gluster/bricksdd1_new/shared on port 49157 >>> >>> After 'gluster volume start shared force' (then with new port 49157): >>> >>> Brick gluster11:/gluster/bricksdd1/shared 49155 0 >>> Y 2506 >>> Brick gluster12:/gluster/bricksdd1_new/shared 49155 0 >>> Y 2097 >>> Brick gluster13:/gluster/bricksdd1_new/shared 49157 0 >>> Y 3994 >>> >>> from /var/log/syslog: >>> >>> Aug 28 06:27:36 gluster13 gluster-bricksdd1_new-shared[2136]: pending >>> frames: >>> Aug 28 06:27:36 gluster13
Re: [Gluster-users] Gluter 3.12.12: performance during heal and in general
Hm, i noticed that in the shared.log (volume log file) on gluster11 and gluster12 (but not on gluster13) i now see these warnings: [2018-08-28 07:18:57.224367] W [MSGID: 109011] [dht-layout.c:186:dht_layout_search] 0-shared-dht: no subvolume for hash (value) = 3054593291 [2018-08-28 07:19:17.733625] W [MSGID: 109011] [dht-layout.c:186:dht_layout_search] 0-shared-dht: no subvolume for hash (value) = 2595205890 [2018-08-28 07:19:27.950355] W [MSGID: 109011] [dht-layout.c:186:dht_layout_search] 0-shared-dht: no subvolume for hash (value) = 3105728076 [2018-08-28 07:19:42.519010] W [MSGID: 109011] [dht-layout.c:186:dht_layout_search] 0-shared-dht: no subvolume for hash (value) = 3740415196 [2018-08-28 07:19:48.194774] W [MSGID: 109011] [dht-layout.c:186:dht_layout_search] 0-shared-dht: no subvolume for hash (value) = 2922795043 [2018-08-28 07:19:52.506135] W [MSGID: 109011] [dht-layout.c:186:dht_layout_search] 0-shared-dht: no subvolume for hash (value) = 2841655539 [2018-08-28 07:19:55.466352] W [MSGID: 109011] [dht-layout.c:186:dht_layout_search] 0-shared-dht: no subvolume for hash (value) = 3049465001 Don't know if that could be related. 2018-08-28 8:54 GMT+02:00 Hu Bert : > a little update after about 2 hours of uptime: still/again high cpu > usage by one brick processes. server load >30. > > gluster11: high cpu; brick /gluster/bricksdd1/; no hdd exchange so far > gluster12: normal cpu; brick /gluster/bricksdd1_new/; hdd change /dev/sdd > gluster13: high cpu; brick /gluster/bricksdd1_new/; hdd change /dev/sdd > > The process for brick bricksdd1 consumes almost all 12 cores. > Interestingly there are more threads for the bricksdd1 process than > for the other bricks. Counted with "ps huH p | wc > -l" > > gluster11: > bricksda1 59 threads, bricksdb1 65 threads, bricksdc1 68 threads, > bricksdd1 85 threads > gluster12: > bricksda1 65 threads, bricksdb1 60 threads, bricksdc1 61 threads, > bricksdd1_new 58 threads > gluster13: > bricksda1 61 threads, bricksdb1 60 threads, bricksdc1 61 threads, > bricksdd1_new 82 threads > > Don't know if that could be relevant. > > 2018-08-28 7:04 GMT+02:00 Hu Bert : >> Good Morning, >> >> today i update + rebooted all gluster servers, kernel update to >> 4.9.0-8 and gluster to 3.12.13. Reboots went fine, but on one of the >> gluster servers (gluster13) one of the bricks did come up at the >> beginning but then lost connection. >> >> OK: >> >> Status of volume: shared >> Gluster process TCP Port RDMA Port Online Pid >> -- >> [...] >> Brick gluster11:/gluster/bricksdd1/shared 49155 0 >> Y 2506 >> Brick gluster12:/gluster/bricksdd1_new/shared49155 0 >> Y 2097 >> Brick gluster13:/gluster/bricksdd1_new/shared49155 0 >> Y 2136 >> >> Lost connection: >> >> Brick gluster11:/gluster/bricksdd1/shared 49155 0 >> Y 2506 >> Brick gluster12:/gluster/bricksdd1_new/shared 49155 0 >> Y 2097 >> Brick gluster13:/gluster/bricksdd1_new/shared N/A N/A >> N N/A >> >> gluster volume heal shared info: >> Brick gluster13:/gluster/bricksdd1_new/shared >> Status: Transport endpoint is not connected >> Number of entries: - >> >> reboot was at 06:15:39; brick then worked for a short period, but then >> somehow disconnected. >> >> from gluster13:/var/log/glusterfs/glusterd.log: >> >> [2018-08-28 04:27:36.944608] I [MSGID: 106005] >> [glusterd-handler.c:6071:__glusterd_brick_rpc_notify] 0-management: >> Brick gluster13:/gluster/bricksdd1_new/shared has disconnected from >> glusterd. >> [2018-08-28 04:28:57.869666] I >> [glusterd-utils.c:6056:glusterd_brick_start] 0-management: starting a >> fresh brick process for brick /gluster/bricksdd1_new/shared >> [2018-08-28 04:35:20.732666] I [MSGID: 106143] >> [glusterd-pmap.c:295:pmap_registry_bind] 0-pmap: adding brick >> /gluster/bricksdd1_new/shared on port 49157 >> >> After 'gluster volume start shared force' (then with new port 49157): >> >> Brick gluster11:/gluster/bricksdd1/shared 49155 0 >> Y 2506 >> Brick gluster12:/gluster/bricksdd1_new/shared 49155 0 >> Y 2097 >> Brick gluster13:/gluster/bricksdd1_new/shared 49157 0 >> Y 3994 >> >> from /var/log/syslog: >> >> Aug 28 06:27:36 gluster13 gluster-bricksdd1_new-shared[2136]: pending frames: >> Aug 28 06:27:36 gluster13 gluster-bricksdd1_new-shared[2136]: frame : >> type(0) op(0) >> Aug 28 06:27:36 gluster13 gluster-bricksdd1_new-shared[2136]: frame : >> type(0) op(0) >> Aug 28 06:27:36 gluster13 gluster-bricksdd1_new-shared[2136]: >> patchset: git://git.gluster.org/glusterfs.git >> Aug 28 06:27:36 gluster13 gluster-bricksdd1_new-shared[2136]: signal >> received: 11 >> Aug 28 06:27:36 gluster13 gluster-bricksdd1_new-shared[2136]: time of crash: >> Aug 28 06:27:36 gluster13 gluster-bricksdd1_new-shared[2136]: >> 2018-08-28
Re: [Gluster-users] Gluter 3.12.12: performance during heal and in general
a little update after about 2 hours of uptime: still/again high cpu usage by one brick processes. server load >30. gluster11: high cpu; brick /gluster/bricksdd1/; no hdd exchange so far gluster12: normal cpu; brick /gluster/bricksdd1_new/; hdd change /dev/sdd gluster13: high cpu; brick /gluster/bricksdd1_new/; hdd change /dev/sdd The process for brick bricksdd1 consumes almost all 12 cores. Interestingly there are more threads for the bricksdd1 process than for the other bricks. Counted with "ps huH p | wc -l" gluster11: bricksda1 59 threads, bricksdb1 65 threads, bricksdc1 68 threads, bricksdd1 85 threads gluster12: bricksda1 65 threads, bricksdb1 60 threads, bricksdc1 61 threads, bricksdd1_new 58 threads gluster13: bricksda1 61 threads, bricksdb1 60 threads, bricksdc1 61 threads, bricksdd1_new 82 threads Don't know if that could be relevant. 2018-08-28 7:04 GMT+02:00 Hu Bert : > Good Morning, > > today i update + rebooted all gluster servers, kernel update to > 4.9.0-8 and gluster to 3.12.13. Reboots went fine, but on one of the > gluster servers (gluster13) one of the bricks did come up at the > beginning but then lost connection. > > OK: > > Status of volume: shared > Gluster process TCP Port RDMA Port Online Pid > -- > [...] > Brick gluster11:/gluster/bricksdd1/shared 49155 0 > Y 2506 > Brick gluster12:/gluster/bricksdd1_new/shared49155 0 > Y 2097 > Brick gluster13:/gluster/bricksdd1_new/shared49155 0 > Y 2136 > > Lost connection: > > Brick gluster11:/gluster/bricksdd1/shared 49155 0 > Y 2506 > Brick gluster12:/gluster/bricksdd1_new/shared 49155 0 > Y 2097 > Brick gluster13:/gluster/bricksdd1_new/shared N/A N/A > N N/A > > gluster volume heal shared info: > Brick gluster13:/gluster/bricksdd1_new/shared > Status: Transport endpoint is not connected > Number of entries: - > > reboot was at 06:15:39; brick then worked for a short period, but then > somehow disconnected. > > from gluster13:/var/log/glusterfs/glusterd.log: > > [2018-08-28 04:27:36.944608] I [MSGID: 106005] > [glusterd-handler.c:6071:__glusterd_brick_rpc_notify] 0-management: > Brick gluster13:/gluster/bricksdd1_new/shared has disconnected from > glusterd. > [2018-08-28 04:28:57.869666] I > [glusterd-utils.c:6056:glusterd_brick_start] 0-management: starting a > fresh brick process for brick /gluster/bricksdd1_new/shared > [2018-08-28 04:35:20.732666] I [MSGID: 106143] > [glusterd-pmap.c:295:pmap_registry_bind] 0-pmap: adding brick > /gluster/bricksdd1_new/shared on port 49157 > > After 'gluster volume start shared force' (then with new port 49157): > > Brick gluster11:/gluster/bricksdd1/shared 49155 0 > Y 2506 > Brick gluster12:/gluster/bricksdd1_new/shared 49155 0 > Y 2097 > Brick gluster13:/gluster/bricksdd1_new/shared 49157 0 > Y 3994 > > from /var/log/syslog: > > Aug 28 06:27:36 gluster13 gluster-bricksdd1_new-shared[2136]: pending frames: > Aug 28 06:27:36 gluster13 gluster-bricksdd1_new-shared[2136]: frame : > type(0) op(0) > Aug 28 06:27:36 gluster13 gluster-bricksdd1_new-shared[2136]: frame : > type(0) op(0) > Aug 28 06:27:36 gluster13 gluster-bricksdd1_new-shared[2136]: > patchset: git://git.gluster.org/glusterfs.git > Aug 28 06:27:36 gluster13 gluster-bricksdd1_new-shared[2136]: signal > received: 11 > Aug 28 06:27:36 gluster13 gluster-bricksdd1_new-shared[2136]: time of crash: > Aug 28 06:27:36 gluster13 gluster-bricksdd1_new-shared[2136]: > 2018-08-28 04:27:36 > Aug 28 06:27:36 gluster13 gluster-bricksdd1_new-shared[2136]: > configuration details: > Aug 28 06:27:36 gluster13 gluster-bricksdd1_new-shared[2136]: argp 1 > Aug 28 06:27:36 gluster13 gluster-bricksdd1_new-shared[2136]: backtrace 1 > Aug 28 06:27:36 gluster13 gluster-bricksdd1_new-shared[2136]: dlfcn 1 > Aug 28 06:27:36 gluster13 gluster-bricksdd1_new-shared[2136]: libpthread 1 > Aug 28 06:27:36 gluster13 gluster-bricksdd1_new-shared[2136]: llistxattr 1 > Aug 28 06:27:36 gluster13 gluster-bricksdd1_new-shared[2136]: setfsid 1 > Aug 28 06:27:36 gluster13 gluster-bricksdd1_new-shared[2136]: spinlock 1 > Aug 28 06:27:36 gluster13 gluster-bricksdd1_new-shared[2136]: epoll.h 1 > Aug 28 06:27:36 gluster13 gluster-bricksdd1_new-shared[2136]: xattr.h 1 > Aug 28 06:27:36 gluster13 gluster-bricksdd1_new-shared[2136]: st_atim.tv_nsec > 1 > Aug 28 06:27:36 gluster13 gluster-bricksdd1_new-shared[2136]: > package-string: glusterfs 3.12.13 > Aug 28 06:27:36 gluster13 gluster-bricksdd1_new-shared[2136]: - > > There are some errors+warnings in the shared.log (volume logfile), but > no error message telling me why > gluster13:/gluster/bricksdd1_new/shared has disconnected. > > Well... at the moment load is ok, all 3 servers at about 15 (but i > think it will go up when more users will cause more
Re: [Gluster-users] Gluter 3.12.12: performance during heal and in general
Good Morning, today i update + rebooted all gluster servers, kernel update to 4.9.0-8 and gluster to 3.12.13. Reboots went fine, but on one of the gluster servers (gluster13) one of the bricks did come up at the beginning but then lost connection. OK: Status of volume: shared Gluster process TCP Port RDMA Port Online Pid -- [...] Brick gluster11:/gluster/bricksdd1/shared 49155 0 Y 2506 Brick gluster12:/gluster/bricksdd1_new/shared49155 0 Y 2097 Brick gluster13:/gluster/bricksdd1_new/shared49155 0 Y 2136 Lost connection: Brick gluster11:/gluster/bricksdd1/shared 49155 0 Y 2506 Brick gluster12:/gluster/bricksdd1_new/shared 49155 0 Y 2097 Brick gluster13:/gluster/bricksdd1_new/shared N/A N/A N N/A gluster volume heal shared info: Brick gluster13:/gluster/bricksdd1_new/shared Status: Transport endpoint is not connected Number of entries: - reboot was at 06:15:39; brick then worked for a short period, but then somehow disconnected. from gluster13:/var/log/glusterfs/glusterd.log: [2018-08-28 04:27:36.944608] I [MSGID: 106005] [glusterd-handler.c:6071:__glusterd_brick_rpc_notify] 0-management: Brick gluster13:/gluster/bricksdd1_new/shared has disconnected from glusterd. [2018-08-28 04:28:57.869666] I [glusterd-utils.c:6056:glusterd_brick_start] 0-management: starting a fresh brick process for brick /gluster/bricksdd1_new/shared [2018-08-28 04:35:20.732666] I [MSGID: 106143] [glusterd-pmap.c:295:pmap_registry_bind] 0-pmap: adding brick /gluster/bricksdd1_new/shared on port 49157 After 'gluster volume start shared force' (then with new port 49157): Brick gluster11:/gluster/bricksdd1/shared 49155 0 Y 2506 Brick gluster12:/gluster/bricksdd1_new/shared 49155 0 Y 2097 Brick gluster13:/gluster/bricksdd1_new/shared 49157 0 Y 3994 from /var/log/syslog: Aug 28 06:27:36 gluster13 gluster-bricksdd1_new-shared[2136]: pending frames: Aug 28 06:27:36 gluster13 gluster-bricksdd1_new-shared[2136]: frame : type(0) op(0) Aug 28 06:27:36 gluster13 gluster-bricksdd1_new-shared[2136]: frame : type(0) op(0) Aug 28 06:27:36 gluster13 gluster-bricksdd1_new-shared[2136]: patchset: git://git.gluster.org/glusterfs.git Aug 28 06:27:36 gluster13 gluster-bricksdd1_new-shared[2136]: signal received: 11 Aug 28 06:27:36 gluster13 gluster-bricksdd1_new-shared[2136]: time of crash: Aug 28 06:27:36 gluster13 gluster-bricksdd1_new-shared[2136]: 2018-08-28 04:27:36 Aug 28 06:27:36 gluster13 gluster-bricksdd1_new-shared[2136]: configuration details: Aug 28 06:27:36 gluster13 gluster-bricksdd1_new-shared[2136]: argp 1 Aug 28 06:27:36 gluster13 gluster-bricksdd1_new-shared[2136]: backtrace 1 Aug 28 06:27:36 gluster13 gluster-bricksdd1_new-shared[2136]: dlfcn 1 Aug 28 06:27:36 gluster13 gluster-bricksdd1_new-shared[2136]: libpthread 1 Aug 28 06:27:36 gluster13 gluster-bricksdd1_new-shared[2136]: llistxattr 1 Aug 28 06:27:36 gluster13 gluster-bricksdd1_new-shared[2136]: setfsid 1 Aug 28 06:27:36 gluster13 gluster-bricksdd1_new-shared[2136]: spinlock 1 Aug 28 06:27:36 gluster13 gluster-bricksdd1_new-shared[2136]: epoll.h 1 Aug 28 06:27:36 gluster13 gluster-bricksdd1_new-shared[2136]: xattr.h 1 Aug 28 06:27:36 gluster13 gluster-bricksdd1_new-shared[2136]: st_atim.tv_nsec 1 Aug 28 06:27:36 gluster13 gluster-bricksdd1_new-shared[2136]: package-string: glusterfs 3.12.13 Aug 28 06:27:36 gluster13 gluster-bricksdd1_new-shared[2136]: - There are some errors+warnings in the shared.log (volume logfile), but no error message telling me why gluster13:/gluster/bricksdd1_new/shared has disconnected. Well... at the moment load is ok, all 3 servers at about 15 (but i think it will go up when more users will cause more traffic -> more work on servers), 'gluster volume heal shared info' shows no entries, status: Status of volume: shared Gluster process TCP Port RDMA Port Online Pid -- Brick gluster11:/gluster/bricksda1/shared 49152 0 Y 2482 Brick gluster12:/gluster/bricksda1/shared 49152 0 Y 2088 Brick gluster13:/gluster/bricksda1/shared 49152 0 Y 2115 Brick gluster11:/gluster/bricksdb1/shared 49153 0 Y 2489 Brick gluster12:/gluster/bricksdb1/shared 49153 0 Y 2094 Brick gluster13:/gluster/bricksdb1/shared 49153 0 Y 2116 Brick gluster11:/gluster/bricksdc1/shared 49154 0 Y 2497 Brick gluster12:/gluster/bricksdc1/shared 49154 0 Y 2095 Brick gluster13:/gluster/bricksdc1/shared 49154 0 Y 2127 Brick gluster11:/gluster/bricksdd1/shared 49155 0 Y 2506 Brick
Re: [Gluster-users] Gluter 3.12.12: performance during heal and in general
yeah, on debian xyz.log.1 is always the former logfile which has been rotated by logrotate. Just checked the 3 servers: now it looks good, i will check it again tomorrow. very strange, maybe logrotate hasn't worked properly. the performance problems remain :-) 2018-08-27 15:41 GMT+02:00 Milind Changire : > On Thu, Aug 23, 2018 at 5:28 PM, Pranith Kumar Karampuri > wrote: >> >> On Wed, Aug 22, 2018 at 12:01 PM Hu Bert wrote: >>> >>> Just an addition: in general there are no log messages in >>> /var/log/glusterfs/ (if you don't all 'gluster volume ...'), but on >>> the node with the lowest load i see in cli.log.1: >>> >>> [2018-08-22 06:20:43.291055] I [socket.c:2474:socket_event_handler] >>> 0-transport: EPOLLERR - disconnecting now >>> [2018-08-22 06:20:46.291327] I [socket.c:2474:socket_event_handler] >>> 0-transport: EPOLLERR - disconnecting now >>> [2018-08-22 06:20:49.291575] I [socket.c:2474:socket_event_handler] >>> 0-transport: EPOLLERR - disconnecting now >>> >>> every 3 seconds. Looks like this bug: >>> https://bugzilla.redhat.com/show_bug.cgi?id=1484885 - but that shoud >>> have been fixed in the 3.12.x release, and network is fine. >> >> >> +Milind Changire > > That's odd. Presuming cli.log.1 is the logrotated file, it should be showing > older log entries than cli.log. But its not the case here. > Or maybe, there's something running on the command-line on the node with the > lowest load. > >> >>> >>> In cli.log there are only these entries: >>> >>> [2018-08-22 06:19:23.428520] I [cli.c:765:main] 0-cli: Started running >>> gluster with version 3.12.12 >>> [2018-08-22 06:19:23.800895] I [MSGID: 101190] >>> [event-epoll.c:613:event_dispatch_epoll_worker] 0-epoll: Started >>> thread with index 1 >>> [2018-08-22 06:19:23.800978] I [socket.c:2474:socket_event_handler] >>> 0-transport: EPOLLERR - disconnecting now >>> [2018-08-22 06:19:23.809366] I [input.c:31:cli_batch] 0-: Exiting with: 0 >>> >>> Just wondered if this could related anyhow. >>> >>> 2018-08-21 8:17 GMT+02:00 Pranith Kumar Karampuri : >>> > >>> > >>> > On Tue, Aug 21, 2018 at 11:40 AM Hu Bert >>> > wrote: >>> >> >>> >> Good morning :-) >>> >> >>> >> gluster11: >>> >> ls -l /gluster/bricksdd1/shared/.glusterfs/indices/xattrop/ >>> >> total 0 >>> >> -- 1 root root 0 Aug 14 06:14 >>> >> xattrop-006b65d8-9e81-4886-b380-89168ea079bd >>> >> >>> >> gluster12: >>> >> ls -l /gluster/bricksdd1_new/shared/.glusterfs/indices/xattrop/ >>> >> total 0 >>> >> -- 1 root root 0 Jul 17 11:24 >>> >> xattrop-c7c6f765-ce17-4361-95fb-2fd7f31c7b82 >>> >> >>> >> gluster13: >>> >> ls -l /gluster/bricksdd1_new/shared/.glusterfs/indices/xattrop/ >>> >> total 0 >>> >> -- 1 root root 0 Aug 16 07:54 >>> >> xattrop-16b696a0-4214-4999-b277-0917c76c983e >>> >> >>> >> >>> >> And here's the output of 'perf ...' which ran almost a minute - file >>> >> grew pretty fast to a size of 17 GB and system load went up heavily. >>> >> Had to wait a while until load dropped :-) >>> >> >>> >> fyi - load at the moment: >>> >> load gluster11: ~90 >>> >> load gluster12: ~10 >>> >> load gluster13: ~50 >>> >> >>> >> perf record --call-graph=dwarf -p 7897 -o >>> >> /tmp/perf.gluster11.bricksdd1.out >>> >> [ perf record: Woken up 9837 times to write data ] >>> >> Warning: >>> >> Processed 2137218 events and lost 33446 chunks! >>> >> >>> >> Check IO/CPU overload! >>> >> >>> >> [ perf record: Captured and wrote 16576.374 MB >>> >> /tmp/perf.gluster11.bricksdd1.out (2047760 samples) ] >>> >> >>> >> Here's an excerpt. >>> >> >>> >> +1.93% 0.00% glusteriotwr0[unknown] [k] >>> >> 0x >>> >> +1.89% 0.00% glusteriotwr28 [unknown] [k] >>> >> 0x >>> >> +1.86% 0.00% glusteriotwr15 [unknown] [k] >>> >> 0x >>> >> +1.85% 0.00% glusteriotwr63 [unknown] [k] >>> >> 0x >>> >> +1.83% 0.01% glusteriotwr0[kernel.kallsyms] [k] >>> >> entry_SYSCALL_64_after_swapgs >>> >> +1.82% 0.00% glusteriotwr38 [unknown] [k] >>> >> 0x >>> >> +1.82% 0.01% glusteriotwr28 [kernel.kallsyms] [k] >>> >> entry_SYSCALL_64_after_swapgs >>> >> +1.82% 0.00% glusteriotwr0[kernel.kallsyms] [k] >>> >> do_syscall_64 >>> >> +1.81% 0.00% glusteriotwr28 [kernel.kallsyms] [k] >>> >> do_syscall_64 >>> >> +1.81% 0.00% glusteriotwr15 [kernel.kallsyms] [k] >>> >> entry_SYSCALL_64_after_swapgs >>> >> +1.81% 0.00% glusteriotwr36 [unknown] [k] >>> >> 0x >>> >> +1.80% 0.00% glusteriotwr15 [kernel.kallsyms] [k] >>> >> do_syscall_64 >>> >> +1.78% 0.01% glusteriotwr63 [kernel.kallsyms] [k] >>> >> entry_SYSCALL_64_after_swapgs >>> >> +1.77% 0.00% glusteriotwr63 [kernel.kallsyms] [k] >>> >> do_syscall_64 >>> >> +1.75% 0.01% glusteriotwr38
Re: [Gluster-users] Gluter 3.12.12: performance during heal and in general
On Thu, Aug 23, 2018 at 5:28 PM, Pranith Kumar Karampuri < pkara...@redhat.com> wrote: > On Wed, Aug 22, 2018 at 12:01 PM Hu Bert wrote: > >> Just an addition: in general there are no log messages in >> /var/log/glusterfs/ (if you don't all 'gluster volume ...'), but on >> the node with the lowest load i see in cli.log.1: >> >> [2018-08-22 06:20:43.291055] I [socket.c:2474:socket_event_handler] >> 0-transport: EPOLLERR - disconnecting now >> [2018-08-22 06:20:46.291327] I [socket.c:2474:socket_event_handler] >> 0-transport: EPOLLERR - disconnecting now >> [2018-08-22 06:20:49.291575] I [socket.c:2474:socket_event_handler] >> 0-transport: EPOLLERR - disconnecting now >> >> every 3 seconds. Looks like this bug: >> https://bugzilla.redhat.com/show_bug.cgi?id=1484885 - but that shoud >> have been fixed in the 3.12.x release, and network is fine. >> > > +Milind Changire > That's odd. Presuming cli.log.1 is the logrotated file, it should be showing older log entries than cli.log. But its not the case here. Or maybe, there's something running on the command-line on the node with the lowest load. > >> In cli.log there are only these entries: >> >> [2018-08-22 06:19:23.428520] I [cli.c:765:main] 0-cli: Started running >> gluster with version 3.12.12 >> [2018-08-22 06:19:23.800895] I [MSGID: 101190] >> [event-epoll.c:613:event_dispatch_epoll_worker] 0-epoll: Started >> thread with index 1 >> [2018-08-22 06:19:23.800978] I [socket.c:2474:socket_event_handler] >> 0-transport: EPOLLERR - disconnecting now >> [2018-08-22 06:19:23.809366] I [input.c:31:cli_batch] 0-: Exiting with: 0 >> >> Just wondered if this could related anyhow. >> >> 2018-08-21 8:17 GMT+02:00 Pranith Kumar Karampuri : >> > >> > >> > On Tue, Aug 21, 2018 at 11:40 AM Hu Bert >> wrote: >> >> >> >> Good morning :-) >> >> >> >> gluster11: >> >> ls -l /gluster/bricksdd1/shared/.glusterfs/indices/xattrop/ >> >> total 0 >> >> -- 1 root root 0 Aug 14 06:14 >> >> xattrop-006b65d8-9e81-4886-b380-89168ea079bd >> >> >> >> gluster12: >> >> ls -l /gluster/bricksdd1_new/shared/.glusterfs/indices/xattrop/ >> >> total 0 >> >> -- 1 root root 0 Jul 17 11:24 >> >> xattrop-c7c6f765-ce17-4361-95fb-2fd7f31c7b82 >> >> >> >> gluster13: >> >> ls -l /gluster/bricksdd1_new/shared/.glusterfs/indices/xattrop/ >> >> total 0 >> >> -- 1 root root 0 Aug 16 07:54 >> >> xattrop-16b696a0-4214-4999-b277-0917c76c983e >> >> >> >> >> >> And here's the output of 'perf ...' which ran almost a minute - file >> >> grew pretty fast to a size of 17 GB and system load went up heavily. >> >> Had to wait a while until load dropped :-) >> >> >> >> fyi - load at the moment: >> >> load gluster11: ~90 >> >> load gluster12: ~10 >> >> load gluster13: ~50 >> >> >> >> perf record --call-graph=dwarf -p 7897 -o >> >> /tmp/perf.gluster11.bricksdd1.out >> >> [ perf record: Woken up 9837 times to write data ] >> >> Warning: >> >> Processed 2137218 events and lost 33446 chunks! >> >> >> >> Check IO/CPU overload! >> >> >> >> [ perf record: Captured and wrote 16576.374 MB >> >> /tmp/perf.gluster11.bricksdd1.out (2047760 samples) ] >> >> >> >> Here's an excerpt. >> >> >> >> +1.93% 0.00% glusteriotwr0[unknown] [k] >> >> 0x >> >> +1.89% 0.00% glusteriotwr28 [unknown] [k] >> >> 0x >> >> +1.86% 0.00% glusteriotwr15 [unknown] [k] >> >> 0x >> >> +1.85% 0.00% glusteriotwr63 [unknown] [k] >> >> 0x >> >> +1.83% 0.01% glusteriotwr0[kernel.kallsyms] [k] >> >> entry_SYSCALL_64_after_swapgs >> >> +1.82% 0.00% glusteriotwr38 [unknown] [k] >> >> 0x >> >> +1.82% 0.01% glusteriotwr28 [kernel.kallsyms] [k] >> >> entry_SYSCALL_64_after_swapgs >> >> +1.82% 0.00% glusteriotwr0[kernel.kallsyms] [k] >> >> do_syscall_64 >> >> +1.81% 0.00% glusteriotwr28 [kernel.kallsyms] [k] >> >> do_syscall_64 >> >> +1.81% 0.00% glusteriotwr15 [kernel.kallsyms] [k] >> >> entry_SYSCALL_64_after_swapgs >> >> +1.81% 0.00% glusteriotwr36 [unknown] [k] >> >> 0x >> >> +1.80% 0.00% glusteriotwr15 [kernel.kallsyms] [k] >> >> do_syscall_64 >> >> +1.78% 0.01% glusteriotwr63 [kernel.kallsyms] [k] >> >> entry_SYSCALL_64_after_swapgs >> >> +1.77% 0.00% glusteriotwr63 [kernel.kallsyms] [k] >> >> do_syscall_64 >> >> +1.75% 0.01% glusteriotwr38 [kernel.kallsyms] [k] >> >> entry_SYSCALL_64_after_swapgs >> >> +1.75% 0.00% glusteriotwr38 [kernel.kallsyms] [k] >> >> do_syscall_64 >> >> +1.74% 0.00% glusteriotwr17 [unknown] [k] >> >> 0x >> >> +1.74% 0.00% glusteriotwr44 [unknown] [k] >> >> 0x >> >> +1.73% 0.00% glusteriotwr6[unknown] [k] >> >>
Re: [Gluster-users] Gluter 3.12.12: performance during heal and in general
On Wed, Aug 22, 2018 at 12:01 PM Hu Bert wrote: > Just an addition: in general there are no log messages in > /var/log/glusterfs/ (if you don't all 'gluster volume ...'), but on > the node with the lowest load i see in cli.log.1: > > [2018-08-22 06:20:43.291055] I [socket.c:2474:socket_event_handler] > 0-transport: EPOLLERR - disconnecting now > [2018-08-22 06:20:46.291327] I [socket.c:2474:socket_event_handler] > 0-transport: EPOLLERR - disconnecting now > [2018-08-22 06:20:49.291575] I [socket.c:2474:socket_event_handler] > 0-transport: EPOLLERR - disconnecting now > > every 3 seconds. Looks like this bug: > https://bugzilla.redhat.com/show_bug.cgi?id=1484885 - but that shoud > have been fixed in the 3.12.x release, and network is fine. > +Milind Changire > In cli.log there are only these entries: > > [2018-08-22 06:19:23.428520] I [cli.c:765:main] 0-cli: Started running > gluster with version 3.12.12 > [2018-08-22 06:19:23.800895] I [MSGID: 101190] > [event-epoll.c:613:event_dispatch_epoll_worker] 0-epoll: Started > thread with index 1 > [2018-08-22 06:19:23.800978] I [socket.c:2474:socket_event_handler] > 0-transport: EPOLLERR - disconnecting now > [2018-08-22 06:19:23.809366] I [input.c:31:cli_batch] 0-: Exiting with: 0 > > Just wondered if this could related anyhow. > > 2018-08-21 8:17 GMT+02:00 Pranith Kumar Karampuri : > > > > > > On Tue, Aug 21, 2018 at 11:40 AM Hu Bert wrote: > >> > >> Good morning :-) > >> > >> gluster11: > >> ls -l /gluster/bricksdd1/shared/.glusterfs/indices/xattrop/ > >> total 0 > >> -- 1 root root 0 Aug 14 06:14 > >> xattrop-006b65d8-9e81-4886-b380-89168ea079bd > >> > >> gluster12: > >> ls -l /gluster/bricksdd1_new/shared/.glusterfs/indices/xattrop/ > >> total 0 > >> -- 1 root root 0 Jul 17 11:24 > >> xattrop-c7c6f765-ce17-4361-95fb-2fd7f31c7b82 > >> > >> gluster13: > >> ls -l /gluster/bricksdd1_new/shared/.glusterfs/indices/xattrop/ > >> total 0 > >> -- 1 root root 0 Aug 16 07:54 > >> xattrop-16b696a0-4214-4999-b277-0917c76c983e > >> > >> > >> And here's the output of 'perf ...' which ran almost a minute - file > >> grew pretty fast to a size of 17 GB and system load went up heavily. > >> Had to wait a while until load dropped :-) > >> > >> fyi - load at the moment: > >> load gluster11: ~90 > >> load gluster12: ~10 > >> load gluster13: ~50 > >> > >> perf record --call-graph=dwarf -p 7897 -o > >> /tmp/perf.gluster11.bricksdd1.out > >> [ perf record: Woken up 9837 times to write data ] > >> Warning: > >> Processed 2137218 events and lost 33446 chunks! > >> > >> Check IO/CPU overload! > >> > >> [ perf record: Captured and wrote 16576.374 MB > >> /tmp/perf.gluster11.bricksdd1.out (2047760 samples) ] > >> > >> Here's an excerpt. > >> > >> +1.93% 0.00% glusteriotwr0[unknown] [k] > >> 0x > >> +1.89% 0.00% glusteriotwr28 [unknown] [k] > >> 0x > >> +1.86% 0.00% glusteriotwr15 [unknown] [k] > >> 0x > >> +1.85% 0.00% glusteriotwr63 [unknown] [k] > >> 0x > >> +1.83% 0.01% glusteriotwr0[kernel.kallsyms] [k] > >> entry_SYSCALL_64_after_swapgs > >> +1.82% 0.00% glusteriotwr38 [unknown] [k] > >> 0x > >> +1.82% 0.01% glusteriotwr28 [kernel.kallsyms] [k] > >> entry_SYSCALL_64_after_swapgs > >> +1.82% 0.00% glusteriotwr0[kernel.kallsyms] [k] > >> do_syscall_64 > >> +1.81% 0.00% glusteriotwr28 [kernel.kallsyms] [k] > >> do_syscall_64 > >> +1.81% 0.00% glusteriotwr15 [kernel.kallsyms] [k] > >> entry_SYSCALL_64_after_swapgs > >> +1.81% 0.00% glusteriotwr36 [unknown] [k] > >> 0x > >> +1.80% 0.00% glusteriotwr15 [kernel.kallsyms] [k] > >> do_syscall_64 > >> +1.78% 0.01% glusteriotwr63 [kernel.kallsyms] [k] > >> entry_SYSCALL_64_after_swapgs > >> +1.77% 0.00% glusteriotwr63 [kernel.kallsyms] [k] > >> do_syscall_64 > >> +1.75% 0.01% glusteriotwr38 [kernel.kallsyms] [k] > >> entry_SYSCALL_64_after_swapgs > >> +1.75% 0.00% glusteriotwr38 [kernel.kallsyms] [k] > >> do_syscall_64 > >> +1.74% 0.00% glusteriotwr17 [unknown] [k] > >> 0x > >> +1.74% 0.00% glusteriotwr44 [unknown] [k] > >> 0x > >> +1.73% 0.00% glusteriotwr6[unknown] [k] > >> 0x > >> +1.73% 0.00% glusteriotwr37 [unknown] [k] > >> 0x > >> +1.73% 0.01% glusteriotwr36 [kernel.kallsyms] [k] > >> entry_SYSCALL_64_after_swapgs > >> +1.72% 0.00% glusteriotwr34 [unknown] [k] > >> 0x > >> +1.72% 0.00% glusteriotwr36 [kernel.kallsyms] [k] > >> do_syscall_64 > >> +1.71% 0.00%
Re: [Gluster-users] Gluter 3.12.12: performance during heal and in general
Just an addition: in general there are no log messages in /var/log/glusterfs/ (if you don't all 'gluster volume ...'), but on the node with the lowest load i see in cli.log.1: [2018-08-22 06:20:43.291055] I [socket.c:2474:socket_event_handler] 0-transport: EPOLLERR - disconnecting now [2018-08-22 06:20:46.291327] I [socket.c:2474:socket_event_handler] 0-transport: EPOLLERR - disconnecting now [2018-08-22 06:20:49.291575] I [socket.c:2474:socket_event_handler] 0-transport: EPOLLERR - disconnecting now every 3 seconds. Looks like this bug: https://bugzilla.redhat.com/show_bug.cgi?id=1484885 - but that shoud have been fixed in the 3.12.x release, and network is fine. In cli.log there are only these entries: [2018-08-22 06:19:23.428520] I [cli.c:765:main] 0-cli: Started running gluster with version 3.12.12 [2018-08-22 06:19:23.800895] I [MSGID: 101190] [event-epoll.c:613:event_dispatch_epoll_worker] 0-epoll: Started thread with index 1 [2018-08-22 06:19:23.800978] I [socket.c:2474:socket_event_handler] 0-transport: EPOLLERR - disconnecting now [2018-08-22 06:19:23.809366] I [input.c:31:cli_batch] 0-: Exiting with: 0 Just wondered if this could related anyhow. 2018-08-21 8:17 GMT+02:00 Pranith Kumar Karampuri : > > > On Tue, Aug 21, 2018 at 11:40 AM Hu Bert wrote: >> >> Good morning :-) >> >> gluster11: >> ls -l /gluster/bricksdd1/shared/.glusterfs/indices/xattrop/ >> total 0 >> -- 1 root root 0 Aug 14 06:14 >> xattrop-006b65d8-9e81-4886-b380-89168ea079bd >> >> gluster12: >> ls -l /gluster/bricksdd1_new/shared/.glusterfs/indices/xattrop/ >> total 0 >> -- 1 root root 0 Jul 17 11:24 >> xattrop-c7c6f765-ce17-4361-95fb-2fd7f31c7b82 >> >> gluster13: >> ls -l /gluster/bricksdd1_new/shared/.glusterfs/indices/xattrop/ >> total 0 >> -- 1 root root 0 Aug 16 07:54 >> xattrop-16b696a0-4214-4999-b277-0917c76c983e >> >> >> And here's the output of 'perf ...' which ran almost a minute - file >> grew pretty fast to a size of 17 GB and system load went up heavily. >> Had to wait a while until load dropped :-) >> >> fyi - load at the moment: >> load gluster11: ~90 >> load gluster12: ~10 >> load gluster13: ~50 >> >> perf record --call-graph=dwarf -p 7897 -o >> /tmp/perf.gluster11.bricksdd1.out >> [ perf record: Woken up 9837 times to write data ] >> Warning: >> Processed 2137218 events and lost 33446 chunks! >> >> Check IO/CPU overload! >> >> [ perf record: Captured and wrote 16576.374 MB >> /tmp/perf.gluster11.bricksdd1.out (2047760 samples) ] >> >> Here's an excerpt. >> >> +1.93% 0.00% glusteriotwr0[unknown] [k] >> 0x >> +1.89% 0.00% glusteriotwr28 [unknown] [k] >> 0x >> +1.86% 0.00% glusteriotwr15 [unknown] [k] >> 0x >> +1.85% 0.00% glusteriotwr63 [unknown] [k] >> 0x >> +1.83% 0.01% glusteriotwr0[kernel.kallsyms] [k] >> entry_SYSCALL_64_after_swapgs >> +1.82% 0.00% glusteriotwr38 [unknown] [k] >> 0x >> +1.82% 0.01% glusteriotwr28 [kernel.kallsyms] [k] >> entry_SYSCALL_64_after_swapgs >> +1.82% 0.00% glusteriotwr0[kernel.kallsyms] [k] >> do_syscall_64 >> +1.81% 0.00% glusteriotwr28 [kernel.kallsyms] [k] >> do_syscall_64 >> +1.81% 0.00% glusteriotwr15 [kernel.kallsyms] [k] >> entry_SYSCALL_64_after_swapgs >> +1.81% 0.00% glusteriotwr36 [unknown] [k] >> 0x >> +1.80% 0.00% glusteriotwr15 [kernel.kallsyms] [k] >> do_syscall_64 >> +1.78% 0.01% glusteriotwr63 [kernel.kallsyms] [k] >> entry_SYSCALL_64_after_swapgs >> +1.77% 0.00% glusteriotwr63 [kernel.kallsyms] [k] >> do_syscall_64 >> +1.75% 0.01% glusteriotwr38 [kernel.kallsyms] [k] >> entry_SYSCALL_64_after_swapgs >> +1.75% 0.00% glusteriotwr38 [kernel.kallsyms] [k] >> do_syscall_64 >> +1.74% 0.00% glusteriotwr17 [unknown] [k] >> 0x >> +1.74% 0.00% glusteriotwr44 [unknown] [k] >> 0x >> +1.73% 0.00% glusteriotwr6[unknown] [k] >> 0x >> +1.73% 0.00% glusteriotwr37 [unknown] [k] >> 0x >> +1.73% 0.01% glusteriotwr36 [kernel.kallsyms] [k] >> entry_SYSCALL_64_after_swapgs >> +1.72% 0.00% glusteriotwr34 [unknown] [k] >> 0x >> +1.72% 0.00% glusteriotwr36 [kernel.kallsyms] [k] >> do_syscall_64 >> +1.71% 0.00% glusteriotwr45 [unknown] [k] >> 0x >> +1.70% 0.00% glusteriotwr7[unknown] [k] >> 0x >> +1.68% 0.00% glusteriotwr15 [kernel.kallsyms] [k] >> sys_getdents >> +1.68% 0.00% glusteriotwr15 [kernel.kallsyms] [k] filldir
Re: [Gluster-users] Gluter 3.12.12: performance during heal and in general
On Tue, Aug 21, 2018 at 11:40 AM Hu Bert wrote: > Good morning :-) > > gluster11: > ls -l /gluster/bricksdd1/shared/.glusterfs/indices/xattrop/ > total 0 > -- 1 root root 0 Aug 14 06:14 > xattrop-006b65d8-9e81-4886-b380-89168ea079bd > > gluster12: > ls -l /gluster/bricksdd1_new/shared/.glusterfs/indices/xattrop/ > total 0 > -- 1 root root 0 Jul 17 11:24 > xattrop-c7c6f765-ce17-4361-95fb-2fd7f31c7b82 > > gluster13: > ls -l /gluster/bricksdd1_new/shared/.glusterfs/indices/xattrop/ > total 0 > -- 1 root root 0 Aug 16 07:54 > xattrop-16b696a0-4214-4999-b277-0917c76c983e > > > And here's the output of 'perf ...' which ran almost a minute - file > grew pretty fast to a size of 17 GB and system load went up heavily. > Had to wait a while until load dropped :-) > > fyi - load at the moment: > load gluster11: ~90 > load gluster12: ~10 > load gluster13: ~50 > > perf record --call-graph=dwarf -p 7897 -o /tmp/perf.gluster11.bricksdd1.out > [ perf record: Woken up 9837 times to write data ] > Warning: > Processed 2137218 events and lost 33446 chunks! > > Check IO/CPU overload! > > [ perf record: Captured and wrote 16576.374 MB > /tmp/perf.gluster11.bricksdd1.out (2047760 samples) ] > > Here's an excerpt. > > +1.93% 0.00% glusteriotwr0[unknown] [k] > 0x > +1.89% 0.00% glusteriotwr28 [unknown] [k] > 0x > +1.86% 0.00% glusteriotwr15 [unknown] [k] > 0x > +1.85% 0.00% glusteriotwr63 [unknown] [k] > 0x > +1.83% 0.01% glusteriotwr0[kernel.kallsyms] [k] > entry_SYSCALL_64_after_swapgs > +1.82% 0.00% glusteriotwr38 [unknown] [k] > 0x > +1.82% 0.01% glusteriotwr28 [kernel.kallsyms] [k] > entry_SYSCALL_64_after_swapgs > +1.82% 0.00% glusteriotwr0[kernel.kallsyms] [k] > do_syscall_64 > +1.81% 0.00% glusteriotwr28 [kernel.kallsyms] [k] > do_syscall_64 > +1.81% 0.00% glusteriotwr15 [kernel.kallsyms] [k] > entry_SYSCALL_64_after_swapgs > +1.81% 0.00% glusteriotwr36 [unknown] [k] > 0x > +1.80% 0.00% glusteriotwr15 [kernel.kallsyms] [k] > do_syscall_64 > +1.78% 0.01% glusteriotwr63 [kernel.kallsyms] [k] > entry_SYSCALL_64_after_swapgs > +1.77% 0.00% glusteriotwr63 [kernel.kallsyms] [k] > do_syscall_64 > +1.75% 0.01% glusteriotwr38 [kernel.kallsyms] [k] > entry_SYSCALL_64_after_swapgs > +1.75% 0.00% glusteriotwr38 [kernel.kallsyms] [k] > do_syscall_64 > +1.74% 0.00% glusteriotwr17 [unknown] [k] > 0x > +1.74% 0.00% glusteriotwr44 [unknown] [k] > 0x > +1.73% 0.00% glusteriotwr6[unknown] [k] > 0x > +1.73% 0.00% glusteriotwr37 [unknown] [k] > 0x > +1.73% 0.01% glusteriotwr36 [kernel.kallsyms] [k] > entry_SYSCALL_64_after_swapgs > +1.72% 0.00% glusteriotwr34 [unknown] [k] > 0x > +1.72% 0.00% glusteriotwr36 [kernel.kallsyms] [k] > do_syscall_64 > +1.71% 0.00% glusteriotwr45 [unknown] [k] > 0x > +1.70% 0.00% glusteriotwr7[unknown] [k] > 0x > +1.68% 0.00% glusteriotwr15 [kernel.kallsyms] [k] > sys_getdents > +1.68% 0.00% glusteriotwr15 [kernel.kallsyms] [k] filldir > +1.68% 0.00% glusteriotwr15 libc-2.24.so [.] > 0x80c60db8ef2b > +1.68% 0.00% glusteriotwr15 libc-2.24.so [.] > readdir64 > +1.68% 0.00% glusteriotwr15 index.so [.] > 0x80c6192a1888 > +1.68% 0.00% glusteriotwr15 [kernel.kallsyms] [k] > iterate_dir > +1.68% 0.00% glusteriotwr15 [kernel.kallsyms] [k] > ext4_htree_fill_tree > +1.68% 0.00% glusteriotwr15 [kernel.kallsyms] [k] > ext4_readdir > > Or do you want to download the file /tmp/perf.gluster11.bricksdd1.out > and examine it yourself? If so i could send you a link. > Thank you! yes a link would be great. I am not as good with kernel side of things. So I will have to show this information to someone else who knows these things so expect delay in response. > > > 2018-08-21 7:13 GMT+02:00 Pranith Kumar Karampuri : > > > > > > On Tue, Aug 21, 2018 at 10:13 AM Pranith Kumar Karampuri > > wrote: > >> > >> > >> > >> On Mon, Aug 20, 2018 at 3:20 PM Hu Bert wrote: > >>> > >>> Regarding hardware the machines are identical. Intel Xeon E5-1650 v3 > >>> Hexa-Core; 64 GB DDR4 ECC; Dell PERC H330 8 Port SAS/SATA 12 GBit/s > >>> RAID Controller; operating system running on a raid1, then 4 disks > >>> (JBOD) as bricks. > >>> > >>> Ok, i ran perf for a few
Re: [Gluster-users] Gluter 3.12.12: performance during heal and in general
Good morning :-) gluster11: ls -l /gluster/bricksdd1/shared/.glusterfs/indices/xattrop/ total 0 -- 1 root root 0 Aug 14 06:14 xattrop-006b65d8-9e81-4886-b380-89168ea079bd gluster12: ls -l /gluster/bricksdd1_new/shared/.glusterfs/indices/xattrop/ total 0 -- 1 root root 0 Jul 17 11:24 xattrop-c7c6f765-ce17-4361-95fb-2fd7f31c7b82 gluster13: ls -l /gluster/bricksdd1_new/shared/.glusterfs/indices/xattrop/ total 0 -- 1 root root 0 Aug 16 07:54 xattrop-16b696a0-4214-4999-b277-0917c76c983e And here's the output of 'perf ...' which ran almost a minute - file grew pretty fast to a size of 17 GB and system load went up heavily. Had to wait a while until load dropped :-) fyi - load at the moment: load gluster11: ~90 load gluster12: ~10 load gluster13: ~50 perf record --call-graph=dwarf -p 7897 -o /tmp/perf.gluster11.bricksdd1.out [ perf record: Woken up 9837 times to write data ] Warning: Processed 2137218 events and lost 33446 chunks! Check IO/CPU overload! [ perf record: Captured and wrote 16576.374 MB /tmp/perf.gluster11.bricksdd1.out (2047760 samples) ] Here's an excerpt. +1.93% 0.00% glusteriotwr0[unknown] [k] 0x +1.89% 0.00% glusteriotwr28 [unknown] [k] 0x +1.86% 0.00% glusteriotwr15 [unknown] [k] 0x +1.85% 0.00% glusteriotwr63 [unknown] [k] 0x +1.83% 0.01% glusteriotwr0[kernel.kallsyms] [k] entry_SYSCALL_64_after_swapgs +1.82% 0.00% glusteriotwr38 [unknown] [k] 0x +1.82% 0.01% glusteriotwr28 [kernel.kallsyms] [k] entry_SYSCALL_64_after_swapgs +1.82% 0.00% glusteriotwr0[kernel.kallsyms] [k] do_syscall_64 +1.81% 0.00% glusteriotwr28 [kernel.kallsyms] [k] do_syscall_64 +1.81% 0.00% glusteriotwr15 [kernel.kallsyms] [k] entry_SYSCALL_64_after_swapgs +1.81% 0.00% glusteriotwr36 [unknown] [k] 0x +1.80% 0.00% glusteriotwr15 [kernel.kallsyms] [k] do_syscall_64 +1.78% 0.01% glusteriotwr63 [kernel.kallsyms] [k] entry_SYSCALL_64_after_swapgs +1.77% 0.00% glusteriotwr63 [kernel.kallsyms] [k] do_syscall_64 +1.75% 0.01% glusteriotwr38 [kernel.kallsyms] [k] entry_SYSCALL_64_after_swapgs +1.75% 0.00% glusteriotwr38 [kernel.kallsyms] [k] do_syscall_64 +1.74% 0.00% glusteriotwr17 [unknown] [k] 0x +1.74% 0.00% glusteriotwr44 [unknown] [k] 0x +1.73% 0.00% glusteriotwr6[unknown] [k] 0x +1.73% 0.00% glusteriotwr37 [unknown] [k] 0x +1.73% 0.01% glusteriotwr36 [kernel.kallsyms] [k] entry_SYSCALL_64_after_swapgs +1.72% 0.00% glusteriotwr34 [unknown] [k] 0x +1.72% 0.00% glusteriotwr36 [kernel.kallsyms] [k] do_syscall_64 +1.71% 0.00% glusteriotwr45 [unknown] [k] 0x +1.70% 0.00% glusteriotwr7[unknown] [k] 0x +1.68% 0.00% glusteriotwr15 [kernel.kallsyms] [k] sys_getdents +1.68% 0.00% glusteriotwr15 [kernel.kallsyms] [k] filldir +1.68% 0.00% glusteriotwr15 libc-2.24.so [.] 0x80c60db8ef2b +1.68% 0.00% glusteriotwr15 libc-2.24.so [.] readdir64 +1.68% 0.00% glusteriotwr15 index.so [.] 0x80c6192a1888 +1.68% 0.00% glusteriotwr15 [kernel.kallsyms] [k] iterate_dir +1.68% 0.00% glusteriotwr15 [kernel.kallsyms] [k] ext4_htree_fill_tree +1.68% 0.00% glusteriotwr15 [kernel.kallsyms] [k] ext4_readdir Or do you want to download the file /tmp/perf.gluster11.bricksdd1.out and examine it yourself? If so i could send you a link. 2018-08-21 7:13 GMT+02:00 Pranith Kumar Karampuri : > > > On Tue, Aug 21, 2018 at 10:13 AM Pranith Kumar Karampuri > wrote: >> >> >> >> On Mon, Aug 20, 2018 at 3:20 PM Hu Bert wrote: >>> >>> Regarding hardware the machines are identical. Intel Xeon E5-1650 v3 >>> Hexa-Core; 64 GB DDR4 ECC; Dell PERC H330 8 Port SAS/SATA 12 GBit/s >>> RAID Controller; operating system running on a raid1, then 4 disks >>> (JBOD) as bricks. >>> >>> Ok, i ran perf for a few seconds. >>> >>> >>> perf record --call-graph=dwarf -p 7897 -o >>> /tmp/perf.gluster11.bricksdd1.out >>> ^C[ perf record: Woken up 378 times to write data ] >>> Warning: >>> Processed 83690 events and lost 96 chunks! >>> >>> Check IO/CPU overload! >>> >>> [ perf record: Captured and wrote 423.087 MB >>> /tmp/perf.gluster11.bricksdd1.out (51744 samples) ] >>> >>> >>> I copied a couple of lines: >>> >>> +8.10% 0.00%
Re: [Gluster-users] Gluter 3.12.12: performance during heal and in general
On Tue, Aug 21, 2018 at 10:13 AM Pranith Kumar Karampuri < pkara...@redhat.com> wrote: > > > On Mon, Aug 20, 2018 at 3:20 PM Hu Bert wrote: > >> Regarding hardware the machines are identical. Intel Xeon E5-1650 v3 >> Hexa-Core; 64 GB DDR4 ECC; Dell PERC H330 8 Port SAS/SATA 12 GBit/s >> RAID Controller; operating system running on a raid1, then 4 disks >> (JBOD) as bricks. >> >> Ok, i ran perf for a few seconds. >> >> >> perf record --call-graph=dwarf -p 7897 -o >> /tmp/perf.gluster11.bricksdd1.out >> ^C[ perf record: Woken up 378 times to write data ] >> Warning: >> Processed 83690 events and lost 96 chunks! >> >> Check IO/CPU overload! >> >> [ perf record: Captured and wrote 423.087 MB >> /tmp/perf.gluster11.bricksdd1.out (51744 samples) ] >> >> >> I copied a couple of lines: >> >> +8.10% 0.00% glusteriotwr22 [unknown] [k] >> 0x >> +8.10% 0.00% glusteriotwr22 [kernel.kallsyms] [k] >> iterate_dir >> +8.10% 0.00% glusteriotwr22 [kernel.kallsyms] [k] >> sys_getdents >> +8.10% 0.00% glusteriotwr22 [kernel.kallsyms] [k] filldir >> +8.10% 0.00% glusteriotwr22 [kernel.kallsyms] [k] >> do_syscall_64 >> +8.10% 0.00% glusteriotwr22 [kernel.kallsyms] [k] >> entry_SYSCALL_64_after_swapgs >> +8.10% 0.00% glusteriotwr22 libc-2.24.so [.] >> 0x80c60db8ef2b >> +8.10% 0.00% glusteriotwr22 libc-2.24.so [.] >> readdir64 >> +8.10% 0.00% glusteriotwr22 index.so [.] >> 0x80c6192a1888 >> +8.10% 0.04% glusteriotwr22 [kernel.kallsyms] [k] >> ext4_htree_fill_tree >> +8.10% 0.00% glusteriotwr22 [kernel.kallsyms] [k] >> ext4_readdir >> +7.95% 0.12% glusteriotwr22 [kernel.kallsyms] [k] >> htree_dirblock_to_tree >> +5.78% 0.96% glusteriotwr22 [kernel.kallsyms] [k] >> __ext4_read_dirblock >> +4.80% 0.02% glusteriotwr22 [kernel.kallsyms] [k] >> ext4_bread >> +4.78% 0.04% glusteriotwr22 [kernel.kallsyms] [k] >> ext4_getblk >> +4.72% 0.02% glusteriotwr22 [kernel.kallsyms] [k] >> __getblk_gfp >> +4.57% 0.00% glusteriotwr3[unknown] [k] >> 0x >> +4.55% 0.00% glusteriotwr3[kernel.kallsyms] [k] >> do_syscall_64 >> >> Do you need different or additional information? >> > > This looks like there are lot of readdirs going on which is different from > what we observed earlier, how many seconds did you do perf record for? Will > it be possible for you to do this for some more time? may be a minute? Just > want to be sure that the data actually represents what we are observing. > I found one code path which on lookup does readdirs. Could you give me the output of ls -l /.glusterfs/indices/xattrop on all the three bricks? It can probably give a correlation to see if it is indeed the same issue or not. > > >> >> >> 2018-08-20 11:20 GMT+02:00 Pranith Kumar Karampuri : >> > Even the brick which doesn't have high CPU seems to have same number of >> > lookups, so that's not it. >> > Is there any difference at all between the machines which have high CPU >> vs >> > low CPU? >> > I think the only other thing I would do is to install perf tools and >> try to >> > figure out the call-graph which is leading to so much CPU >> > >> > This affects performance of the brick I think, so you may have to do it >> > quickly and for less time. >> > >> > perf record --call-graph=dwarf -p-o >> > then >> > perf report -i >> > >> > >> > On Mon, Aug 20, 2018 at 2:40 PM Hu Bert wrote: >> >> >> >> gluster volume heal shared info | grep -i number >> >> Number of entries: 0 >> >> Number of entries: 0 >> >> Number of entries: 0 >> >> Number of entries: 0 >> >> Number of entries: 0 >> >> Number of entries: 0 >> >> Number of entries: 0 >> >> Number of entries: 0 >> >> Number of entries: 0 >> >> Number of entries: 0 >> >> Number of entries: 0 >> >> Number of entries: 0 >> >> >> >> Looks good to me. >> >> >> >> >> >> 2018-08-20 10:51 GMT+02:00 Pranith Kumar Karampuri < >> pkara...@redhat.com>: >> >> > There are a lot of Lookup operations in the system. But I am not >> able to >> >> > find why. Could you check the output of >> >> > >> >> > # gluster volume heal info | grep -i number >> >> > >> >> > it should print all zeros. >> >> > >> >> > On Fri, Aug 17, 2018 at 1:49 PM Hu Bert >> wrote: >> >> >> >> >> >> I don't know what you exactly mean with workload, but the main >> >> >> function of the volume is storing (incl. writing, reading) images >> >> >> (from hundreds of bytes up to 30 MBs, overall ~7TB). The work is >> done >> >> >> by apache tomcat servers writing to / reading from the volume. >> Besides >> >> >> images there are some text files and binaries that are stored on the >> >> >> volume and get updated regularly (every x hours); we'll try to >>
Re: [Gluster-users] Gluter 3.12.12: performance during heal and in general
On Mon, Aug 20, 2018 at 3:20 PM Hu Bert wrote: > Regarding hardware the machines are identical. Intel Xeon E5-1650 v3 > Hexa-Core; 64 GB DDR4 ECC; Dell PERC H330 8 Port SAS/SATA 12 GBit/s > RAID Controller; operating system running on a raid1, then 4 disks > (JBOD) as bricks. > > Ok, i ran perf for a few seconds. > > > perf record --call-graph=dwarf -p 7897 -o > /tmp/perf.gluster11.bricksdd1.out > ^C[ perf record: Woken up 378 times to write data ] > Warning: > Processed 83690 events and lost 96 chunks! > > Check IO/CPU overload! > > [ perf record: Captured and wrote 423.087 MB > /tmp/perf.gluster11.bricksdd1.out (51744 samples) ] > > > I copied a couple of lines: > > +8.10% 0.00% glusteriotwr22 [unknown] [k] > 0x > +8.10% 0.00% glusteriotwr22 [kernel.kallsyms] [k] > iterate_dir > +8.10% 0.00% glusteriotwr22 [kernel.kallsyms] [k] > sys_getdents > +8.10% 0.00% glusteriotwr22 [kernel.kallsyms] [k] filldir > +8.10% 0.00% glusteriotwr22 [kernel.kallsyms] [k] > do_syscall_64 > +8.10% 0.00% glusteriotwr22 [kernel.kallsyms] [k] > entry_SYSCALL_64_after_swapgs > +8.10% 0.00% glusteriotwr22 libc-2.24.so [.] > 0x80c60db8ef2b > +8.10% 0.00% glusteriotwr22 libc-2.24.so [.] > readdir64 > +8.10% 0.00% glusteriotwr22 index.so [.] > 0x80c6192a1888 > +8.10% 0.04% glusteriotwr22 [kernel.kallsyms] [k] > ext4_htree_fill_tree > +8.10% 0.00% glusteriotwr22 [kernel.kallsyms] [k] > ext4_readdir > +7.95% 0.12% glusteriotwr22 [kernel.kallsyms] [k] > htree_dirblock_to_tree > +5.78% 0.96% glusteriotwr22 [kernel.kallsyms] [k] > __ext4_read_dirblock > +4.80% 0.02% glusteriotwr22 [kernel.kallsyms] [k] > ext4_bread > +4.78% 0.04% glusteriotwr22 [kernel.kallsyms] [k] > ext4_getblk > +4.72% 0.02% glusteriotwr22 [kernel.kallsyms] [k] > __getblk_gfp > +4.57% 0.00% glusteriotwr3[unknown] [k] > 0x > +4.55% 0.00% glusteriotwr3[kernel.kallsyms] [k] > do_syscall_64 > > Do you need different or additional information? > This looks like there are lot of readdirs going on which is different from what we observed earlier, how many seconds did you do perf record for? Will it be possible for you to do this for some more time? may be a minute? Just want to be sure that the data actually represents what we are observing. > > > 2018-08-20 11:20 GMT+02:00 Pranith Kumar Karampuri : > > Even the brick which doesn't have high CPU seems to have same number of > > lookups, so that's not it. > > Is there any difference at all between the machines which have high CPU > vs > > low CPU? > > I think the only other thing I would do is to install perf tools and try > to > > figure out the call-graph which is leading to so much CPU > > > > This affects performance of the brick I think, so you may have to do it > > quickly and for less time. > > > > perf record --call-graph=dwarf -p-o > > then > > perf report -i > > > > > > On Mon, Aug 20, 2018 at 2:40 PM Hu Bert wrote: > >> > >> gluster volume heal shared info | grep -i number > >> Number of entries: 0 > >> Number of entries: 0 > >> Number of entries: 0 > >> Number of entries: 0 > >> Number of entries: 0 > >> Number of entries: 0 > >> Number of entries: 0 > >> Number of entries: 0 > >> Number of entries: 0 > >> Number of entries: 0 > >> Number of entries: 0 > >> Number of entries: 0 > >> > >> Looks good to me. > >> > >> > >> 2018-08-20 10:51 GMT+02:00 Pranith Kumar Karampuri >: > >> > There are a lot of Lookup operations in the system. But I am not able > to > >> > find why. Could you check the output of > >> > > >> > # gluster volume heal info | grep -i number > >> > > >> > it should print all zeros. > >> > > >> > On Fri, Aug 17, 2018 at 1:49 PM Hu Bert > wrote: > >> >> > >> >> I don't know what you exactly mean with workload, but the main > >> >> function of the volume is storing (incl. writing, reading) images > >> >> (from hundreds of bytes up to 30 MBs, overall ~7TB). The work is done > >> >> by apache tomcat servers writing to / reading from the volume. > Besides > >> >> images there are some text files and binaries that are stored on the > >> >> volume and get updated regularly (every x hours); we'll try to > migrate > >> >> the latter ones to local storage asap. > >> >> > >> >> Interestingly it's only one process (and its threads) of the same > >> >> brick on 2 of the gluster servers that consumes the CPU. > >> >> > >> >> gluster11: bricksdd1; not healed; full CPU > >> >> gluster12: bricksdd1; got healed; normal CPU > >> >> gluster13: bricksdd1; got healed; full CPU > >> >> > >> >> Besides: performance during heal (e.g. gluster12, bricksdd1) was way > >> >> better than it is now. I've
Re: [Gluster-users] Gluter 3.12.12: performance during heal and in general
Regarding hardware the machines are identical. Intel Xeon E5-1650 v3 Hexa-Core; 64 GB DDR4 ECC; Dell PERC H330 8 Port SAS/SATA 12 GBit/s RAID Controller; operating system running on a raid1, then 4 disks (JBOD) as bricks. Ok, i ran perf for a few seconds. perf record --call-graph=dwarf -p 7897 -o /tmp/perf.gluster11.bricksdd1.out ^C[ perf record: Woken up 378 times to write data ] Warning: Processed 83690 events and lost 96 chunks! Check IO/CPU overload! [ perf record: Captured and wrote 423.087 MB /tmp/perf.gluster11.bricksdd1.out (51744 samples) ] I copied a couple of lines: +8.10% 0.00% glusteriotwr22 [unknown] [k] 0x +8.10% 0.00% glusteriotwr22 [kernel.kallsyms] [k] iterate_dir +8.10% 0.00% glusteriotwr22 [kernel.kallsyms] [k] sys_getdents +8.10% 0.00% glusteriotwr22 [kernel.kallsyms] [k] filldir +8.10% 0.00% glusteriotwr22 [kernel.kallsyms] [k] do_syscall_64 +8.10% 0.00% glusteriotwr22 [kernel.kallsyms] [k] entry_SYSCALL_64_after_swapgs +8.10% 0.00% glusteriotwr22 libc-2.24.so [.] 0x80c60db8ef2b +8.10% 0.00% glusteriotwr22 libc-2.24.so [.] readdir64 +8.10% 0.00% glusteriotwr22 index.so [.] 0x80c6192a1888 +8.10% 0.04% glusteriotwr22 [kernel.kallsyms] [k] ext4_htree_fill_tree +8.10% 0.00% glusteriotwr22 [kernel.kallsyms] [k] ext4_readdir +7.95% 0.12% glusteriotwr22 [kernel.kallsyms] [k] htree_dirblock_to_tree +5.78% 0.96% glusteriotwr22 [kernel.kallsyms] [k] __ext4_read_dirblock +4.80% 0.02% glusteriotwr22 [kernel.kallsyms] [k] ext4_bread +4.78% 0.04% glusteriotwr22 [kernel.kallsyms] [k] ext4_getblk +4.72% 0.02% glusteriotwr22 [kernel.kallsyms] [k] __getblk_gfp +4.57% 0.00% glusteriotwr3[unknown] [k] 0x +4.55% 0.00% glusteriotwr3[kernel.kallsyms] [k] do_syscall_64 Do you need different or additional information? 2018-08-20 11:20 GMT+02:00 Pranith Kumar Karampuri : > Even the brick which doesn't have high CPU seems to have same number of > lookups, so that's not it. > Is there any difference at all between the machines which have high CPU vs > low CPU? > I think the only other thing I would do is to install perf tools and try to > figure out the call-graph which is leading to so much CPU > > This affects performance of the brick I think, so you may have to do it > quickly and for less time. > > perf record --call-graph=dwarf -p-o > then > perf report -i > > > On Mon, Aug 20, 2018 at 2:40 PM Hu Bert wrote: >> >> gluster volume heal shared info | grep -i number >> Number of entries: 0 >> Number of entries: 0 >> Number of entries: 0 >> Number of entries: 0 >> Number of entries: 0 >> Number of entries: 0 >> Number of entries: 0 >> Number of entries: 0 >> Number of entries: 0 >> Number of entries: 0 >> Number of entries: 0 >> Number of entries: 0 >> >> Looks good to me. >> >> >> 2018-08-20 10:51 GMT+02:00 Pranith Kumar Karampuri : >> > There are a lot of Lookup operations in the system. But I am not able to >> > find why. Could you check the output of >> > >> > # gluster volume heal info | grep -i number >> > >> > it should print all zeros. >> > >> > On Fri, Aug 17, 2018 at 1:49 PM Hu Bert wrote: >> >> >> >> I don't know what you exactly mean with workload, but the main >> >> function of the volume is storing (incl. writing, reading) images >> >> (from hundreds of bytes up to 30 MBs, overall ~7TB). The work is done >> >> by apache tomcat servers writing to / reading from the volume. Besides >> >> images there are some text files and binaries that are stored on the >> >> volume and get updated regularly (every x hours); we'll try to migrate >> >> the latter ones to local storage asap. >> >> >> >> Interestingly it's only one process (and its threads) of the same >> >> brick on 2 of the gluster servers that consumes the CPU. >> >> >> >> gluster11: bricksdd1; not healed; full CPU >> >> gluster12: bricksdd1; got healed; normal CPU >> >> gluster13: bricksdd1; got healed; full CPU >> >> >> >> Besides: performance during heal (e.g. gluster12, bricksdd1) was way >> >> better than it is now. I've attached 2 pngs showing the differing cpu >> >> usage of last week before/after heal. >> >> >> >> 2018-08-17 9:30 GMT+02:00 Pranith Kumar Karampuri >> >> : >> >> > There seems to be too many lookup operations compared to any other >> >> > operations. What is the workload on the volume? >> >> > >> >> > On Fri, Aug 17, 2018 at 12:47 PM Hu Bert >> >> > wrote: >> >> >> >> >> >> i hope i did get it right. >> >> >> >> >> >> gluster volume profile shared start >> >> >> wait 10 minutes >> >> >> gluster volume profile shared info >> >> >> gluster volume profile shared stop >> >> >> >> >> >> If that's ok,
Re: [Gluster-users] Gluter 3.12.12: performance during heal and in general
gluster volume heal shared info | grep -i number Number of entries: 0 Number of entries: 0 Number of entries: 0 Number of entries: 0 Number of entries: 0 Number of entries: 0 Number of entries: 0 Number of entries: 0 Number of entries: 0 Number of entries: 0 Number of entries: 0 Number of entries: 0 Looks good to me. 2018-08-20 10:51 GMT+02:00 Pranith Kumar Karampuri : > There are a lot of Lookup operations in the system. But I am not able to > find why. Could you check the output of > > # gluster volume heal info | grep -i number > > it should print all zeros. > > On Fri, Aug 17, 2018 at 1:49 PM Hu Bert wrote: >> >> I don't know what you exactly mean with workload, but the main >> function of the volume is storing (incl. writing, reading) images >> (from hundreds of bytes up to 30 MBs, overall ~7TB). The work is done >> by apache tomcat servers writing to / reading from the volume. Besides >> images there are some text files and binaries that are stored on the >> volume and get updated regularly (every x hours); we'll try to migrate >> the latter ones to local storage asap. >> >> Interestingly it's only one process (and its threads) of the same >> brick on 2 of the gluster servers that consumes the CPU. >> >> gluster11: bricksdd1; not healed; full CPU >> gluster12: bricksdd1; got healed; normal CPU >> gluster13: bricksdd1; got healed; full CPU >> >> Besides: performance during heal (e.g. gluster12, bricksdd1) was way >> better than it is now. I've attached 2 pngs showing the differing cpu >> usage of last week before/after heal. >> >> 2018-08-17 9:30 GMT+02:00 Pranith Kumar Karampuri : >> > There seems to be too many lookup operations compared to any other >> > operations. What is the workload on the volume? >> > >> > On Fri, Aug 17, 2018 at 12:47 PM Hu Bert wrote: >> >> >> >> i hope i did get it right. >> >> >> >> gluster volume profile shared start >> >> wait 10 minutes >> >> gluster volume profile shared info >> >> gluster volume profile shared stop >> >> >> >> If that's ok, i've attached the output of the info command. >> >> >> >> >> >> 2018-08-17 8:31 GMT+02:00 Pranith Kumar Karampuri >> >> : >> >> > Please do volume profile also for around 10 minutes when CPU% is >> >> > high. >> >> > >> >> > On Fri, Aug 17, 2018 at 11:56 AM Pranith Kumar Karampuri >> >> > wrote: >> >> >> >> >> >> As per the output, all io-threads are using a lot of CPU. It is >> >> >> better >> >> >> to >> >> >> check what the volume profile is to see what is leading to so much >> >> >> work >> >> >> for >> >> >> io-threads. Please follow the documentation at >> >> >> >> >> >> >> >> >> https://gluster.readthedocs.io/en/latest/Administrator%20Guide/Monitoring%20Workload/ >> >> >> section: " >> >> >> >> >> >> Running GlusterFS Volume Profile Command" >> >> >> >> >> >> and attach output of "gluster volume profile info", >> >> >> >> >> >> On Fri, Aug 17, 2018 at 11:24 AM Hu Bert >> >> >> wrote: >> >> >>> >> >> >>> Good morning, >> >> >>> >> >> >>> i ran the command during 100% CPU usage and attached the file. >> >> >>> Hopefully it helps. >> >> >>> >> >> >>> 2018-08-17 7:33 GMT+02:00 Pranith Kumar Karampuri >> >> >>> : >> >> >>> > Could you do the following on one of the nodes where you are >> >> >>> > observing >> >> >>> > high >> >> >>> > CPU usage and attach that file to this thread? We can find what >> >> >>> > threads/processes are leading to high usage. Do this for say 10 >> >> >>> > minutes >> >> >>> > when >> >> >>> > you see the ~100% CPU. >> >> >>> > >> >> >>> > top -bHd 5 > /tmp/top.${HOSTNAME}.txt >> >> >>> > >> >> >>> > On Wed, Aug 15, 2018 at 2:37 PM Hu Bert >> >> >>> > wrote: >> >> >>> >> >> >> >>> >> Hello again :-) >> >> >>> >> >> >> >>> >> The self heal must have finished as there are no log entries in >> >> >>> >> glustershd.log files anymore. According to munin disk latency >> >> >>> >> (average >> >> >>> >> io wait) has gone down to 100 ms, and disk utilization has gone >> >> >>> >> down >> >> >>> >> to ~60% - both on all servers and hard disks. >> >> >>> >> >> >> >>> >> But now system load on 2 servers (which were in the good state) >> >> >>> >> fluctuates between 60 and 100; the server with the formerly >> >> >>> >> failed >> >> >>> >> disk has a load of 20-30.I've uploaded some munin graphics of >> >> >>> >> the >> >> >>> >> cpu >> >> >>> >> usage: >> >> >>> >> >> >> >>> >> https://abload.de/img/gluster11_cpu31d3a.png >> >> >>> >> https://abload.de/img/gluster12_cpu8sem7.png >> >> >>> >> https://abload.de/img/gluster13_cpud7eni.png >> >> >>> >> >> >> >>> >> This can't be normal. 2 of the servers under heavy load and one >> >> >>> >> not >> >> >>> >> that much. Does anyone have an explanation of this strange >> >> >>> >> behaviour? >> >> >>> >> >> >> >>> >> >> >> >>> >> Thx :-) >> >> >>> >> >> >> >>> >> 2018-08-14 9:37 GMT+02:00 Hu Bert : >> >> >>> >> > Hi there, >> >> >>> >> > >> >> >>> >> > well, it seems the heal has finally finished. Couldn't >> >> >>> >> > see/find >> >> >>> >> > any >> >> >>> >>
Re: [Gluster-users] Gluter 3.12.12: performance during heal and in general
There are a lot of Lookup operations in the system. But I am not able to find why. Could you check the output of # gluster volume heal info | grep -i number it should print all zeros. On Fri, Aug 17, 2018 at 1:49 PM Hu Bert wrote: > I don't know what you exactly mean with workload, but the main > function of the volume is storing (incl. writing, reading) images > (from hundreds of bytes up to 30 MBs, overall ~7TB). The work is done > by apache tomcat servers writing to / reading from the volume. Besides > images there are some text files and binaries that are stored on the > volume and get updated regularly (every x hours); we'll try to migrate > the latter ones to local storage asap. > > Interestingly it's only one process (and its threads) of the same > brick on 2 of the gluster servers that consumes the CPU. > > gluster11: bricksdd1; not healed; full CPU > gluster12: bricksdd1; got healed; normal CPU > gluster13: bricksdd1; got healed; full CPU > > Besides: performance during heal (e.g. gluster12, bricksdd1) was way > better than it is now. I've attached 2 pngs showing the differing cpu > usage of last week before/after heal. > > 2018-08-17 9:30 GMT+02:00 Pranith Kumar Karampuri : > > There seems to be too many lookup operations compared to any other > > operations. What is the workload on the volume? > > > > On Fri, Aug 17, 2018 at 12:47 PM Hu Bert wrote: > >> > >> i hope i did get it right. > >> > >> gluster volume profile shared start > >> wait 10 minutes > >> gluster volume profile shared info > >> gluster volume profile shared stop > >> > >> If that's ok, i've attached the output of the info command. > >> > >> > >> 2018-08-17 8:31 GMT+02:00 Pranith Kumar Karampuri >: > >> > Please do volume profile also for around 10 minutes when CPU% is high. > >> > > >> > On Fri, Aug 17, 2018 at 11:56 AM Pranith Kumar Karampuri > >> > wrote: > >> >> > >> >> As per the output, all io-threads are using a lot of CPU. It is > better > >> >> to > >> >> check what the volume profile is to see what is leading to so much > work > >> >> for > >> >> io-threads. Please follow the documentation at > >> >> > >> >> > https://gluster.readthedocs.io/en/latest/Administrator%20Guide/Monitoring%20Workload/ > >> >> section: " > >> >> > >> >> Running GlusterFS Volume Profile Command" > >> >> > >> >> and attach output of "gluster volume profile info", > >> >> > >> >> On Fri, Aug 17, 2018 at 11:24 AM Hu Bert > >> >> wrote: > >> >>> > >> >>> Good morning, > >> >>> > >> >>> i ran the command during 100% CPU usage and attached the file. > >> >>> Hopefully it helps. > >> >>> > >> >>> 2018-08-17 7:33 GMT+02:00 Pranith Kumar Karampuri > >> >>> : > >> >>> > Could you do the following on one of the nodes where you are > >> >>> > observing > >> >>> > high > >> >>> > CPU usage and attach that file to this thread? We can find what > >> >>> > threads/processes are leading to high usage. Do this for say 10 > >> >>> > minutes > >> >>> > when > >> >>> > you see the ~100% CPU. > >> >>> > > >> >>> > top -bHd 5 > /tmp/top.${HOSTNAME}.txt > >> >>> > > >> >>> > On Wed, Aug 15, 2018 at 2:37 PM Hu Bert > >> >>> > wrote: > >> >>> >> > >> >>> >> Hello again :-) > >> >>> >> > >> >>> >> The self heal must have finished as there are no log entries in > >> >>> >> glustershd.log files anymore. According to munin disk latency > >> >>> >> (average > >> >>> >> io wait) has gone down to 100 ms, and disk utilization has gone > >> >>> >> down > >> >>> >> to ~60% - both on all servers and hard disks. > >> >>> >> > >> >>> >> But now system load on 2 servers (which were in the good state) > >> >>> >> fluctuates between 60 and 100; the server with the formerly > failed > >> >>> >> disk has a load of 20-30.I've uploaded some munin graphics of the > >> >>> >> cpu > >> >>> >> usage: > >> >>> >> > >> >>> >> https://abload.de/img/gluster11_cpu31d3a.png > >> >>> >> https://abload.de/img/gluster12_cpu8sem7.png > >> >>> >> https://abload.de/img/gluster13_cpud7eni.png > >> >>> >> > >> >>> >> This can't be normal. 2 of the servers under heavy load and one > not > >> >>> >> that much. Does anyone have an explanation of this strange > >> >>> >> behaviour? > >> >>> >> > >> >>> >> > >> >>> >> Thx :-) > >> >>> >> > >> >>> >> 2018-08-14 9:37 GMT+02:00 Hu Bert : > >> >>> >> > Hi there, > >> >>> >> > > >> >>> >> > well, it seems the heal has finally finished. Couldn't see/find > >> >>> >> > any > >> >>> >> > related log message; is there such a message in a specific log > >> >>> >> > file? > >> >>> >> > > >> >>> >> > But i see the same behaviour when the last heal finished: all > CPU > >> >>> >> > cores are consumed by brick processes; not only by the formerly > >> >>> >> > failed > >> >>> >> > bricksdd1, but by all 4 brick processes (and their threads). > Load > >> >>> >> > goes > >> >>> >> > up to > 100 on the 2 servers with the not-failed brick, and > >> >>> >> > glustershd.log gets filled with a lot of entries. Load on the > >> >>> >> > server > >> >>> >> > with the then
Re: [Gluster-users] Gluter 3.12.12: performance during heal and in general
I don't know what you exactly mean with workload, but the main function of the volume is storing (incl. writing, reading) images (from hundreds of bytes up to 30 MBs, overall ~7TB). The work is done by apache tomcat servers writing to / reading from the volume. Besides images there are some text files and binaries that are stored on the volume and get updated regularly (every x hours); we'll try to migrate the latter ones to local storage asap. Interestingly it's only one process (and its threads) of the same brick on 2 of the gluster servers that consumes the CPU. gluster11: bricksdd1; not healed; full CPU gluster12: bricksdd1; got healed; normal CPU gluster13: bricksdd1; got healed; full CPU Besides: performance during heal (e.g. gluster12, bricksdd1) was way better than it is now. I've attached 2 pngs showing the differing cpu usage of last week before/after heal. 2018-08-17 9:30 GMT+02:00 Pranith Kumar Karampuri : > There seems to be too many lookup operations compared to any other > operations. What is the workload on the volume? > > On Fri, Aug 17, 2018 at 12:47 PM Hu Bert wrote: >> >> i hope i did get it right. >> >> gluster volume profile shared start >> wait 10 minutes >> gluster volume profile shared info >> gluster volume profile shared stop >> >> If that's ok, i've attached the output of the info command. >> >> >> 2018-08-17 8:31 GMT+02:00 Pranith Kumar Karampuri : >> > Please do volume profile also for around 10 minutes when CPU% is high. >> > >> > On Fri, Aug 17, 2018 at 11:56 AM Pranith Kumar Karampuri >> > wrote: >> >> >> >> As per the output, all io-threads are using a lot of CPU. It is better >> >> to >> >> check what the volume profile is to see what is leading to so much work >> >> for >> >> io-threads. Please follow the documentation at >> >> >> >> https://gluster.readthedocs.io/en/latest/Administrator%20Guide/Monitoring%20Workload/ >> >> section: " >> >> >> >> Running GlusterFS Volume Profile Command" >> >> >> >> and attach output of "gluster volume profile info", >> >> >> >> On Fri, Aug 17, 2018 at 11:24 AM Hu Bert >> >> wrote: >> >>> >> >>> Good morning, >> >>> >> >>> i ran the command during 100% CPU usage and attached the file. >> >>> Hopefully it helps. >> >>> >> >>> 2018-08-17 7:33 GMT+02:00 Pranith Kumar Karampuri >> >>> : >> >>> > Could you do the following on one of the nodes where you are >> >>> > observing >> >>> > high >> >>> > CPU usage and attach that file to this thread? We can find what >> >>> > threads/processes are leading to high usage. Do this for say 10 >> >>> > minutes >> >>> > when >> >>> > you see the ~100% CPU. >> >>> > >> >>> > top -bHd 5 > /tmp/top.${HOSTNAME}.txt >> >>> > >> >>> > On Wed, Aug 15, 2018 at 2:37 PM Hu Bert >> >>> > wrote: >> >>> >> >> >>> >> Hello again :-) >> >>> >> >> >>> >> The self heal must have finished as there are no log entries in >> >>> >> glustershd.log files anymore. According to munin disk latency >> >>> >> (average >> >>> >> io wait) has gone down to 100 ms, and disk utilization has gone >> >>> >> down >> >>> >> to ~60% - both on all servers and hard disks. >> >>> >> >> >>> >> But now system load on 2 servers (which were in the good state) >> >>> >> fluctuates between 60 and 100; the server with the formerly failed >> >>> >> disk has a load of 20-30.I've uploaded some munin graphics of the >> >>> >> cpu >> >>> >> usage: >> >>> >> >> >>> >> https://abload.de/img/gluster11_cpu31d3a.png >> >>> >> https://abload.de/img/gluster12_cpu8sem7.png >> >>> >> https://abload.de/img/gluster13_cpud7eni.png >> >>> >> >> >>> >> This can't be normal. 2 of the servers under heavy load and one not >> >>> >> that much. Does anyone have an explanation of this strange >> >>> >> behaviour? >> >>> >> >> >>> >> >> >>> >> Thx :-) >> >>> >> >> >>> >> 2018-08-14 9:37 GMT+02:00 Hu Bert : >> >>> >> > Hi there, >> >>> >> > >> >>> >> > well, it seems the heal has finally finished. Couldn't see/find >> >>> >> > any >> >>> >> > related log message; is there such a message in a specific log >> >>> >> > file? >> >>> >> > >> >>> >> > But i see the same behaviour when the last heal finished: all CPU >> >>> >> > cores are consumed by brick processes; not only by the formerly >> >>> >> > failed >> >>> >> > bricksdd1, but by all 4 brick processes (and their threads). Load >> >>> >> > goes >> >>> >> > up to > 100 on the 2 servers with the not-failed brick, and >> >>> >> > glustershd.log gets filled with a lot of entries. Load on the >> >>> >> > server >> >>> >> > with the then failed brick not that high, but still ~60. >> >>> >> > >> >>> >> > Is this behaviour normal? Is there some post-heal after a heal >> >>> >> > has >> >>> >> > finished? >> >>> >> > >> >>> >> > thx in advance :-) >> >>> > >> >>> > >> >>> > >> >>> > -- >> >>> > Pranith >> >> >> >> >> >> >> >> -- >> >> Pranith >> > >> > >> > >> > -- >> > Pranith > > > > -- > Pranith ___ Gluster-users mailing list Gluster-users@gluster.org
Re: [Gluster-users] Gluter 3.12.12: performance during heal and in general
There seems to be too many lookup operations compared to any other operations. What is the workload on the volume? On Fri, Aug 17, 2018 at 12:47 PM Hu Bert wrote: > i hope i did get it right. > > gluster volume profile shared start > wait 10 minutes > gluster volume profile shared info > gluster volume profile shared stop > > If that's ok, i've attached the output of the info command. > > > 2018-08-17 8:31 GMT+02:00 Pranith Kumar Karampuri : > > Please do volume profile also for around 10 minutes when CPU% is high. > > > > On Fri, Aug 17, 2018 at 11:56 AM Pranith Kumar Karampuri > > wrote: > >> > >> As per the output, all io-threads are using a lot of CPU. It is better > to > >> check what the volume profile is to see what is leading to so much work > for > >> io-threads. Please follow the documentation at > >> > https://gluster.readthedocs.io/en/latest/Administrator%20Guide/Monitoring%20Workload/ > >> section: " > >> > >> Running GlusterFS Volume Profile Command" > >> > >> and attach output of "gluster volume profile info", > >> > >> On Fri, Aug 17, 2018 at 11:24 AM Hu Bert > wrote: > >>> > >>> Good morning, > >>> > >>> i ran the command during 100% CPU usage and attached the file. > >>> Hopefully it helps. > >>> > >>> 2018-08-17 7:33 GMT+02:00 Pranith Kumar Karampuri >: > >>> > Could you do the following on one of the nodes where you are > observing > >>> > high > >>> > CPU usage and attach that file to this thread? We can find what > >>> > threads/processes are leading to high usage. Do this for say 10 > minutes > >>> > when > >>> > you see the ~100% CPU. > >>> > > >>> > top -bHd 5 > /tmp/top.${HOSTNAME}.txt > >>> > > >>> > On Wed, Aug 15, 2018 at 2:37 PM Hu Bert > wrote: > >>> >> > >>> >> Hello again :-) > >>> >> > >>> >> The self heal must have finished as there are no log entries in > >>> >> glustershd.log files anymore. According to munin disk latency > (average > >>> >> io wait) has gone down to 100 ms, and disk utilization has gone down > >>> >> to ~60% - both on all servers and hard disks. > >>> >> > >>> >> But now system load on 2 servers (which were in the good state) > >>> >> fluctuates between 60 and 100; the server with the formerly failed > >>> >> disk has a load of 20-30.I've uploaded some munin graphics of the > cpu > >>> >> usage: > >>> >> > >>> >> https://abload.de/img/gluster11_cpu31d3a.png > >>> >> https://abload.de/img/gluster12_cpu8sem7.png > >>> >> https://abload.de/img/gluster13_cpud7eni.png > >>> >> > >>> >> This can't be normal. 2 of the servers under heavy load and one not > >>> >> that much. Does anyone have an explanation of this strange > behaviour? > >>> >> > >>> >> > >>> >> Thx :-) > >>> >> > >>> >> 2018-08-14 9:37 GMT+02:00 Hu Bert : > >>> >> > Hi there, > >>> >> > > >>> >> > well, it seems the heal has finally finished. Couldn't see/find > any > >>> >> > related log message; is there such a message in a specific log > file? > >>> >> > > >>> >> > But i see the same behaviour when the last heal finished: all CPU > >>> >> > cores are consumed by brick processes; not only by the formerly > >>> >> > failed > >>> >> > bricksdd1, but by all 4 brick processes (and their threads). Load > >>> >> > goes > >>> >> > up to > 100 on the 2 servers with the not-failed brick, and > >>> >> > glustershd.log gets filled with a lot of entries. Load on the > server > >>> >> > with the then failed brick not that high, but still ~60. > >>> >> > > >>> >> > Is this behaviour normal? Is there some post-heal after a heal has > >>> >> > finished? > >>> >> > > >>> >> > thx in advance :-) > >>> > > >>> > > >>> > > >>> > -- > >>> > Pranith > >> > >> > >> > >> -- > >> Pranith > > > > > > > > -- > > Pranith > -- Pranith ___ Gluster-users mailing list Gluster-users@gluster.org https://lists.gluster.org/mailman/listinfo/gluster-users
Re: [Gluster-users] Gluter 3.12.12: performance during heal and in general
i hope i did get it right. gluster volume profile shared start wait 10 minutes gluster volume profile shared info gluster volume profile shared stop If that's ok, i've attached the output of the info command. 2018-08-17 8:31 GMT+02:00 Pranith Kumar Karampuri : > Please do volume profile also for around 10 minutes when CPU% is high. > > On Fri, Aug 17, 2018 at 11:56 AM Pranith Kumar Karampuri > wrote: >> >> As per the output, all io-threads are using a lot of CPU. It is better to >> check what the volume profile is to see what is leading to so much work for >> io-threads. Please follow the documentation at >> https://gluster.readthedocs.io/en/latest/Administrator%20Guide/Monitoring%20Workload/ >> section: " >> >> Running GlusterFS Volume Profile Command" >> >> and attach output of "gluster volume profile info", >> >> On Fri, Aug 17, 2018 at 11:24 AM Hu Bert wrote: >>> >>> Good morning, >>> >>> i ran the command during 100% CPU usage and attached the file. >>> Hopefully it helps. >>> >>> 2018-08-17 7:33 GMT+02:00 Pranith Kumar Karampuri : >>> > Could you do the following on one of the nodes where you are observing >>> > high >>> > CPU usage and attach that file to this thread? We can find what >>> > threads/processes are leading to high usage. Do this for say 10 minutes >>> > when >>> > you see the ~100% CPU. >>> > >>> > top -bHd 5 > /tmp/top.${HOSTNAME}.txt >>> > >>> > On Wed, Aug 15, 2018 at 2:37 PM Hu Bert wrote: >>> >> >>> >> Hello again :-) >>> >> >>> >> The self heal must have finished as there are no log entries in >>> >> glustershd.log files anymore. According to munin disk latency (average >>> >> io wait) has gone down to 100 ms, and disk utilization has gone down >>> >> to ~60% - both on all servers and hard disks. >>> >> >>> >> But now system load on 2 servers (which were in the good state) >>> >> fluctuates between 60 and 100; the server with the formerly failed >>> >> disk has a load of 20-30.I've uploaded some munin graphics of the cpu >>> >> usage: >>> >> >>> >> https://abload.de/img/gluster11_cpu31d3a.png >>> >> https://abload.de/img/gluster12_cpu8sem7.png >>> >> https://abload.de/img/gluster13_cpud7eni.png >>> >> >>> >> This can't be normal. 2 of the servers under heavy load and one not >>> >> that much. Does anyone have an explanation of this strange behaviour? >>> >> >>> >> >>> >> Thx :-) >>> >> >>> >> 2018-08-14 9:37 GMT+02:00 Hu Bert : >>> >> > Hi there, >>> >> > >>> >> > well, it seems the heal has finally finished. Couldn't see/find any >>> >> > related log message; is there such a message in a specific log file? >>> >> > >>> >> > But i see the same behaviour when the last heal finished: all CPU >>> >> > cores are consumed by brick processes; not only by the formerly >>> >> > failed >>> >> > bricksdd1, but by all 4 brick processes (and their threads). Load >>> >> > goes >>> >> > up to > 100 on the 2 servers with the not-failed brick, and >>> >> > glustershd.log gets filled with a lot of entries. Load on the server >>> >> > with the then failed brick not that high, but still ~60. >>> >> > >>> >> > Is this behaviour normal? Is there some post-heal after a heal has >>> >> > finished? >>> >> > >>> >> > thx in advance :-) >>> > >>> > >>> > >>> > -- >>> > Pranith >> >> >> >> -- >> Pranith > > > > -- > Pranith Brick: gluster11:/gluster/bricksda1/shared -- Cumulative Stats: Block Size: 1b+ 2b+ 4b+ No. of Reads:42346 No. of Writes: 2768 130 Block Size: 8b+ 16b+ 32b+ No. of Reads: 87 162 386 No. of Writes: 252 496 1524 Block Size: 64b+ 128b+ 256b+ No. of Reads: 600 1236 2849 No. of Writes: 5862 18485 8234 Block Size:512b+1024b+2048b+ No. of Reads: 8971 12202 22124 No. of Writes:17856 89024 94131 Block Size: 4096b+8192b+ 16384b+ No. of Reads:52532109761340599 No. of Writes: 231908 2099445401203 Block Size: 32768b+ 65536b+ 131072b+ No. of Reads: 241543 1160765 3331181 No. of Writes: 843531 1576773767822 %-latency Avg-latency Min-Latency Max-Latency No. of calls Fop - --- --- ---
Re: [Gluster-users] Gluter 3.12.12: performance during heal and in general
Please do volume profile also for around 10 minutes when CPU% is high. On Fri, Aug 17, 2018 at 11:56 AM Pranith Kumar Karampuri < pkara...@redhat.com> wrote: > As per the output, all io-threads are using a lot of CPU. It is better to > check what the volume profile is to see what is leading to so much work for > io-threads. Please follow the documentation at > https://gluster.readthedocs.io/en/latest/Administrator%20Guide/Monitoring%20Workload/ > section: "Running GlusterFS Volume Profile Command"and attach output of > "gluster > volume profile info", > > On Fri, Aug 17, 2018 at 11:24 AM Hu Bert wrote: > >> Good morning, >> >> i ran the command during 100% CPU usage and attached the file. >> Hopefully it helps. >> >> 2018-08-17 7:33 GMT+02:00 Pranith Kumar Karampuri : >> > Could you do the following on one of the nodes where you are observing >> high >> > CPU usage and attach that file to this thread? We can find what >> > threads/processes are leading to high usage. Do this for say 10 minutes >> when >> > you see the ~100% CPU. >> > >> > top -bHd 5 > /tmp/top.${HOSTNAME}.txt >> > >> > On Wed, Aug 15, 2018 at 2:37 PM Hu Bert wrote: >> >> >> >> Hello again :-) >> >> >> >> The self heal must have finished as there are no log entries in >> >> glustershd.log files anymore. According to munin disk latency (average >> >> io wait) has gone down to 100 ms, and disk utilization has gone down >> >> to ~60% - both on all servers and hard disks. >> >> >> >> But now system load on 2 servers (which were in the good state) >> >> fluctuates between 60 and 100; the server with the formerly failed >> >> disk has a load of 20-30.I've uploaded some munin graphics of the cpu >> >> usage: >> >> >> >> https://abload.de/img/gluster11_cpu31d3a.png >> >> https://abload.de/img/gluster12_cpu8sem7.png >> >> https://abload.de/img/gluster13_cpud7eni.png >> >> >> >> This can't be normal. 2 of the servers under heavy load and one not >> >> that much. Does anyone have an explanation of this strange behaviour? >> >> >> >> >> >> Thx :-) >> >> >> >> 2018-08-14 9:37 GMT+02:00 Hu Bert : >> >> > Hi there, >> >> > >> >> > well, it seems the heal has finally finished. Couldn't see/find any >> >> > related log message; is there such a message in a specific log file? >> >> > >> >> > But i see the same behaviour when the last heal finished: all CPU >> >> > cores are consumed by brick processes; not only by the formerly >> failed >> >> > bricksdd1, but by all 4 brick processes (and their threads). Load >> goes >> >> > up to > 100 on the 2 servers with the not-failed brick, and >> >> > glustershd.log gets filled with a lot of entries. Load on the server >> >> > with the then failed brick not that high, but still ~60. >> >> > >> >> > Is this behaviour normal? Is there some post-heal after a heal has >> >> > finished? >> >> > >> >> > thx in advance :-) >> > >> > >> > >> > -- >> > Pranith >> > > > -- > Pranith > -- Pranith ___ Gluster-users mailing list Gluster-users@gluster.org https://lists.gluster.org/mailman/listinfo/gluster-users
Re: [Gluster-users] Gluter 3.12.12: performance during heal and in general
As per the output, all io-threads are using a lot of CPU. It is better to check what the volume profile is to see what is leading to so much work for io-threads. Please follow the documentation at https://gluster.readthedocs.io/en/latest/Administrator%20Guide/Monitoring%20Workload/ section: "Running GlusterFS Volume Profile Command"and attach output of "gluster volume profile info", On Fri, Aug 17, 2018 at 11:24 AM Hu Bert wrote: > Good morning, > > i ran the command during 100% CPU usage and attached the file. > Hopefully it helps. > > 2018-08-17 7:33 GMT+02:00 Pranith Kumar Karampuri : > > Could you do the following on one of the nodes where you are observing > high > > CPU usage and attach that file to this thread? We can find what > > threads/processes are leading to high usage. Do this for say 10 minutes > when > > you see the ~100% CPU. > > > > top -bHd 5 > /tmp/top.${HOSTNAME}.txt > > > > On Wed, Aug 15, 2018 at 2:37 PM Hu Bert wrote: > >> > >> Hello again :-) > >> > >> The self heal must have finished as there are no log entries in > >> glustershd.log files anymore. According to munin disk latency (average > >> io wait) has gone down to 100 ms, and disk utilization has gone down > >> to ~60% - both on all servers and hard disks. > >> > >> But now system load on 2 servers (which were in the good state) > >> fluctuates between 60 and 100; the server with the formerly failed > >> disk has a load of 20-30.I've uploaded some munin graphics of the cpu > >> usage: > >> > >> https://abload.de/img/gluster11_cpu31d3a.png > >> https://abload.de/img/gluster12_cpu8sem7.png > >> https://abload.de/img/gluster13_cpud7eni.png > >> > >> This can't be normal. 2 of the servers under heavy load and one not > >> that much. Does anyone have an explanation of this strange behaviour? > >> > >> > >> Thx :-) > >> > >> 2018-08-14 9:37 GMT+02:00 Hu Bert : > >> > Hi there, > >> > > >> > well, it seems the heal has finally finished. Couldn't see/find any > >> > related log message; is there such a message in a specific log file? > >> > > >> > But i see the same behaviour when the last heal finished: all CPU > >> > cores are consumed by brick processes; not only by the formerly failed > >> > bricksdd1, but by all 4 brick processes (and their threads). Load goes > >> > up to > 100 on the 2 servers with the not-failed brick, and > >> > glustershd.log gets filled with a lot of entries. Load on the server > >> > with the then failed brick not that high, but still ~60. > >> > > >> > Is this behaviour normal? Is there some post-heal after a heal has > >> > finished? > >> > > >> > thx in advance :-) > > > > > > > > -- > > Pranith > -- Pranith ___ Gluster-users mailing list Gluster-users@gluster.org https://lists.gluster.org/mailman/listinfo/gluster-users
Re: [Gluster-users] Gluter 3.12.12: performance during heal and in general
Could you do the following on one of the nodes where you are observing high CPU usage and attach that file to this thread? We can find what threads/processes are leading to high usage. Do this for say 10 minutes when you see the ~100% CPU. top -bHd 5 > /tmp/top.${HOSTNAME}.txt On Wed, Aug 15, 2018 at 2:37 PM Hu Bert wrote: > Hello again :-) > > The self heal must have finished as there are no log entries in > glustershd.log files anymore. According to munin disk latency (average > io wait) has gone down to 100 ms, and disk utilization has gone down > to ~60% - both on all servers and hard disks. > > But now system load on 2 servers (which were in the good state) > fluctuates between 60 and 100; the server with the formerly failed > disk has a load of 20-30.I've uploaded some munin graphics of the cpu > usage: > > https://abload.de/img/gluster11_cpu31d3a.png > https://abload.de/img/gluster12_cpu8sem7.png > https://abload.de/img/gluster13_cpud7eni.png > > This can't be normal. 2 of the servers under heavy load and one not > that much. Does anyone have an explanation of this strange behaviour? > > > Thx :-) > > 2018-08-14 9:37 GMT+02:00 Hu Bert : > > Hi there, > > > > well, it seems the heal has finally finished. Couldn't see/find any > > related log message; is there such a message in a specific log file? > > > > But i see the same behaviour when the last heal finished: all CPU > > cores are consumed by brick processes; not only by the formerly failed > > bricksdd1, but by all 4 brick processes (and their threads). Load goes > > up to > 100 on the 2 servers with the not-failed brick, and > > glustershd.log gets filled with a lot of entries. Load on the server > > with the then failed brick not that high, but still ~60. > > > > Is this behaviour normal? Is there some post-heal after a heal has > finished? > > > > thx in advance :-) > -- Pranith ___ Gluster-users mailing list Gluster-users@gluster.org https://lists.gluster.org/mailman/listinfo/gluster-users
Re: [Gluster-users] Gluter 3.12.12: performance during heal and in general
Hi, well, as the situation doesn't get better, we're quite helpless and mostly in the dark, so we're thinking about hiring some professional support. Any hint? :-) 2018-08-15 11:07 GMT+02:00 Hu Bert : > Hello again :-) > > The self heal must have finished as there are no log entries in > glustershd.log files anymore. According to munin disk latency (average > io wait) has gone down to 100 ms, and disk utilization has gone down > to ~60% - both on all servers and hard disks. > > But now system load on 2 servers (which were in the good state) > fluctuates between 60 and 100; the server with the formerly failed > disk has a load of 20-30.I've uploaded some munin graphics of the cpu > usage: > > https://abload.de/img/gluster11_cpu31d3a.png > https://abload.de/img/gluster12_cpu8sem7.png > https://abload.de/img/gluster13_cpud7eni.png > > This can't be normal. 2 of the servers under heavy load and one not > that much. Does anyone have an explanation of this strange behaviour? > > > Thx :-) > > 2018-08-14 9:37 GMT+02:00 Hu Bert : >> Hi there, >> >> well, it seems the heal has finally finished. Couldn't see/find any >> related log message; is there such a message in a specific log file? >> >> But i see the same behaviour when the last heal finished: all CPU >> cores are consumed by brick processes; not only by the formerly failed >> bricksdd1, but by all 4 brick processes (and their threads). Load goes >> up to > 100 on the 2 servers with the not-failed brick, and >> glustershd.log gets filled with a lot of entries. Load on the server >> with the then failed brick not that high, but still ~60. >> >> Is this behaviour normal? Is there some post-heal after a heal has finished? >> >> thx in advance :-) ___ Gluster-users mailing list Gluster-users@gluster.org https://lists.gluster.org/mailman/listinfo/gluster-users
Re: [Gluster-users] Gluter 3.12.12: performance during heal and in general
Hello again :-) The self heal must have finished as there are no log entries in glustershd.log files anymore. According to munin disk latency (average io wait) has gone down to 100 ms, and disk utilization has gone down to ~60% - both on all servers and hard disks. But now system load on 2 servers (which were in the good state) fluctuates between 60 and 100; the server with the formerly failed disk has a load of 20-30.I've uploaded some munin graphics of the cpu usage: https://abload.de/img/gluster11_cpu31d3a.png https://abload.de/img/gluster12_cpu8sem7.png https://abload.de/img/gluster13_cpud7eni.png This can't be normal. 2 of the servers under heavy load and one not that much. Does anyone have an explanation of this strange behaviour? Thx :-) 2018-08-14 9:37 GMT+02:00 Hu Bert : > Hi there, > > well, it seems the heal has finally finished. Couldn't see/find any > related log message; is there such a message in a specific log file? > > But i see the same behaviour when the last heal finished: all CPU > cores are consumed by brick processes; not only by the formerly failed > bricksdd1, but by all 4 brick processes (and their threads). Load goes > up to > 100 on the 2 servers with the not-failed brick, and > glustershd.log gets filled with a lot of entries. Load on the server > with the then failed brick not that high, but still ~60. > > Is this behaviour normal? Is there some post-heal after a heal has finished? > > thx in advance :-) ___ Gluster-users mailing list Gluster-users@gluster.org https://lists.gluster.org/mailman/listinfo/gluster-users
Re: [Gluster-users] Gluter 3.12.12: performance during heal and in general
Hi there, well, it seems the heal has finally finished. Couldn't see/find any related log message; is there such a message in a specific log file? But i see the same behaviour when the last heal finished: all CPU cores are consumed by brick processes; not only by the formerly failed bricksdd1, but by all 4 brick processes (and their threads). Load goes up to > 100 on the 2 servers with the not-failed brick, and glustershd.log gets filled with a lot of entries. Load on the server with the then failed brick not that high, but still ~60. Is this behaviour normal? Is there some post-heal after a heal has finished? thx in advance :-) ___ Gluster-users mailing list Gluster-users@gluster.org https://lists.gluster.org/mailman/listinfo/gluster-users
Re: [Gluster-users] Gluter 3.12.12: performance during heal and in general
Hello :-) Just wanted to give a short report... >> It could be saturating in the day. But if enough self-heals are going on, >> even in the night it should have been close to 100%. > > Lowest utilization was 70% over night, but i'll check this > evening/weekend. Also that 'stat...' is running. At the moment 1.1TB of 2.0TB got healed, disk utilization still between 100% (day) and 70% (night). So this will take another 10-14 days. > What, in your opinion, would be better for permormance? > > - Having 3 servers and RAID10 (with conventional disks) > - Having 3 additional servers with 4 hdds (JBOD) each (distribute > replicate, replica 3) > - SSDs? (would be quite expensive to reach the storage amount we have > at the moment) > > Just curious. It seems we'll have to adjust our setup during winter anyway :-) Well, we'll definitely rethink our setup this autumn :-) ___ Gluster-users mailing list Gluster-users@gluster.org https://lists.gluster.org/mailman/listinfo/gluster-users
Re: [Gluster-users] Gluter 3.12.12: performance during heal and in general
>> Btw.: i've seen in the munin stats that the disk utilization for >> bricksdd1 on the healthy gluster servers is between 70% (night) and >> almost 99% (daytime). So it looks like that the basic problem is the >> disk which seems not to be able to work faster? If so (heal) >> performance won't improve with this setup, i assume. > > > It could be saturating in the day. But if enough self-heals are going on, > even in the night it should have been close to 100%. Lowest utilization was 70% over night, but i'll check this evening/weekend. Also that 'stat...' is running. >> Maybe switching >> to RAID10 (conventional hard disks), SSDs or even add 3 additional >> gluster servers (distributed replicated) could help? > > > It definitely will give better protection against hardware failure. Failure > domain will be lesser. What, in your opinion, would be better for permormance? - Having 3 servers and RAID10 (with conventional disks) - Having 3 additional servers with 4 hdds (JBOD) each (distribute replicate, replica 3) - SSDs? (would be quite expensive to reach the storage amount we have at the moment) Just curious. It seems we'll have to adjust our setup during winter anyway :-) Thanx again :-) ___ Gluster-users mailing list Gluster-users@gluster.org https://lists.gluster.org/mailman/listinfo/gluster-users
Re: [Gluster-users] Gluter 3.12.12: performance during heal and in general
On Fri, Jul 27, 2018 at 1:32 PM, Hu Bert wrote: > 2018-07-27 9:22 GMT+02:00 Pranith Kumar Karampuri : > > > > > > On Fri, Jul 27, 2018 at 12:36 PM, Hu Bert > wrote: > >> > >> 2018-07-27 8:52 GMT+02:00 Pranith Kumar Karampuri >: > >> > > >> > > >> > On Fri, Jul 27, 2018 at 11:53 AM, Hu Bert > >> > wrote: > >> >> > >> >> > Do you already have all the 19 directories already created? If > >> >> > not > >> >> > could you find out which of the paths need it and do a stat > directly > >> >> > instead > >> >> > of find? > >> >> > >> >> Quite probable not all of them have been created (but counting how > >> >> much would take very long...). Hm, maybe running stat in a double > loop > >> >> (thx to our directory structure) would help. Something like this (may > >> >> be not 100% correct): > >> >> > >> >> for a in ${100..999}; do > >> >> for b in ${100..999}; do > >> >> stat /$a/$b/ > >> >> done > >> >> done > >> >> > >> >> Should run stat on all directories. I think i'll give this a try. > >> > > >> > > >> > Just to prevent these served from a cache, it is probably better to do > >> > this > >> > from a fresh mount? > >> > > >> > -- > >> > Pranith > >> > >> Good idea. I'll install glusterfs client on a little used machine, so > >> there should be no caching. Thx! Have a good weekend when the time > >> comes :-) > > > > > > If this proves effective, what you need to also do is unmount and mount > > again, something like: > > > > mount > > for a in ${100..999}; do > > for b in ${100..999}; do > > stat /$a/$b/ > > done > > done > > unmount > > I'll see what is possible over the weekend. > > Btw.: i've seen in the munin stats that the disk utilization for > bricksdd1 on the healthy gluster servers is between 70% (night) and > almost 99% (daytime). So it looks like that the basic problem is the > disk which seems not to be able to work faster? If so (heal) > performance won't improve with this setup, i assume. It could be saturating in the day. But if enough self-heals are going on, even in the night it should have been close to 100%. > Maybe switching > to RAID10 (conventional hard disks), SSDs or even add 3 additional > gluster servers (distributed replicated) could help? > It definitely will give better protection against hardware failure. Failure domain will be lesser. -- Pranith ___ Gluster-users mailing list Gluster-users@gluster.org https://lists.gluster.org/mailman/listinfo/gluster-users
Re: [Gluster-users] Gluter 3.12.12: performance during heal and in general
2018-07-27 9:22 GMT+02:00 Pranith Kumar Karampuri : > > > On Fri, Jul 27, 2018 at 12:36 PM, Hu Bert wrote: >> >> 2018-07-27 8:52 GMT+02:00 Pranith Kumar Karampuri : >> > >> > >> > On Fri, Jul 27, 2018 at 11:53 AM, Hu Bert >> > wrote: >> >> >> >> > Do you already have all the 19 directories already created? If >> >> > not >> >> > could you find out which of the paths need it and do a stat directly >> >> > instead >> >> > of find? >> >> >> >> Quite probable not all of them have been created (but counting how >> >> much would take very long...). Hm, maybe running stat in a double loop >> >> (thx to our directory structure) would help. Something like this (may >> >> be not 100% correct): >> >> >> >> for a in ${100..999}; do >> >> for b in ${100..999}; do >> >> stat /$a/$b/ >> >> done >> >> done >> >> >> >> Should run stat on all directories. I think i'll give this a try. >> > >> > >> > Just to prevent these served from a cache, it is probably better to do >> > this >> > from a fresh mount? >> > >> > -- >> > Pranith >> >> Good idea. I'll install glusterfs client on a little used machine, so >> there should be no caching. Thx! Have a good weekend when the time >> comes :-) > > > If this proves effective, what you need to also do is unmount and mount > again, something like: > > mount > for a in ${100..999}; do > for b in ${100..999}; do > stat /$a/$b/ > done > done > unmount I'll see what is possible over the weekend. Btw.: i've seen in the munin stats that the disk utilization for bricksdd1 on the healthy gluster servers is between 70% (night) and almost 99% (daytime). So it looks like that the basic problem is the disk which seems not to be able to work faster? If so (heal) performance won't improve with this setup, i assume. Maybe switching to RAID10 (conventional hard disks), SSDs or even add 3 additional gluster servers (distributed replicated) could help? ___ Gluster-users mailing list Gluster-users@gluster.org https://lists.gluster.org/mailman/listinfo/gluster-users
Re: [Gluster-users] Gluter 3.12.12: performance during heal and in general
On Fri, Jul 27, 2018 at 12:36 PM, Hu Bert wrote: > 2018-07-27 8:52 GMT+02:00 Pranith Kumar Karampuri : > > > > > > On Fri, Jul 27, 2018 at 11:53 AM, Hu Bert > wrote: > >> > >> > Do you already have all the 19 directories already created? If not > >> > could you find out which of the paths need it and do a stat directly > instead > >> > of find? > >> > >> Quite probable not all of them have been created (but counting how > >> much would take very long...). Hm, maybe running stat in a double loop > >> (thx to our directory structure) would help. Something like this (may > >> be not 100% correct): > >> > >> for a in ${100..999}; do > >> for b in ${100..999}; do > >> stat /$a/$b/ > >> done > >> done > >> > >> Should run stat on all directories. I think i'll give this a try. > > > > > > Just to prevent these served from a cache, it is probably better to do > this > > from a fresh mount? > > > > -- > > Pranith > > Good idea. I'll install glusterfs client on a little used machine, so > there should be no caching. Thx! Have a good weekend when the time > comes :-) > If this proves effective, what you need to also do is unmount and mount again, something like: mount for a in ${100..999}; do for b in ${100..999}; do stat /$a/$b/ done done unmount -- Pranith ___ Gluster-users mailing list Gluster-users@gluster.org https://lists.gluster.org/mailman/listinfo/gluster-users
Re: [Gluster-users] Gluter 3.12.12: performance during heal and in general
2018-07-27 8:52 GMT+02:00 Pranith Kumar Karampuri : > > > On Fri, Jul 27, 2018 at 11:53 AM, Hu Bert wrote: >> >> > Do you already have all the 19 directories already created? If not >> > could you find out which of the paths need it and do a stat directly >> > instead >> > of find? >> >> Quite probable not all of them have been created (but counting how >> much would take very long...). Hm, maybe running stat in a double loop >> (thx to our directory structure) would help. Something like this (may >> be not 100% correct): >> >> for a in ${100..999}; do >> for b in ${100..999}; do >> stat /$a/$b/ >> done >> done >> >> Should run stat on all directories. I think i'll give this a try. > > > Just to prevent these served from a cache, it is probably better to do this > from a fresh mount? > > -- > Pranith Good idea. I'll install glusterfs client on a little used machine, so there should be no caching. Thx! Have a good weekend when the time comes :-) ___ Gluster-users mailing list Gluster-users@gluster.org https://lists.gluster.org/mailman/listinfo/gluster-users
Re: [Gluster-users] Gluter 3.12.12: performance during heal and in general
On Fri, Jul 27, 2018 at 11:53 AM, Hu Bert wrote: > > Do you already have all the 19 directories already created? If not > could you find out which of the paths need it and do a stat directly > instead of find? > > Quite probable not all of them have been created (but counting how > much would take very long...). Hm, maybe running stat in a double loop > (thx to our directory structure) would help. Something like this (may > be not 100% correct): > > for a in ${100..999}; do > for b in ${100..999}; do > stat /$a/$b/ > done > done > > Should run stat on all directories. I think i'll give this a try. > Just to prevent these served from a cache, it is probably better to do this from a fresh mount? -- Pranith ___ Gluster-users mailing list Gluster-users@gluster.org https://lists.gluster.org/mailman/listinfo/gluster-users
Re: [Gluster-users] Gluter 3.12.12: performance during heal and in general
> Do you already have all the 19 directories already created? If not could > you find out which of the paths need it and do a stat directly instead of > find? Quite probable not all of them have been created (but counting how much would take very long...). Hm, maybe running stat in a double loop (thx to our directory structure) would help. Something like this (may be not 100% correct): for a in ${100..999}; do for b in ${100..999}; do stat /$a/$b/ done done Should run stat on all directories. I think i'll give this a try. ___ Gluster-users mailing list Gluster-users@gluster.org https://lists.gluster.org/mailman/listinfo/gluster-users
Re: [Gluster-users] Gluter 3.12.12: performance during heal and in general
On Fri, Jul 27, 2018 at 11:11 AM, Hu Bert wrote: > Good Morning :-) > > on server gluster11 about 1.25 million and on gluster13 about 1.35 > million log entries in glustershd.log file. About 70 GB got healed, > overall ~700GB of 2.0TB. Doesn't seem to run faster. I'm calling > 'find...' whenever i notice that it has finished. Hmm... is it > possible and reasonable to run 2 finds in parallel, maybe on different > subdirectories? E.g. running one one $volume/public/ and on one > $volume/private/ ? > Do you already have all the 19 directories already created? If not could you find out which of the paths need it and do a stat directly instead of find? > > 2018-07-26 11:29 GMT+02:00 Pranith Kumar Karampuri : > > > > > > On Thu, Jul 26, 2018 at 2:41 PM, Hu Bert wrote: > >> > >> > Sorry, bad copy/paste :-(. > >> > >> np :-) > >> > >> The question regarding version 4.1 was meant more generally: does > >> gluster v4.0 etc. have a better performance than version 3.12 etc.? > >> Just curious :-) Sooner or later we have to upgrade anyway. > > > > > > You can check what changed @ > > https://github.com/gluster/glusterfs/blob/release-4.0/ > doc/release-notes/4.0.0.md#performance > > https://github.com/gluster/glusterfs/blob/release-4.1/ > doc/release-notes/4.1.0.md#performance > > > >> > >> > >> btw.: gluster12 was the node with the failed brick, and i started the > >> full heal on this node (has the biggest uuid as well). Is it normal > >> that the glustershd.log on this node is rather empty (some hundred > >> entries), but the glustershd.log files on the 2 other nodes have > >> hundreds of thousands of entries? > > > > > > heals happen on the good bricks, so this is expected. > > > >> > >> (sry, mail twice, didn't go to the list, but maybe others are > >> interested... :-) ) > >> > >> 2018-07-26 10:17 GMT+02:00 Pranith Kumar Karampuri >: > >> > > >> > > >> > On Thu, Jul 26, 2018 at 12:59 PM, Hu Bert > >> > wrote: > >> >> > >> >> Hi Pranith, > >> >> > >> >> thanks a lot for your efforts and for tracking "my" problem with an > >> >> issue. > >> >> :-) > >> >> > >> >> > >> >> I've set this params on the gluster volume and will start the > >> >> 'find...' command within a short time. I'll probably add another > >> >> answer to the list to document the progress. > >> >> > >> >> btw. - you had some typos: > >> >> gluster volume set cluster.cluster.heal-wait-queue-length > >> >> 1 => cluster is doubled > >> >> gluster volume set cluster.data-self-heal-window-size 16 > => > >> >> it's actually cluster.self-heal-window-size > >> >> > >> >> but actually no problem :-) > >> > > >> > > >> > Sorry, bad copy/paste :-(. > >> > > >> >> > >> >> > >> >> Just curious: would gluster 4.1 improve the performance for healing > >> >> and in general for "my" scenario? > >> > > >> > > >> > No, this issue is present in all the existing releases. But it is > >> > solvable. > >> > You can follow that issue to see progress and when it is fixed etc. > >> > > >> >> > >> >> > >> >> 2018-07-26 8:56 GMT+02:00 Pranith Kumar Karampuri > >> >> : > >> >> > Thanks a lot for detailed write-up, this helps find the bottlenecks > >> >> > easily. > >> >> > On a high level, to handle this directory hierarchy i.e. lots of > >> >> > directories > >> >> > with files, we need to improve healing > >> >> > algorithms. Based on the data you provided, we need to make the > >> >> > following > >> >> > enhancements: > >> >> > > >> >> > 1) At the moment directories are healed one at a time, but files > can > >> >> > be > >> >> > healed upto 64 in parallel per replica subvolume. > >> >> > So if you have nX2 or nX3 distributed subvolumes, it can heal 64n > >> >> > number > >> >> > of > >> >> > files in parallel. > >> >> > > >> >> > I raised https://github.com/gluster/glusterfs/issues/477 to track > >> >> > this. > >> >> > In > >> >> > the mean-while you can use the following work-around: > >> >> > a) Increase background heals on the mount: > >> >> > gluster volume set cluster.background-self-heal-count > 256 > >> >> > gluster volume set cluster.cluster.heal-wait- > queue-length > >> >> > 1 > >> >> > find -type d | xargs stat > >> >> > > >> >> > one 'find' will trigger 10256 directories. So you may have to do > this > >> >> > periodically until all directories are healed. > >> >> > > >> >> > 2) Self-heal heals a file 128KB at a > >> >> > time(data-self-heal-window-size). I > >> >> > think for your environment bumping upto MBs is better. Say 2MB i.e. > >> >> > 16*128KB? > >> >> > > >> >> > Command to do that is: > >> >> > gluster volume set cluster.data-self-heal-window-size 16 > >> >> > > >> >> > > >> >> > On Thu, Jul 26, 2018 at 10:40 AM, Hu Bert > >> >> > wrote: > >> >> >> > >> >> >> Hi Pranith, > >> >> >> > >> >> >> Sry, it took a while to count the directories. I'll try to answer > >> >> >> your > >> >> >> questions as good as possible. > >> >> >> > >> >> >> > What kind of data do you have? > >> >> >> > How many directories in the filesystem? >
Re: [Gluster-users] Gluter 3.12.12: performance during heal and in general
Good Morning :-) on server gluster11 about 1.25 million and on gluster13 about 1.35 million log entries in glustershd.log file. About 70 GB got healed, overall ~700GB of 2.0TB. Doesn't seem to run faster. I'm calling 'find...' whenever i notice that it has finished. Hmm... is it possible and reasonable to run 2 finds in parallel, maybe on different subdirectories? E.g. running one one $volume/public/ and on one $volume/private/ ? 2018-07-26 11:29 GMT+02:00 Pranith Kumar Karampuri : > > > On Thu, Jul 26, 2018 at 2:41 PM, Hu Bert wrote: >> >> > Sorry, bad copy/paste :-(. >> >> np :-) >> >> The question regarding version 4.1 was meant more generally: does >> gluster v4.0 etc. have a better performance than version 3.12 etc.? >> Just curious :-) Sooner or later we have to upgrade anyway. > > > You can check what changed @ > https://github.com/gluster/glusterfs/blob/release-4.0/doc/release-notes/4.0.0.md#performance > https://github.com/gluster/glusterfs/blob/release-4.1/doc/release-notes/4.1.0.md#performance > >> >> >> btw.: gluster12 was the node with the failed brick, and i started the >> full heal on this node (has the biggest uuid as well). Is it normal >> that the glustershd.log on this node is rather empty (some hundred >> entries), but the glustershd.log files on the 2 other nodes have >> hundreds of thousands of entries? > > > heals happen on the good bricks, so this is expected. > >> >> (sry, mail twice, didn't go to the list, but maybe others are >> interested... :-) ) >> >> 2018-07-26 10:17 GMT+02:00 Pranith Kumar Karampuri : >> > >> > >> > On Thu, Jul 26, 2018 at 12:59 PM, Hu Bert >> > wrote: >> >> >> >> Hi Pranith, >> >> >> >> thanks a lot for your efforts and for tracking "my" problem with an >> >> issue. >> >> :-) >> >> >> >> >> >> I've set this params on the gluster volume and will start the >> >> 'find...' command within a short time. I'll probably add another >> >> answer to the list to document the progress. >> >> >> >> btw. - you had some typos: >> >> gluster volume set cluster.cluster.heal-wait-queue-length >> >> 1 => cluster is doubled >> >> gluster volume set cluster.data-self-heal-window-size 16 => >> >> it's actually cluster.self-heal-window-size >> >> >> >> but actually no problem :-) >> > >> > >> > Sorry, bad copy/paste :-(. >> > >> >> >> >> >> >> Just curious: would gluster 4.1 improve the performance for healing >> >> and in general for "my" scenario? >> > >> > >> > No, this issue is present in all the existing releases. But it is >> > solvable. >> > You can follow that issue to see progress and when it is fixed etc. >> > >> >> >> >> >> >> 2018-07-26 8:56 GMT+02:00 Pranith Kumar Karampuri >> >> : >> >> > Thanks a lot for detailed write-up, this helps find the bottlenecks >> >> > easily. >> >> > On a high level, to handle this directory hierarchy i.e. lots of >> >> > directories >> >> > with files, we need to improve healing >> >> > algorithms. Based on the data you provided, we need to make the >> >> > following >> >> > enhancements: >> >> > >> >> > 1) At the moment directories are healed one at a time, but files can >> >> > be >> >> > healed upto 64 in parallel per replica subvolume. >> >> > So if you have nX2 or nX3 distributed subvolumes, it can heal 64n >> >> > number >> >> > of >> >> > files in parallel. >> >> > >> >> > I raised https://github.com/gluster/glusterfs/issues/477 to track >> >> > this. >> >> > In >> >> > the mean-while you can use the following work-around: >> >> > a) Increase background heals on the mount: >> >> > gluster volume set cluster.background-self-heal-count 256 >> >> > gluster volume set cluster.cluster.heal-wait-queue-length >> >> > 1 >> >> > find -type d | xargs stat >> >> > >> >> > one 'find' will trigger 10256 directories. So you may have to do this >> >> > periodically until all directories are healed. >> >> > >> >> > 2) Self-heal heals a file 128KB at a >> >> > time(data-self-heal-window-size). I >> >> > think for your environment bumping upto MBs is better. Say 2MB i.e. >> >> > 16*128KB? >> >> > >> >> > Command to do that is: >> >> > gluster volume set cluster.data-self-heal-window-size 16 >> >> > >> >> > >> >> > On Thu, Jul 26, 2018 at 10:40 AM, Hu Bert >> >> > wrote: >> >> >> >> >> >> Hi Pranith, >> >> >> >> >> >> Sry, it took a while to count the directories. I'll try to answer >> >> >> your >> >> >> questions as good as possible. >> >> >> >> >> >> > What kind of data do you have? >> >> >> > How many directories in the filesystem? >> >> >> > On average how many files per directory? >> >> >> > What is the depth of your directory hierarchy on average? >> >> >> > What is average filesize? >> >> >> >> >> >> We have mostly images (more than 95% of disk usage, 90% of file >> >> >> count), some text files (like css, jsp, gpx etc.) and some binaries. >> >> >> >> >> >> There are about 190.000 directories in the file system; maybe there >> >> >> are some more because we're hit by bug 1512371 (parallel-readdir = >> >> >> TRUE prevents
Re: [Gluster-users] Gluter 3.12.12: performance during heal and in general
On Thu, Jul 26, 2018 at 2:41 PM, Hu Bert wrote: > > Sorry, bad copy/paste :-(. > > np :-) > > The question regarding version 4.1 was meant more generally: does > gluster v4.0 etc. have a better performance than version 3.12 etc.? > Just curious :-) Sooner or later we have to upgrade anyway. You can check what changed @ https://github.com/gluster/glusterfs/blob/release-4.0/doc/release-notes/4.0.0.md#performance https://github.com/gluster/glusterfs/blob/release-4.1/doc/release-notes/4.1.0.md#performance > > > btw.: gluster12 was the node with the failed brick, and i started the > full heal on this node (has the biggest uuid as well). Is it normal > that the glustershd.log on this node is rather empty (some hundred > entries), but the glustershd.log files on the 2 other nodes have > hundreds of thousands of entries? > heals happen on the good bricks, so this is expected. > (sry, mail twice, didn't go to the list, but maybe others are > interested... :-) ) > > 2018-07-26 10:17 GMT+02:00 Pranith Kumar Karampuri : > > > > > > On Thu, Jul 26, 2018 at 12:59 PM, Hu Bert > wrote: > >> > >> Hi Pranith, > >> > >> thanks a lot for your efforts and for tracking "my" problem with an > issue. > >> :-) > >> > >> > >> I've set this params on the gluster volume and will start the > >> 'find...' command within a short time. I'll probably add another > >> answer to the list to document the progress. > >> > >> btw. - you had some typos: > >> gluster volume set cluster.cluster.heal-wait-queue-length > >> 1 => cluster is doubled > >> gluster volume set cluster.data-self-heal-window-size 16 => > >> it's actually cluster.self-heal-window-size > >> > >> but actually no problem :-) > > > > > > Sorry, bad copy/paste :-(. > > > >> > >> > >> Just curious: would gluster 4.1 improve the performance for healing > >> and in general for "my" scenario? > > > > > > No, this issue is present in all the existing releases. But it is > solvable. > > You can follow that issue to see progress and when it is fixed etc. > > > >> > >> > >> 2018-07-26 8:56 GMT+02:00 Pranith Kumar Karampuri >: > >> > Thanks a lot for detailed write-up, this helps find the bottlenecks > >> > easily. > >> > On a high level, to handle this directory hierarchy i.e. lots of > >> > directories > >> > with files, we need to improve healing > >> > algorithms. Based on the data you provided, we need to make the > >> > following > >> > enhancements: > >> > > >> > 1) At the moment directories are healed one at a time, but files can > be > >> > healed upto 64 in parallel per replica subvolume. > >> > So if you have nX2 or nX3 distributed subvolumes, it can heal 64n > number > >> > of > >> > files in parallel. > >> > > >> > I raised https://github.com/gluster/glusterfs/issues/477 to track > this. > >> > In > >> > the mean-while you can use the following work-around: > >> > a) Increase background heals on the mount: > >> > gluster volume set cluster.background-self-heal-count 256 > >> > gluster volume set cluster.cluster.heal-wait-queue-length > >> > 1 > >> > find -type d | xargs stat > >> > > >> > one 'find' will trigger 10256 directories. So you may have to do this > >> > periodically until all directories are healed. > >> > > >> > 2) Self-heal heals a file 128KB at a time(data-self-heal-window-size). > I > >> > think for your environment bumping upto MBs is better. Say 2MB i.e. > >> > 16*128KB? > >> > > >> > Command to do that is: > >> > gluster volume set cluster.data-self-heal-window-size 16 > >> > > >> > > >> > On Thu, Jul 26, 2018 at 10:40 AM, Hu Bert > >> > wrote: > >> >> > >> >> Hi Pranith, > >> >> > >> >> Sry, it took a while to count the directories. I'll try to answer > your > >> >> questions as good as possible. > >> >> > >> >> > What kind of data do you have? > >> >> > How many directories in the filesystem? > >> >> > On average how many files per directory? > >> >> > What is the depth of your directory hierarchy on average? > >> >> > What is average filesize? > >> >> > >> >> We have mostly images (more than 95% of disk usage, 90% of file > >> >> count), some text files (like css, jsp, gpx etc.) and some binaries. > >> >> > >> >> There are about 190.000 directories in the file system; maybe there > >> >> are some more because we're hit by bug 1512371 (parallel-readdir = > >> >> TRUE prevents directories listing). But the number of directories > >> >> could/will rise in the future (maybe millions). > >> >> > >> >> files per directory: ranges from 0 to 100, on average it should be 20 > >> >> files per directory (well, at least in the deepest dirs, see > >> >> explanation below). > >> >> > >> >> Average filesize: ranges from a few hundred bytes up to 30 MB, on > >> >> average it should be 2-3 MB. > >> >> > >> >> Directory hierarchy: maximum depth as seen from within the volume is > >> >> 6, the average should be 3. > >> >> > >> >> volume name: shared > >> >> mount point on clients: /data/repository/shared/ > >> >> below /shared/ there are 2
Re: [Gluster-users] Gluter 3.12.12: performance during heal and in general
> Sorry, bad copy/paste :-(. np :-) The question regarding version 4.1 was meant more generally: does gluster v4.0 etc. have a better performance than version 3.12 etc.? Just curious :-) Sooner or later we have to upgrade anyway. btw.: gluster12 was the node with the failed brick, and i started the full heal on this node (has the biggest uuid as well). Is it normal that the glustershd.log on this node is rather empty (some hundred entries), but the glustershd.log files on the 2 other nodes have hundreds of thousands of entries? (sry, mail twice, didn't go to the list, but maybe others are interested... :-) ) 2018-07-26 10:17 GMT+02:00 Pranith Kumar Karampuri : > > > On Thu, Jul 26, 2018 at 12:59 PM, Hu Bert wrote: >> >> Hi Pranith, >> >> thanks a lot for your efforts and for tracking "my" problem with an issue. >> :-) >> >> >> I've set this params on the gluster volume and will start the >> 'find...' command within a short time. I'll probably add another >> answer to the list to document the progress. >> >> btw. - you had some typos: >> gluster volume set cluster.cluster.heal-wait-queue-length >> 1 => cluster is doubled >> gluster volume set cluster.data-self-heal-window-size 16 => >> it's actually cluster.self-heal-window-size >> >> but actually no problem :-) > > > Sorry, bad copy/paste :-(. > >> >> >> Just curious: would gluster 4.1 improve the performance for healing >> and in general for "my" scenario? > > > No, this issue is present in all the existing releases. But it is solvable. > You can follow that issue to see progress and when it is fixed etc. > >> >> >> 2018-07-26 8:56 GMT+02:00 Pranith Kumar Karampuri : >> > Thanks a lot for detailed write-up, this helps find the bottlenecks >> > easily. >> > On a high level, to handle this directory hierarchy i.e. lots of >> > directories >> > with files, we need to improve healing >> > algorithms. Based on the data you provided, we need to make the >> > following >> > enhancements: >> > >> > 1) At the moment directories are healed one at a time, but files can be >> > healed upto 64 in parallel per replica subvolume. >> > So if you have nX2 or nX3 distributed subvolumes, it can heal 64n number >> > of >> > files in parallel. >> > >> > I raised https://github.com/gluster/glusterfs/issues/477 to track this. >> > In >> > the mean-while you can use the following work-around: >> > a) Increase background heals on the mount: >> > gluster volume set cluster.background-self-heal-count 256 >> > gluster volume set cluster.cluster.heal-wait-queue-length >> > 1 >> > find -type d | xargs stat >> > >> > one 'find' will trigger 10256 directories. So you may have to do this >> > periodically until all directories are healed. >> > >> > 2) Self-heal heals a file 128KB at a time(data-self-heal-window-size). I >> > think for your environment bumping upto MBs is better. Say 2MB i.e. >> > 16*128KB? >> > >> > Command to do that is: >> > gluster volume set cluster.data-self-heal-window-size 16 >> > >> > >> > On Thu, Jul 26, 2018 at 10:40 AM, Hu Bert >> > wrote: >> >> >> >> Hi Pranith, >> >> >> >> Sry, it took a while to count the directories. I'll try to answer your >> >> questions as good as possible. >> >> >> >> > What kind of data do you have? >> >> > How many directories in the filesystem? >> >> > On average how many files per directory? >> >> > What is the depth of your directory hierarchy on average? >> >> > What is average filesize? >> >> >> >> We have mostly images (more than 95% of disk usage, 90% of file >> >> count), some text files (like css, jsp, gpx etc.) and some binaries. >> >> >> >> There are about 190.000 directories in the file system; maybe there >> >> are some more because we're hit by bug 1512371 (parallel-readdir = >> >> TRUE prevents directories listing). But the number of directories >> >> could/will rise in the future (maybe millions). >> >> >> >> files per directory: ranges from 0 to 100, on average it should be 20 >> >> files per directory (well, at least in the deepest dirs, see >> >> explanation below). >> >> >> >> Average filesize: ranges from a few hundred bytes up to 30 MB, on >> >> average it should be 2-3 MB. >> >> >> >> Directory hierarchy: maximum depth as seen from within the volume is >> >> 6, the average should be 3. >> >> >> >> volume name: shared >> >> mount point on clients: /data/repository/shared/ >> >> below /shared/ there are 2 directories: >> >> - public/: mainly calculated images (file sizes from a few KB up to >> >> max 1 MB) and some resouces (small PNGs with a size of a few hundred >> >> bytes). >> >> - private/: mainly source images; file sizes from 50 KB up to 30MB >> >> >> >> We migrated from a NFS server (SPOF) to glusterfs and simply copied >> >> our files. The images (which have an ID) are stored in the deepest >> >> directories of the dir tree. I'll better explain it :-) >> >> >> >> directory structure for the images (i'll omit some other miscellaneous >> >> stuff, but it looks quite similar): >> >> -
Re: [Gluster-users] Gluter 3.12.12: performance during heal and in general
On Thu, Jul 26, 2018 at 12:59 PM, Hu Bert wrote: > Hi Pranith, > > thanks a lot for your efforts and for tracking "my" problem with an issue. > :-) > > I've set this params on the gluster volume and will start the > 'find...' command within a short time. I'll probably add another > answer to the list to document the progress. > > btw. - you had some typos: > gluster volume set cluster.cluster.heal-wait-queue-length > 1 => cluster is doubled > gluster volume set cluster.data-self-heal-window-size 16 => > it's actually cluster.self-heal-window-size > > but actually no problem :-) > Sorry, bad copy/paste :-(. > > Just curious: would gluster 4.1 improve the performance for healing > and in general for "my" scenario? > No, this issue is present in all the existing releases. But it is solvable. You can follow that issue to see progress and when it is fixed etc. > > 2018-07-26 8:56 GMT+02:00 Pranith Kumar Karampuri : > > Thanks a lot for detailed write-up, this helps find the bottlenecks > easily. > > On a high level, to handle this directory hierarchy i.e. lots of > directories > > with files, we need to improve healing > > algorithms. Based on the data you provided, we need to make the following > > enhancements: > > > > 1) At the moment directories are healed one at a time, but files can be > > healed upto 64 in parallel per replica subvolume. > > So if you have nX2 or nX3 distributed subvolumes, it can heal 64n number > of > > files in parallel. > > > > I raised https://github.com/gluster/glusterfs/issues/477 to track this. > In > > the mean-while you can use the following work-around: > > a) Increase background heals on the mount: > > gluster volume set cluster.background-self-heal-count 256 > > gluster volume set cluster.cluster.heal-wait-queue-length > 1 > > find -type d | xargs stat > > > > one 'find' will trigger 10256 directories. So you may have to do this > > periodically until all directories are healed. > > > > 2) Self-heal heals a file 128KB at a time(data-self-heal-window-size). I > > think for your environment bumping upto MBs is better. Say 2MB i.e. > > 16*128KB? > > > > Command to do that is: > > gluster volume set cluster.data-self-heal-window-size 16 > > > > > > On Thu, Jul 26, 2018 at 10:40 AM, Hu Bert > wrote: > >> > >> Hi Pranith, > >> > >> Sry, it took a while to count the directories. I'll try to answer your > >> questions as good as possible. > >> > >> > What kind of data do you have? > >> > How many directories in the filesystem? > >> > On average how many files per directory? > >> > What is the depth of your directory hierarchy on average? > >> > What is average filesize? > >> > >> We have mostly images (more than 95% of disk usage, 90% of file > >> count), some text files (like css, jsp, gpx etc.) and some binaries. > >> > >> There are about 190.000 directories in the file system; maybe there > >> are some more because we're hit by bug 1512371 (parallel-readdir = > >> TRUE prevents directories listing). But the number of directories > >> could/will rise in the future (maybe millions). > >> > >> files per directory: ranges from 0 to 100, on average it should be 20 > >> files per directory (well, at least in the deepest dirs, see > >> explanation below). > >> > >> Average filesize: ranges from a few hundred bytes up to 30 MB, on > >> average it should be 2-3 MB. > >> > >> Directory hierarchy: maximum depth as seen from within the volume is > >> 6, the average should be 3. > >> > >> volume name: shared > >> mount point on clients: /data/repository/shared/ > >> below /shared/ there are 2 directories: > >> - public/: mainly calculated images (file sizes from a few KB up to > >> max 1 MB) and some resouces (small PNGs with a size of a few hundred > >> bytes). > >> - private/: mainly source images; file sizes from 50 KB up to 30MB > >> > >> We migrated from a NFS server (SPOF) to glusterfs and simply copied > >> our files. The images (which have an ID) are stored in the deepest > >> directories of the dir tree. I'll better explain it :-) > >> > >> directory structure for the images (i'll omit some other miscellaneous > >> stuff, but it looks quite similar): > >> - ID of an image has 7 or 8 digits > >> - /shared/private/: /(first 3 digits of ID)/(next 3 digits of > ID)/$ID.jpg > >> - /shared/public/: /(first 3 digits of ID)/(next 3 digits of > >> ID)/$ID/$misc_formats.jpg > >> > >> That's why we have that many (sub-)directories. Files are only stored > >> in the lowest directory hierarchy. I hope i could make our structure > >> at least a bit more transparent. > >> > >> i hope there's something we can do to raise performance a bit. thx in > >> advance :-) > >> > >> > >> 2018-07-24 10:40 GMT+02:00 Pranith Kumar Karampuri >: > >> > > >> > > >> > On Mon, Jul 23, 2018 at 4:16 PM, Hu Bert > wrote: > >> >> > >> >> Well, over the weekend about 200GB were copied, so now there are > >> >> ~400GB copied to the brick. That's far beyond a speed of 10GB per > >> >> hour. If I
Re: [Gluster-users] Gluter 3.12.12: performance during heal and in general
Hi Pranith, thanks a lot for your efforts and for tracking "my" problem with an issue. :-) I've set this params on the gluster volume and will start the 'find...' command within a short time. I'll probably add another answer to the list to document the progress. btw. - you had some typos: gluster volume set cluster.cluster.heal-wait-queue-length 1 => cluster is doubled gluster volume set cluster.data-self-heal-window-size 16 => it's actually cluster.self-heal-window-size but actually no problem :-) Just curious: would gluster 4.1 improve the performance for healing and in general for "my" scenario? 2018-07-26 8:56 GMT+02:00 Pranith Kumar Karampuri : > Thanks a lot for detailed write-up, this helps find the bottlenecks easily. > On a high level, to handle this directory hierarchy i.e. lots of directories > with files, we need to improve healing > algorithms. Based on the data you provided, we need to make the following > enhancements: > > 1) At the moment directories are healed one at a time, but files can be > healed upto 64 in parallel per replica subvolume. > So if you have nX2 or nX3 distributed subvolumes, it can heal 64n number of > files in parallel. > > I raised https://github.com/gluster/glusterfs/issues/477 to track this. In > the mean-while you can use the following work-around: > a) Increase background heals on the mount: > gluster volume set cluster.background-self-heal-count 256 > gluster volume set cluster.cluster.heal-wait-queue-length 1 > find -type d | xargs stat > > one 'find' will trigger 10256 directories. So you may have to do this > periodically until all directories are healed. > > 2) Self-heal heals a file 128KB at a time(data-self-heal-window-size). I > think for your environment bumping upto MBs is better. Say 2MB i.e. > 16*128KB? > > Command to do that is: > gluster volume set cluster.data-self-heal-window-size 16 > > > On Thu, Jul 26, 2018 at 10:40 AM, Hu Bert wrote: >> >> Hi Pranith, >> >> Sry, it took a while to count the directories. I'll try to answer your >> questions as good as possible. >> >> > What kind of data do you have? >> > How many directories in the filesystem? >> > On average how many files per directory? >> > What is the depth of your directory hierarchy on average? >> > What is average filesize? >> >> We have mostly images (more than 95% of disk usage, 90% of file >> count), some text files (like css, jsp, gpx etc.) and some binaries. >> >> There are about 190.000 directories in the file system; maybe there >> are some more because we're hit by bug 1512371 (parallel-readdir = >> TRUE prevents directories listing). But the number of directories >> could/will rise in the future (maybe millions). >> >> files per directory: ranges from 0 to 100, on average it should be 20 >> files per directory (well, at least in the deepest dirs, see >> explanation below). >> >> Average filesize: ranges from a few hundred bytes up to 30 MB, on >> average it should be 2-3 MB. >> >> Directory hierarchy: maximum depth as seen from within the volume is >> 6, the average should be 3. >> >> volume name: shared >> mount point on clients: /data/repository/shared/ >> below /shared/ there are 2 directories: >> - public/: mainly calculated images (file sizes from a few KB up to >> max 1 MB) and some resouces (small PNGs with a size of a few hundred >> bytes). >> - private/: mainly source images; file sizes from 50 KB up to 30MB >> >> We migrated from a NFS server (SPOF) to glusterfs and simply copied >> our files. The images (which have an ID) are stored in the deepest >> directories of the dir tree. I'll better explain it :-) >> >> directory structure for the images (i'll omit some other miscellaneous >> stuff, but it looks quite similar): >> - ID of an image has 7 or 8 digits >> - /shared/private/: /(first 3 digits of ID)/(next 3 digits of ID)/$ID.jpg >> - /shared/public/: /(first 3 digits of ID)/(next 3 digits of >> ID)/$ID/$misc_formats.jpg >> >> That's why we have that many (sub-)directories. Files are only stored >> in the lowest directory hierarchy. I hope i could make our structure >> at least a bit more transparent. >> >> i hope there's something we can do to raise performance a bit. thx in >> advance :-) >> >> >> 2018-07-24 10:40 GMT+02:00 Pranith Kumar Karampuri : >> > >> > >> > On Mon, Jul 23, 2018 at 4:16 PM, Hu Bert wrote: >> >> >> >> Well, over the weekend about 200GB were copied, so now there are >> >> ~400GB copied to the brick. That's far beyond a speed of 10GB per >> >> hour. If I copied the 1.6 TB directly, that would be done within max 2 >> >> days. But with the self heal this will take at least 20 days minimum. >> >> >> >> Why is the performance that bad? No chance of speeding this up? >> > >> > >> > What kind of data do you have? >> > How many directories in the filesystem? >> > On average how many files per directory? >> > What is the depth of your directory hierarchy on average? >> > What is average filesize? >> > >> > Based on this data we can
Re: [Gluster-users] Gluter 3.12.12: performance during heal and in general
Thanks a lot for detailed write-up, this helps find the bottlenecks easily. On a high level, to handle this directory hierarchy i.e. lots of directories with files, we need to improve healing algorithms. Based on the data you provided, we need to make the following enhancements: 1) At the moment directories are healed one at a time, but files can be healed upto 64 in parallel per replica subvolume. So if you have nX2 or nX3 distributed subvolumes, it can heal 64n number of files in parallel. I raised https://github.com/gluster/glusterfs/issues/477 to track this. In the mean-while you can use the following work-around: a) Increase background heals on the mount: gluster volume set cluster.background-self-heal-count 256 gluster volume set cluster.cluster.heal-wait-queue-length 1 find -type d | xargs stat one 'find' will trigger 10256 directories. So you may have to do this periodically until all directories are healed. 2) Self-heal heals a file 128KB at a time(data-self-heal-window-size). I think for your environment bumping upto MBs is better. Say 2MB i.e. 16*128KB? Command to do that is: gluster volume set cluster.data-self-heal-window-size 16 On Thu, Jul 26, 2018 at 10:40 AM, Hu Bert wrote: > Hi Pranith, > > Sry, it took a while to count the directories. I'll try to answer your > questions as good as possible. > > > What kind of data do you have? > > How many directories in the filesystem? > > On average how many files per directory? > > What is the depth of your directory hierarchy on average? > > What is average filesize? > > We have mostly images (more than 95% of disk usage, 90% of file > count), some text files (like css, jsp, gpx etc.) and some binaries. > > There are about 190.000 directories in the file system; maybe there > are some more because we're hit by bug 1512371 (parallel-readdir = > TRUE prevents directories listing). But the number of directories > could/will rise in the future (maybe millions). > > files per directory: ranges from 0 to 100, on average it should be 20 > files per directory (well, at least in the deepest dirs, see > explanation below). > > Average filesize: ranges from a few hundred bytes up to 30 MB, on > average it should be 2-3 MB. > > Directory hierarchy: maximum depth as seen from within the volume is > 6, the average should be 3. > > volume name: shared > mount point on clients: /data/repository/shared/ > below /shared/ there are 2 directories: > - public/: mainly calculated images (file sizes from a few KB up to > max 1 MB) and some resouces (small PNGs with a size of a few hundred > bytes). > - private/: mainly source images; file sizes from 50 KB up to 30MB > > We migrated from a NFS server (SPOF) to glusterfs and simply copied > our files. The images (which have an ID) are stored in the deepest > directories of the dir tree. I'll better explain it :-) > > directory structure for the images (i'll omit some other miscellaneous > stuff, but it looks quite similar): > - ID of an image has 7 or 8 digits > - /shared/private/: /(first 3 digits of ID)/(next 3 digits of ID)/$ID.jpg > - /shared/public/: /(first 3 digits of ID)/(next 3 digits of > ID)/$ID/$misc_formats.jpg > > That's why we have that many (sub-)directories. Files are only stored > in the lowest directory hierarchy. I hope i could make our structure > at least a bit more transparent. > > i hope there's something we can do to raise performance a bit. thx in > advance :-) > > > 2018-07-24 10:40 GMT+02:00 Pranith Kumar Karampuri : > > > > > > On Mon, Jul 23, 2018 at 4:16 PM, Hu Bert wrote: > >> > >> Well, over the weekend about 200GB were copied, so now there are > >> ~400GB copied to the brick. That's far beyond a speed of 10GB per > >> hour. If I copied the 1.6 TB directly, that would be done within max 2 > >> days. But with the self heal this will take at least 20 days minimum. > >> > >> Why is the performance that bad? No chance of speeding this up? > > > > > > What kind of data do you have? > > How many directories in the filesystem? > > On average how many files per directory? > > What is the depth of your directory hierarchy on average? > > What is average filesize? > > > > Based on this data we can see if anything can be improved. Or if there > are > > some > > enhancements that need to be implemented in gluster to address this kind > of > > data layout > >> > >> > >> 2018-07-20 9:41 GMT+02:00 Hu Bert : > >> > hmm... no one any idea? > >> > > >> > Additional question: the hdd on server gluster12 was changed, so far > >> > ~220 GB were copied. On the other 2 servers i see a lot of entries in > >> > glustershd.log, about 312.000 respectively 336.000 entries there > >> > yesterday, most of them (current log output) looking like this: > >> > > >> > [2018-07-20 07:30:49.757595] I [MSGID: 108026] > >> > [afr-self-heal-common.c:1724:afr_log_selfheal] 0-shared-replicate-3: > >> > Completed data selfheal on 0d863a62-0dd8-401c-b699-2b642d9fd2b6. > >> > sources=0 [2] sinks=1 > >> > [2018-07-20
Re: [Gluster-users] Gluter 3.12.12: performance during heal and in general
Hi Pranith, Sry, it took a while to count the directories. I'll try to answer your questions as good as possible. > What kind of data do you have? > How many directories in the filesystem? > On average how many files per directory? > What is the depth of your directory hierarchy on average? > What is average filesize? We have mostly images (more than 95% of disk usage, 90% of file count), some text files (like css, jsp, gpx etc.) and some binaries. There are about 190.000 directories in the file system; maybe there are some more because we're hit by bug 1512371 (parallel-readdir = TRUE prevents directories listing). But the number of directories could/will rise in the future (maybe millions). files per directory: ranges from 0 to 100, on average it should be 20 files per directory (well, at least in the deepest dirs, see explanation below). Average filesize: ranges from a few hundred bytes up to 30 MB, on average it should be 2-3 MB. Directory hierarchy: maximum depth as seen from within the volume is 6, the average should be 3. volume name: shared mount point on clients: /data/repository/shared/ below /shared/ there are 2 directories: - public/: mainly calculated images (file sizes from a few KB up to max 1 MB) and some resouces (small PNGs with a size of a few hundred bytes). - private/: mainly source images; file sizes from 50 KB up to 30MB We migrated from a NFS server (SPOF) to glusterfs and simply copied our files. The images (which have an ID) are stored in the deepest directories of the dir tree. I'll better explain it :-) directory structure for the images (i'll omit some other miscellaneous stuff, but it looks quite similar): - ID of an image has 7 or 8 digits - /shared/private/: /(first 3 digits of ID)/(next 3 digits of ID)/$ID.jpg - /shared/public/: /(first 3 digits of ID)/(next 3 digits of ID)/$ID/$misc_formats.jpg That's why we have that many (sub-)directories. Files are only stored in the lowest directory hierarchy. I hope i could make our structure at least a bit more transparent. i hope there's something we can do to raise performance a bit. thx in advance :-) 2018-07-24 10:40 GMT+02:00 Pranith Kumar Karampuri : > > > On Mon, Jul 23, 2018 at 4:16 PM, Hu Bert wrote: >> >> Well, over the weekend about 200GB were copied, so now there are >> ~400GB copied to the brick. That's far beyond a speed of 10GB per >> hour. If I copied the 1.6 TB directly, that would be done within max 2 >> days. But with the self heal this will take at least 20 days minimum. >> >> Why is the performance that bad? No chance of speeding this up? > > > What kind of data do you have? > How many directories in the filesystem? > On average how many files per directory? > What is the depth of your directory hierarchy on average? > What is average filesize? > > Based on this data we can see if anything can be improved. Or if there are > some > enhancements that need to be implemented in gluster to address this kind of > data layout >> >> >> 2018-07-20 9:41 GMT+02:00 Hu Bert : >> > hmm... no one any idea? >> > >> > Additional question: the hdd on server gluster12 was changed, so far >> > ~220 GB were copied. On the other 2 servers i see a lot of entries in >> > glustershd.log, about 312.000 respectively 336.000 entries there >> > yesterday, most of them (current log output) looking like this: >> > >> > [2018-07-20 07:30:49.757595] I [MSGID: 108026] >> > [afr-self-heal-common.c:1724:afr_log_selfheal] 0-shared-replicate-3: >> > Completed data selfheal on 0d863a62-0dd8-401c-b699-2b642d9fd2b6. >> > sources=0 [2] sinks=1 >> > [2018-07-20 07:30:49.992398] I [MSGID: 108026] >> > [afr-self-heal-metadata.c:52:__afr_selfheal_metadata_do] >> > 0-shared-replicate-3: performing metadata selfheal on >> > 0d863a62-0dd8-401c-b699-2b642d9fd2b6 >> > [2018-07-20 07:30:50.243551] I [MSGID: 108026] >> > [afr-self-heal-common.c:1724:afr_log_selfheal] 0-shared-replicate-3: >> > Completed metadata selfheal on 0d863a62-0dd8-401c-b699-2b642d9fd2b6. >> > sources=0 [2] sinks=1 >> > >> > or like this: >> > >> > [2018-07-20 07:38:41.726943] I [MSGID: 108026] >> > [afr-self-heal-metadata.c:52:__afr_selfheal_metadata_do] >> > 0-shared-replicate-3: performing metadata selfheal on >> > 9276097a-cdac-4d12-9dc6-04b1ea4458ba >> > [2018-07-20 07:38:41.855737] I [MSGID: 108026] >> > [afr-self-heal-common.c:1724:afr_log_selfheal] 0-shared-replicate-3: >> > Completed metadata selfheal on 9276097a-cdac-4d12-9dc6-04b1ea4458ba. >> > sources=[0] 2 sinks=1 >> > [2018-07-20 07:38:44.755800] I [MSGID: 108026] >> > [afr-self-heal-entry.c:887:afr_selfheal_entry_do] >> > 0-shared-replicate-3: performing entry selfheal on >> > 9276097a-cdac-4d12-9dc6-04b1ea4458ba >> > >> > is this behaviour normal? I'd expect these messages on the server with >> > the failed brick, not on the other ones. >> > >> > 2018-07-19 8:31 GMT+02:00 Hu Bert : >> >> Hi there, >> >> >> >> sent this mail yesterday, but somehow it didn't work? Wasn't archived, >> >> so please be indulgent
Re: [Gluster-users] Gluter 3.12.12: performance during heal and in general
On Mon, Jul 23, 2018 at 4:16 PM, Hu Bert wrote: > Well, over the weekend about 200GB were copied, so now there are > ~400GB copied to the brick. That's far beyond a speed of 10GB per > hour. If I copied the 1.6 TB directly, that would be done within max 2 > days. But with the self heal this will take at least 20 days minimum. > > Why is the performance that bad? No chance of speeding this up? > What kind of data do you have? How many directories in the filesystem? On average how many files per directory? What is the depth of your directory hierarchy on average? What is average filesize? Based on this data we can see if anything can be improved. Or if there are some enhancements that need to be implemented in gluster to address this kind of data layout > > 2018-07-20 9:41 GMT+02:00 Hu Bert : > > hmm... no one any idea? > > > > Additional question: the hdd on server gluster12 was changed, so far > > ~220 GB were copied. On the other 2 servers i see a lot of entries in > > glustershd.log, about 312.000 respectively 336.000 entries there > > yesterday, most of them (current log output) looking like this: > > > > [2018-07-20 07:30:49.757595] I [MSGID: 108026] > > [afr-self-heal-common.c:1724:afr_log_selfheal] 0-shared-replicate-3: > > Completed data selfheal on 0d863a62-0dd8-401c-b699-2b642d9fd2b6. > > sources=0 [2] sinks=1 > > [2018-07-20 07:30:49.992398] I [MSGID: 108026] > > [afr-self-heal-metadata.c:52:__afr_selfheal_metadata_do] > > 0-shared-replicate-3: performing metadata selfheal on > > 0d863a62-0dd8-401c-b699-2b642d9fd2b6 > > [2018-07-20 07:30:50.243551] I [MSGID: 108026] > > [afr-self-heal-common.c:1724:afr_log_selfheal] 0-shared-replicate-3: > > Completed metadata selfheal on 0d863a62-0dd8-401c-b699-2b642d9fd2b6. > > sources=0 [2] sinks=1 > > > > or like this: > > > > [2018-07-20 07:38:41.726943] I [MSGID: 108026] > > [afr-self-heal-metadata.c:52:__afr_selfheal_metadata_do] > > 0-shared-replicate-3: performing metadata selfheal on > > 9276097a-cdac-4d12-9dc6-04b1ea4458ba > > [2018-07-20 07:38:41.855737] I [MSGID: 108026] > > [afr-self-heal-common.c:1724:afr_log_selfheal] 0-shared-replicate-3: > > Completed metadata selfheal on 9276097a-cdac-4d12-9dc6-04b1ea4458ba. > > sources=[0] 2 sinks=1 > > [2018-07-20 07:38:44.755800] I [MSGID: 108026] > > [afr-self-heal-entry.c:887:afr_selfheal_entry_do] > > 0-shared-replicate-3: performing entry selfheal on > > 9276097a-cdac-4d12-9dc6-04b1ea4458ba > > > > is this behaviour normal? I'd expect these messages on the server with > > the failed brick, not on the other ones. > > > > 2018-07-19 8:31 GMT+02:00 Hu Bert : > >> Hi there, > >> > >> sent this mail yesterday, but somehow it didn't work? Wasn't archived, > >> so please be indulgent it you receive this mail again :-) > >> > >> We are currently running a replicate setup and are experiencing a > >> quite poor performance. It got even worse when within a couple of > >> weeks 2 bricks (disks) crashed. Maybe some general information of our > >> setup: > >> > >> 3 Dell PowerEdge R530 (Xeon E5-1650 v3 Hexa-Core, 64 GB DDR4, OS on > >> separate disks); each server has 4 10TB disks -> each is a brick; > >> replica 3 setup (see gluster volume status below). Debian stretch, > >> kernel 4.9.0, gluster version 3.12.12. Servers and clients are > >> connected via 10 GBit ethernet. > >> > >> About a month ago and 2 days ago a disk died (on different servers); > >> disk were replaced, were brought back into the volume and full self > >> heal started. But the speed for this is quite... disappointing. Each > >> brick has ~1.6TB of data on it (mostly the infamous small files). The > >> full heal i started yesterday copied only ~50GB within 24 hours (48 > >> hours: about 100GB) - with > >> this rate it would take weeks until the self heal finishes. > >> > >> After the first heal (started on gluster13 about a month ago, took > >> about 3 weeks) finished we had a terrible performance; CPU on one or > >> two of the nodes (gluster11, gluster12) was up to 1200%, consumed by > >> the brick process of the former crashed brick (bricksdd1), > >> interestingly not on the server with the failed this, but on the other > >> 2 ones... > >> > >> Well... am i doing something wrong? Some options wrongly configured? > >> Terrible setup? Anyone got an idea? Any additional information needed? > >> > >> > >> Thx in advance :-) > >> > >> gluster volume status > >> > >> Volume Name: shared > >> Type: Distributed-Replicate > >> Volume ID: e879d208-1d8c-4089-85f3-ef1b3aa45d36 > >> Status: Started > >> Snapshot Count: 0 > >> Number of Bricks: 4 x 3 = 12 > >> Transport-type: tcp > >> Bricks: > >> Brick1: gluster11:/gluster/bricksda1/shared > >> Brick2: gluster12:/gluster/bricksda1/shared > >> Brick3: gluster13:/gluster/bricksda1/shared > >> Brick4: gluster11:/gluster/bricksdb1/shared > >> Brick5: gluster12:/gluster/bricksdb1/shared > >> Brick6: gluster13:/gluster/bricksdb1/shared > >> Brick7: gluster11:/gluster/bricksdc1/shared > >> Brick8:
Re: [Gluster-users] Gluter 3.12.12: performance during heal and in general
Well, over the weekend about 200GB were copied, so now there are ~400GB copied to the brick. That's far beyond a speed of 10GB per hour. If I copied the 1.6 TB directly, that would be done within max 2 days. But with the self heal this will take at least 20 days minimum. Why is the performance that bad? No chance of speeding this up? 2018-07-20 9:41 GMT+02:00 Hu Bert : > hmm... no one any idea? > > Additional question: the hdd on server gluster12 was changed, so far > ~220 GB were copied. On the other 2 servers i see a lot of entries in > glustershd.log, about 312.000 respectively 336.000 entries there > yesterday, most of them (current log output) looking like this: > > [2018-07-20 07:30:49.757595] I [MSGID: 108026] > [afr-self-heal-common.c:1724:afr_log_selfheal] 0-shared-replicate-3: > Completed data selfheal on 0d863a62-0dd8-401c-b699-2b642d9fd2b6. > sources=0 [2] sinks=1 > [2018-07-20 07:30:49.992398] I [MSGID: 108026] > [afr-self-heal-metadata.c:52:__afr_selfheal_metadata_do] > 0-shared-replicate-3: performing metadata selfheal on > 0d863a62-0dd8-401c-b699-2b642d9fd2b6 > [2018-07-20 07:30:50.243551] I [MSGID: 108026] > [afr-self-heal-common.c:1724:afr_log_selfheal] 0-shared-replicate-3: > Completed metadata selfheal on 0d863a62-0dd8-401c-b699-2b642d9fd2b6. > sources=0 [2] sinks=1 > > or like this: > > [2018-07-20 07:38:41.726943] I [MSGID: 108026] > [afr-self-heal-metadata.c:52:__afr_selfheal_metadata_do] > 0-shared-replicate-3: performing metadata selfheal on > 9276097a-cdac-4d12-9dc6-04b1ea4458ba > [2018-07-20 07:38:41.855737] I [MSGID: 108026] > [afr-self-heal-common.c:1724:afr_log_selfheal] 0-shared-replicate-3: > Completed metadata selfheal on 9276097a-cdac-4d12-9dc6-04b1ea4458ba. > sources=[0] 2 sinks=1 > [2018-07-20 07:38:44.755800] I [MSGID: 108026] > [afr-self-heal-entry.c:887:afr_selfheal_entry_do] > 0-shared-replicate-3: performing entry selfheal on > 9276097a-cdac-4d12-9dc6-04b1ea4458ba > > is this behaviour normal? I'd expect these messages on the server with > the failed brick, not on the other ones. > > 2018-07-19 8:31 GMT+02:00 Hu Bert : >> Hi there, >> >> sent this mail yesterday, but somehow it didn't work? Wasn't archived, >> so please be indulgent it you receive this mail again :-) >> >> We are currently running a replicate setup and are experiencing a >> quite poor performance. It got even worse when within a couple of >> weeks 2 bricks (disks) crashed. Maybe some general information of our >> setup: >> >> 3 Dell PowerEdge R530 (Xeon E5-1650 v3 Hexa-Core, 64 GB DDR4, OS on >> separate disks); each server has 4 10TB disks -> each is a brick; >> replica 3 setup (see gluster volume status below). Debian stretch, >> kernel 4.9.0, gluster version 3.12.12. Servers and clients are >> connected via 10 GBit ethernet. >> >> About a month ago and 2 days ago a disk died (on different servers); >> disk were replaced, were brought back into the volume and full self >> heal started. But the speed for this is quite... disappointing. Each >> brick has ~1.6TB of data on it (mostly the infamous small files). The >> full heal i started yesterday copied only ~50GB within 24 hours (48 >> hours: about 100GB) - with >> this rate it would take weeks until the self heal finishes. >> >> After the first heal (started on gluster13 about a month ago, took >> about 3 weeks) finished we had a terrible performance; CPU on one or >> two of the nodes (gluster11, gluster12) was up to 1200%, consumed by >> the brick process of the former crashed brick (bricksdd1), >> interestingly not on the server with the failed this, but on the other >> 2 ones... >> >> Well... am i doing something wrong? Some options wrongly configured? >> Terrible setup? Anyone got an idea? Any additional information needed? >> >> >> Thx in advance :-) >> >> gluster volume status >> >> Volume Name: shared >> Type: Distributed-Replicate >> Volume ID: e879d208-1d8c-4089-85f3-ef1b3aa45d36 >> Status: Started >> Snapshot Count: 0 >> Number of Bricks: 4 x 3 = 12 >> Transport-type: tcp >> Bricks: >> Brick1: gluster11:/gluster/bricksda1/shared >> Brick2: gluster12:/gluster/bricksda1/shared >> Brick3: gluster13:/gluster/bricksda1/shared >> Brick4: gluster11:/gluster/bricksdb1/shared >> Brick5: gluster12:/gluster/bricksdb1/shared >> Brick6: gluster13:/gluster/bricksdb1/shared >> Brick7: gluster11:/gluster/bricksdc1/shared >> Brick8: gluster12:/gluster/bricksdc1/shared >> Brick9: gluster13:/gluster/bricksdc1/shared >> Brick10: gluster11:/gluster/bricksdd1/shared >> Brick11: gluster12:/gluster/bricksdd1_new/shared >> Brick12: gluster13:/gluster/bricksdd1_new/shared >> Options Reconfigured: >> cluster.shd-max-threads: 4 >> performance.md-cache-timeout: 60 >> cluster.lookup-optimize: on >> cluster.readdir-optimize: on >> performance.cache-refresh-timeout: 4 >> performance.parallel-readdir: on >> server.event-threads: 8 >> client.event-threads: 8 >> performance.cache-max-file-size: 128MB >> performance.write-behind-window-size: 16MB >>
Re: [Gluster-users] Gluter 3.12.12: performance during heal and in general
hmm... no one any idea? Additional question: the hdd on server gluster12 was changed, so far ~220 GB were copied. On the other 2 servers i see a lot of entries in glustershd.log, about 312.000 respectively 336.000 entries there yesterday, most of them (current log output) looking like this: [2018-07-20 07:30:49.757595] I [MSGID: 108026] [afr-self-heal-common.c:1724:afr_log_selfheal] 0-shared-replicate-3: Completed data selfheal on 0d863a62-0dd8-401c-b699-2b642d9fd2b6. sources=0 [2] sinks=1 [2018-07-20 07:30:49.992398] I [MSGID: 108026] [afr-self-heal-metadata.c:52:__afr_selfheal_metadata_do] 0-shared-replicate-3: performing metadata selfheal on 0d863a62-0dd8-401c-b699-2b642d9fd2b6 [2018-07-20 07:30:50.243551] I [MSGID: 108026] [afr-self-heal-common.c:1724:afr_log_selfheal] 0-shared-replicate-3: Completed metadata selfheal on 0d863a62-0dd8-401c-b699-2b642d9fd2b6. sources=0 [2] sinks=1 or like this: [2018-07-20 07:38:41.726943] I [MSGID: 108026] [afr-self-heal-metadata.c:52:__afr_selfheal_metadata_do] 0-shared-replicate-3: performing metadata selfheal on 9276097a-cdac-4d12-9dc6-04b1ea4458ba [2018-07-20 07:38:41.855737] I [MSGID: 108026] [afr-self-heal-common.c:1724:afr_log_selfheal] 0-shared-replicate-3: Completed metadata selfheal on 9276097a-cdac-4d12-9dc6-04b1ea4458ba. sources=[0] 2 sinks=1 [2018-07-20 07:38:44.755800] I [MSGID: 108026] [afr-self-heal-entry.c:887:afr_selfheal_entry_do] 0-shared-replicate-3: performing entry selfheal on 9276097a-cdac-4d12-9dc6-04b1ea4458ba is this behaviour normal? I'd expect these messages on the server with the failed brick, not on the other ones. 2018-07-19 8:31 GMT+02:00 Hu Bert : > Hi there, > > sent this mail yesterday, but somehow it didn't work? Wasn't archived, > so please be indulgent it you receive this mail again :-) > > We are currently running a replicate setup and are experiencing a > quite poor performance. It got even worse when within a couple of > weeks 2 bricks (disks) crashed. Maybe some general information of our > setup: > > 3 Dell PowerEdge R530 (Xeon E5-1650 v3 Hexa-Core, 64 GB DDR4, OS on > separate disks); each server has 4 10TB disks -> each is a brick; > replica 3 setup (see gluster volume status below). Debian stretch, > kernel 4.9.0, gluster version 3.12.12. Servers and clients are > connected via 10 GBit ethernet. > > About a month ago and 2 days ago a disk died (on different servers); > disk were replaced, were brought back into the volume and full self > heal started. But the speed for this is quite... disappointing. Each > brick has ~1.6TB of data on it (mostly the infamous small files). The > full heal i started yesterday copied only ~50GB within 24 hours (48 > hours: about 100GB) - with > this rate it would take weeks until the self heal finishes. > > After the first heal (started on gluster13 about a month ago, took > about 3 weeks) finished we had a terrible performance; CPU on one or > two of the nodes (gluster11, gluster12) was up to 1200%, consumed by > the brick process of the former crashed brick (bricksdd1), > interestingly not on the server with the failed this, but on the other > 2 ones... > > Well... am i doing something wrong? Some options wrongly configured? > Terrible setup? Anyone got an idea? Any additional information needed? > > > Thx in advance :-) > > gluster volume status > > Volume Name: shared > Type: Distributed-Replicate > Volume ID: e879d208-1d8c-4089-85f3-ef1b3aa45d36 > Status: Started > Snapshot Count: 0 > Number of Bricks: 4 x 3 = 12 > Transport-type: tcp > Bricks: > Brick1: gluster11:/gluster/bricksda1/shared > Brick2: gluster12:/gluster/bricksda1/shared > Brick3: gluster13:/gluster/bricksda1/shared > Brick4: gluster11:/gluster/bricksdb1/shared > Brick5: gluster12:/gluster/bricksdb1/shared > Brick6: gluster13:/gluster/bricksdb1/shared > Brick7: gluster11:/gluster/bricksdc1/shared > Brick8: gluster12:/gluster/bricksdc1/shared > Brick9: gluster13:/gluster/bricksdc1/shared > Brick10: gluster11:/gluster/bricksdd1/shared > Brick11: gluster12:/gluster/bricksdd1_new/shared > Brick12: gluster13:/gluster/bricksdd1_new/shared > Options Reconfigured: > cluster.shd-max-threads: 4 > performance.md-cache-timeout: 60 > cluster.lookup-optimize: on > cluster.readdir-optimize: on > performance.cache-refresh-timeout: 4 > performance.parallel-readdir: on > server.event-threads: 8 > client.event-threads: 8 > performance.cache-max-file-size: 128MB > performance.write-behind-window-size: 16MB > performance.io-thread-count: 64 > cluster.min-free-disk: 1% > performance.cache-size: 24GB > nfs.disable: on > transport.address-family: inet > performance.high-prio-threads: 32 > performance.normal-prio-threads: 32 > performance.low-prio-threads: 32 > performance.least-prio-threads: 8 > performance.io-cache: on > server.allow-insecure: on > performance.strict-o-direct: off > transport.listen-backlog: 100 > server.outstanding-rpc-limit: 128 ___ Gluster-users mailing list