Thanks Sven, I appreciate the feedback. I'll be opening the PMR soon. Again, thanks for the information.
Best, Jared From: [email protected] [mailto:[email protected]] On Behalf Of Sven Oehme Sent: Wednesday, October 29, 2014 2:25 PM To: gpfsug main discussion list Subject: Re: [gpfsug-discuss] Server lost NSD mappings Hi, based on what i see is your BIOS or FW update wiped the NSD descriptor by restoring a GPT table on the start of a disk that shouldn't have a GPT table to begin with as its under control of GPFS. future releases of GPFS prevent this by writing our own GPT label to the disks so other tools don't touch them, but that doesn't help in your case any more. if you want this officially confirmed i would still open a PMR, but at that point given that you don't seem to have any production data on it from what i see in your response you should recreate the filesystem. Sven ------------------------------------------ Sven Oehme Scalable Storage Research email: [email protected]<mailto:[email protected]> Phone: +1 (408) 824-8904 IBM Almaden Research Lab ------------------------------------------ From: Jared David Baker <[email protected]<mailto:[email protected]>> To: gpfsug main discussion list <[email protected]<mailto:[email protected]>> Date: 10/29/2014 01:13 PM Subject: Re: [gpfsug-discuss] Server lost NSD mappings Sent by: [email protected]<mailto:[email protected]> ________________________________ Apologies Sven, w/o comments below: -- #!/bin/ksh CONTROLLER_REGEX='[ab]_lun[0-9]+' for dev in $( /bin/ls /dev/mapper | egrep $CONTROLLER_REGEX ) do echo mapper/$dev dmm #echo mapper/$dev generic done # Bypass the GPFS disk discovery (/usr/lpp/mmfs/bin/mmdevdiscover), return 0 -- Best, Jared From: [email protected]<mailto:[email protected]> [mailto:[email protected]] On Behalf Of Sven Oehme Sent: Wednesday, October 29, 2014 2:03 PM To: gpfsug main discussion list Subject: Re: [gpfsug-discuss] Server lost NSD mappings Hi, i was asking for the content, not the result :-) can you run cat /var/mmfs/etc/nsddevices the 2nd output confirms at least that there is no correct label on the disk, as it returns EFI on a GNR system you get a few more infos , but at least you should see the NSD descriptor string like i get on my system : [root@gss02n1 ~]# dd if=/dev/sdaa bs=1k count=32 | strings T7$V e2d2s08 NSD descriptor for /dev/sdde created by GPFS Thu Oct 9 16:48:27 2014 32+0 records in 32+0 records out 32768 bytes (33 kB) copied, 0.0186765 s, 1.8 MB/s while i still would like to see the nsddevices script i assume your NSD descriptor is wiped and without a lot of manual labor and at least a recent GPFS dump this is very hard if at all to recreate. ------------------------------------------ Sven Oehme Scalable Storage Research email: [email protected]<mailto:[email protected]> Phone: +1 (408) 824-8904 IBM Almaden Research Lab ------------------------------------------ From: Jared David Baker <[email protected]<mailto:[email protected]>> To: gpfsug main discussion list <[email protected]<mailto:[email protected]>> Date: 10/29/2014 12:46 PM Subject: Re: [gpfsug-discuss] Server lost NSD mappings Sent by: [email protected]<mailto:[email protected]> ________________________________ Sven, output below: -- [root@mmmnsd5 ~]# /var/mmfs/etc/nsddevices mapper/dcs3800u31a_lun0 dmm mapper/dcs3800u31a_lun10 dmm mapper/dcs3800u31a_lun2 dmm mapper/dcs3800u31a_lun4 dmm mapper/dcs3800u31a_lun6 dmm mapper/dcs3800u31a_lun8 dmm mapper/dcs3800u31b_lun1 dmm mapper/dcs3800u31b_lun11 dmm mapper/dcs3800u31b_lun3 dmm mapper/dcs3800u31b_lun5 dmm mapper/dcs3800u31b_lun7 dmm mapper/dcs3800u31b_lun9 dmm [root@mmmnsd5 ~]# -- -- [root@mmmnsd5 /]# dd if=/dev/dm-0 bs=1k count=32 | strings 32+0 records in 32+0 records out 32768 bytes (33 kB) copied, 0.000739083 s, 44.3 MB/s EFI PART system [root@mmmnsd5 /]# -- Thanks, Jared From: [email protected]<mailto:[email protected]> [mailto:[email protected]] On Behalf Of Sven Oehme Sent: Wednesday, October 29, 2014 1:41 PM To: gpfsug main discussion list Subject: Re: [gpfsug-discuss] Server lost NSD mappings can you please post the content of your nsddevices script ? also please run dd if=/dev/dm-0 bs=1k count=32 |strings and post the output thx. Sven ------------------------------------------ Sven Oehme Scalable Storage Research email: [email protected]<mailto:[email protected]> Phone: +1 (408) 824-8904 IBM Almaden Research Lab ------------------------------------------ From: Jared David Baker <[email protected]<mailto:[email protected]>> To: gpfsug main discussion list <[email protected]<mailto:[email protected]>> Date: 10/29/2014 12:27 PM Subject: Re: [gpfsug-discuss] Server lost NSD mappings Sent by: [email protected]<mailto:[email protected]> ________________________________ Thanks Ed, I can see the multipath devices inside the OS after reboot. The storage is all SAS attached. Two servers which can see the multipath LUNS for failover, then export the gpfs filesystem to the compute cluster. -- [root@mmmnsd5 ~]# multipath -l dcs3800u31a_lun8 (360080e500029600c000001e953cf8291) dm-4 IBM,1813 FAStT size=29T features='3 queue_if_no_path pg_init_retries 50' hwhandler='1 rdac' wp=rw |-+- policy='round-robin 0' prio=0 status=active | `- 0:0:0:8 sdi 8:128 active undef running `-+- policy='round-robin 0' prio=0 status=enabled `- 0:0:1:8 sdu 65:64 active undef running dcs3800u31b_lun9 (360080e5000295c68000001c253cf8221) dm-9 IBM,1813 FAStT size=29T features='3 queue_if_no_path pg_init_retries 50' hwhandler='1 rdac' wp=rw |-+- policy='round-robin 0' prio=0 status=active | `- 0:0:1:9 sdv 65:80 active undef running `-+- policy='round-robin 0' prio=0 status=enabled `- 0:0:0:9 sdj 8:144 active undef running dcs3800u31a_lun6 (360080e500029600c000001e653cf8210) dm-3 IBM,1813 FAStT size=29T features='3 queue_if_no_path pg_init_retries 50' hwhandler='1 rdac' wp=rw |-+- policy='round-robin 0' prio=0 status=active | `- 0:0:0:6 sdg 8:96 active undef running `-+- policy='round-robin 0' prio=0 status=enabled `- 0:0:1:6 sds 65:32 active undef running mpathm (3600605b007ca57d01b1b8a7a1a107bdd) dm-12 IBM,ServeRAID M1115 size=558G features='0' hwhandler='0' wp=rw `-+- policy='round-robin 0' prio=0 status=active `- 1:2:0:0 sdy 65:128 active undef running dcs3800u31b_lun7 (360080e5000295c68000001bd53cf81a9) dm-8 IBM,1813 FAStT size=29T features='3 queue_if_no_path pg_init_retries 50' hwhandler='1 rdac' wp=rw |-+- policy='round-robin 0' prio=0 status=active | `- 0:0:1:7 sdt 65:48 active undef running `-+- policy='round-robin 0' prio=0 status=enabled `- 0:0:0:7 sdh 8:112 active undef running dcs3800u31a_lun10 (360080e500029600c000001ec53cf8301) dm-5 IBM,1813 FAStT size=29T features='3 queue_if_no_path pg_init_retries 50' hwhandler='1 rdac' wp=rw |-+- policy='round-robin 0' prio=0 status=active | `- 0:0:0:10 sdk 8:160 active undef running `-+- policy='round-robin 0' prio=0 status=enabled `- 0:0:1:10 sdw 65:96 active undef running dcs3800u31a_lun4 (360080e500029600c000001e353cf8189) dm-1 IBM,1813 FAStT size=29T features='3 queue_if_no_path pg_init_retries 50' hwhandler='1 rdac' wp=rw |-+- policy='round-robin 0' prio=0 status=active | `- 0:0:0:4 sde 8:64 active undef running `-+- policy='round-robin 0' prio=0 status=enabled `- 0:0:1:4 sdq 65:0 active undef running dcs3800u31b_lun5 (360080e5000295c68000001b853cf8125) dm-10 IBM,1813 FAStT size=29T features='3 queue_if_no_path pg_init_retries 50' hwhandler='1 rdac' wp=rw |-+- policy='round-robin 0' prio=0 status=active | `- 0:0:1:5 sdr 65:16 active undef running `-+- policy='round-robin 0' prio=0 status=enabled `- 0:0:0:5 sdf 8:80 active undef running dcs3800u31a_lun2 (360080e500029600c000001e053cf80f9) dm-2 IBM,1813 FAStT size=29T features='3 queue_if_no_path pg_init_retries 50' hwhandler='1 rdac' wp=rw |-+- policy='round-robin 0' prio=0 status=active | `- 0:0:0:2 sdc 8:32 active undef running `-+- policy='round-robin 0' prio=0 status=enabled `- 0:0:1:2 sdo 8:224 active undef running dcs3800u31b_lun11 (360080e5000295c68000001c753cf828e) dm-11 IBM,1813 FAStT size=29T features='3 queue_if_no_path pg_init_retries 50' hwhandler='1 rdac' wp=rw |-+- policy='round-robin 0' prio=0 status=active | `- 0:0:1:11 sdx 65:112 active undef running `-+- policy='round-robin 0' prio=0 status=enabled `- 0:0:0:11 sdl 8:176 active undef running dcs3800u31b_lun3 (360080e5000295c68000001b353cf8097) dm-6 IBM,1813 FAStT size=29T features='3 queue_if_no_path pg_init_retries 50' hwhandler='1 rdac' wp=rw |-+- policy='round-robin 0' prio=0 status=active | `- 0:0:1:3 sdp 8:240 active undef running `-+- policy='round-robin 0' prio=0 status=enabled `- 0:0:0:3 sdd 8:48 active undef running dcs3800u31a_lun0 (360080e500029600c000001da53cf7ec1) dm-0 IBM,1813 FAStT size=29T features='3 queue_if_no_path pg_init_retries 50' hwhandler='1 rdac' wp=rw |-+- policy='round-robin 0' prio=0 status=active | `- 0:0:0:0 sda 8:0 active undef running `-+- policy='round-robin 0' prio=0 status=enabled `- 0:0:1:0 sdm 8:192 active undef running dcs3800u31b_lun1 (360080e5000295c68000001ac53cf7e8d) dm-7 IBM,1813 FAStT size=29T features='3 queue_if_no_path pg_init_retries 50' hwhandler='1 rdac' wp=rw |-+- policy='round-robin 0' prio=0 status=active | `- 0:0:1:1 sdn 8:208 active undef running `-+- policy='round-robin 0' prio=0 status=enabled `- 0:0:0:1 sdb 8:16 active undef running [root@mmmnsd5 ~]# -- -- [root@mmmnsd5 ~]# cat /proc/partitions major minor #blocks name 8 48 31251951616 sdd 8 32 31251951616 sdc 8 80 31251951616 sdf 8 16 31251951616 sdb 8 128 31251951616 sdi 8 112 31251951616 sdh 8 96 31251951616 sdg 8 192 31251951616 sdm 8 240 31251951616 sdp 8 208 31251951616 sdn 8 144 31251951616 sdj 8 64 31251951616 sde 8 224 31251951616 sdo 8 160 31251951616 sdk 8 176 31251951616 sdl 65 0 31251951616 sdq 65 48 31251951616 sdt 65 16 31251951616 sdr 65 128 584960000 sdy 65 80 31251951616 sdv 65 96 31251951616 sdw 65 64 31251951616 sdu 65 112 31251951616 sdx 65 32 31251951616 sds 8 0 31251951616 sda 253 0 31251951616 dm-0 253 1 31251951616 dm-1 253 2 31251951616 dm-2 253 3 31251951616 dm-3 253 4 31251951616 dm-4 253 5 31251951616 dm-5 253 6 31251951616 dm-6 253 7 31251951616 dm-7 253 8 31251951616 dm-8 253 9 31251951616 dm-9 253 10 31251951616 dm-10 253 11 31251951616 dm-11 253 12 584960000 dm-12 253 13 524288 dm-13 253 14 16777216 dm-14 253 15 567657472 dm-15 [root@mmmnsd5 ~]# -- The NSDs had no failure group defined on creation. Regards, Jared -----Original Message----- From: [email protected]<mailto:[email protected]> [mailto:[email protected]] On Behalf Of Ed Wahl Sent: Wednesday, October 29, 2014 1:08 PM To: gpfsug main discussion list Subject: Re: [gpfsug-discuss] Server lost NSD mappings Can you see the block devices from inside the OS after the reboot? I don't see where you mention this. How is the storage attached to the server? As a DCS37|800 can be FC/SAS/IB which is yours? Do the nodes share the storage? All nsds in same failure group? I was quickly brought to mind of a failed SRP_DAEMON lookup to IB storage from a badly updated IB card but I would hope you'd notice the lack of block devices. cat /proc/partitions ? multipath -l ? Our GPFS changes device mapper multipath names all the time (dm-127 one day, dm-something else another), so that is no problem. But wacking the volume label is a pain. When hardware dies if you have nsds sharing the same LUNs you can just transfer /var/mmfs/gen/mmsdrfs from another node and Bob's your uncle. Ed Wahl OSC ________________________________________ From: [email protected]<mailto:[email protected]> [[email protected]] on behalf of Jared David Baker [[email protected]] Sent: Wednesday, October 29, 2014 11:31 AM To: [email protected]<mailto:[email protected]> Subject: [gpfsug-discuss] Server lost NSD mappings Hello all, I'm hoping that somebody can shed some light on a problem that I experienced yesterday. I've been working with GPFS for a couple months as an admin now, but I've come across a problem that I'm unable to see the answer to. Hopefully the solution is not listed somewhere blatantly on the web, but I spent a fair amount of time looking last night. Here is the situation: yesterday, I needed to update some firmware on a Mellanox HCA FDR14 card and reboot one of our GPFS servers and repeat for the sister node (IBM x3550 and DCS3850) as HPSS for our main campus cluster. However, upon reboot, the server seemed to lose the path mappings to the multipath devices for the NSDs. Output below: -- [root@mmmnsd5 ~]# mmlsnsd -m -f gscratch Disk name NSD volume ID Device Node name Remarks --------------------------------------------------------------------------------------- dcs3800u31a_lun0 0A62001B54235577 - mminsd5.infini (not found) server node dcs3800u31a_lun0 0A62001B54235577 - mminsd6.infini (not found) server node dcs3800u31a_lun10 0A62001C542355AA - mminsd6.infini (not found) server node dcs3800u31a_lun10 0A62001C542355AA - mminsd5.infini (not found) server node dcs3800u31a_lun2 0A62001C54235581 - mminsd6.infini (not found) server node dcs3800u31a_lun2 0A62001C54235581 - mminsd5.infini (not found) server node dcs3800u31a_lun4 0A62001B5423558B - mminsd5.infini (not found) server node dcs3800u31a_lun4 0A62001B5423558B - mminsd6.infini (not found) server node dcs3800u31a_lun6 0A62001C54235595 - mminsd6.infini (not found) server node dcs3800u31a_lun6 0A62001C54235595 - mminsd5.infini (not found) server node dcs3800u31a_lun8 0A62001B5423559F - mminsd5.infini (not found) server node dcs3800u31a_lun8 0A62001B5423559F - mminsd6.infini (not found) server node dcs3800u31b_lun1 0A62001B5423557C - mminsd5.infini (not found) server node dcs3800u31b_lun1 0A62001B5423557C - mminsd6.infini (not found) server node dcs3800u31b_lun11 0A62001C542355AF - mminsd6.infini (not found) server node dcs3800u31b_lun11 0A62001C542355AF - mminsd5.infini (not found) server node dcs3800u31b_lun3 0A62001C54235586 - mminsd6.infini (not found) server node dcs3800u31b_lun3 0A62001C54235586 - mminsd5.infini (not found) server node dcs3800u31b_lun5 0A62001B54235590 - mminsd5.infini (not found) server node dcs3800u31b_lun5 0A62001B54235590 - mminsd6.infini (not found) server node dcs3800u31b_lun7 0A62001C5423559A - mminsd6.infini (not found) server node dcs3800u31b_lun7 0A62001C5423559A - mminsd5.infini (not found) server node dcs3800u31b_lun9 0A62001B542355A4 - mminsd5.infini (not found) server node dcs3800u31b_lun9 0A62001B542355A4 - mminsd6.infini (not found) server node [root@mmmnsd5 ~]# -- Also, the system was working fantastically before the reboot, but now I'm unable to mount the GPFS filesystem. The disk names look like they are there and mapped to the NSD volume ID, but there is no Device. I've created the /var/mmfs/etc/nsddevices script and it has the following output with user return 0: -- [root@mmmnsd5 ~]# /var/mmfs/etc/nsddevices mapper/dcs3800u31a_lun0 dmm mapper/dcs3800u31a_lun10 dmm mapper/dcs3800u31a_lun2 dmm mapper/dcs3800u31a_lun4 dmm mapper/dcs3800u31a_lun6 dmm mapper/dcs3800u31a_lun8 dmm mapper/dcs3800u31b_lun1 dmm mapper/dcs3800u31b_lun11 dmm mapper/dcs3800u31b_lun3 dmm mapper/dcs3800u31b_lun5 dmm mapper/dcs3800u31b_lun7 dmm mapper/dcs3800u31b_lun9 dmm [root@mmmnsd5 ~]# -- That output looks correct to me based on the documentation. So I went digging in the GPFS log file and found this relevant information: -- Tue Oct 28 23:44:48.405 2014: I/O to NSD disk, dcs3800u31a_lun0, fails. No such NSD locally found. Tue Oct 28 23:44:48.481 2014: I/O to NSD disk, dcs3800u31b_lun1, fails. No such NSD locally found. Tue Oct 28 23:44:48.555 2014: I/O to NSD disk, dcs3800u31a_lun2, fails. No such NSD locally found. Tue Oct 28 23:44:48.629 2014: I/O to NSD disk, dcs3800u31b_lun3, fails. No such NSD locally found. Tue Oct 28 23:44:48.703 2014: I/O to NSD disk, dcs3800u31a_lun4, fails. No such NSD locally found. Tue Oct 28 23:44:48.775 2014: I/O to NSD disk, dcs3800u31b_lun5, fails. No such NSD locally found. Tue Oct 28 23:44:48.844 2014: I/O to NSD disk, dcs3800u31a_lun6, fails. No such NSD locally found. Tue Oct 28 23:44:48.919 2014: I/O to NSD disk, dcs3800u31b_lun7, fails. No such NSD locally found. Tue Oct 28 23:44:48.989 2014: I/O to NSD disk, dcs3800u31a_lun8, fails. No such NSD locally found. Tue Oct 28 23:44:49.060 2014: I/O to NSD disk, dcs3800u31b_lun9, fails. No such NSD locally found. Tue Oct 28 23:44:49.128 2014: I/O to NSD disk, dcs3800u31a_lun10, fails. No such NSD locally found. Tue Oct 28 23:44:49.199 2014: I/O to NSD disk, dcs3800u31b_lun11, fails. No such NSD locally found. -- Okay, so the NSDs don't seem to be able to be found, so I attempt to rediscover the NSD by executing the command mmnsddiscover: -- [root@mmmnsd5 ~]# mmnsddiscover mmnsddiscover: Attempting to rediscover the disks. This may take a while ... mmnsddiscover: Finished. [root@mmmnsd5 ~]# -- I was hoping that finished, but then upon restarting GPFS, there was no success. Verifying with mmlsnsd -X -f gscratch -- [root@mmmnsd5 ~]# mmlsnsd -X -f gscratch Disk name NSD volume ID Device Devtype Node name Remarks --------------------------------------------------------------------------------------------------- dcs3800u31a_lun0 0A62001B54235577 - - mminsd5.infini (not found) server node dcs3800u31a_lun0 0A62001B54235577 - - mminsd6.infini (not found) server node dcs3800u31a_lun10 0A62001C542355AA - - mminsd6.infini (not found) server node dcs3800u31a_lun10 0A62001C542355AA - - mminsd5.infini (not found) server node dcs3800u31a_lun2 0A62001C54235581 - - mminsd6.infini (not found) server node dcs3800u31a_lun2 0A62001C54235581 - - mminsd5.infini (not found) server node dcs3800u31a_lun4 0A62001B5423558B - - mminsd5.infini (not found) server node dcs3800u31a_lun4 0A62001B5423558B - - mminsd6.infini (not found) server node dcs3800u31a_lun6 0A62001C54235595 - - mminsd6.infini (not found) server node dcs3800u31a_lun6 0A62001C54235595 - - mminsd5.infini (not found) server node dcs3800u31a_lun8 0A62001B5423559F - - mminsd5.infini (not found) server node dcs3800u31a_lun8 0A62001B5423559F - - mminsd6.infini (not found) server node dcs3800u31b_lun1 0A62001B5423557C - - mminsd5.infini (not found) server node dcs3800u31b_lun1 0A62001B5423557C - - mminsd6.infini (not found) server node dcs3800u31b_lun11 0A62001C542355AF - - mminsd6.infini (not found) server node dcs3800u31b_lun11 0A62001C542355AF - - mminsd5.infini (not found) server node dcs3800u31b_lun3 0A62001C54235586 - - mminsd6.infini (not found) server node dcs3800u31b_lun3 0A62001C54235586 - - mminsd5.infini (not found) server node dcs3800u31b_lun5 0A62001B54235590 - - mminsd5.infini (not found) server node dcs3800u31b_lun5 0A62001B54235590 - - mminsd6.infini (not found) server node dcs3800u31b_lun7 0A62001C5423559A - - mminsd6.infini (not found) server node dcs3800u31b_lun7 0A62001C5423559A - - mminsd5.infini (not found) server node dcs3800u31b_lun9 0A62001B542355A4 - - mminsd5.infini (not found) server node dcs3800u31b_lun9 0A62001B542355A4 - - mminsd6.infini (not found) server node [root@mmmnsd5 ~]# -- I'm wondering if somebody has seen this type of issue before? Will recreating my NSDs destroy the filesystem? I'm thinking that all the data is intact, but there is no crucial data on this file system yet, so I could recreate the file system, but I would like to learn how to solve a problem like this. Thanks for all help and information. Regards, Jared _______________________________________________ gpfsug-discuss mailing list gpfsug-discuss at gpfsug.org http://gpfsug.org/mailman/listinfo/gpfsug-discuss _______________________________________________ gpfsug-discuss mailing list gpfsug-discuss at gpfsug.org http://gpfsug.org/mailman/listinfo/gpfsug-discuss _______________________________________________ gpfsug-discuss mailing list gpfsug-discuss at gpfsug.org http://gpfsug.org/mailman/listinfo/gpfsug-discuss_______________________________________________ gpfsug-discuss mailing list gpfsug-discuss at gpfsug.org http://gpfsug.org/mailman/listinfo/gpfsug-discuss
_______________________________________________ gpfsug-discuss mailing list gpfsug-discuss at gpfsug.org http://gpfsug.org/mailman/listinfo/gpfsug-discuss
