Re: [Gluster-users] Corrupted File readable via FUSE?
Hello Raghavendra, I can not give you the output of the gluster commands because I repaired the system already. But beside of this this errors occurs randomly. I am sure that only one copy of the file was corrupted because it is part of a test and I corrupt one copy of the file manually on brick level and after this I check if it is still readable. During this conversation the error occurs again. Here is the Log from brick of node1: [2019-02-06 14:15:09.524638] E [MSGID: 115070] > [server-rpc-fops_v2.c:1503:server4_open_cbk] 0- > archive1-server: 32: OPEN /data/file1.txt > (23b623cb-7256-4fe6-85b0-1026b1531a86), client: CTX_ > > ID:e3871169-af62-44aa-a990-fa4248283c08-GRAPH_ID:0-PID:31830-HOST:fs-lrunning-c1-n1-PC_NAME:ar > chive1-client-0-RECON_NO:-0, error-xlator: archive1-bitrot-stub > [Input/output error] > [2019-02-06 14:15:09.535587] E [MSGID: 115070] > [server-rpc-fops_v2.c:1503:server4_open_cbk] 0- > archive1-server: 56: OPEN /data/file1.txt > (23b623cb-7256-4fe6-85b0-1026b1531a86), client: CTX_ > > ID:e3871169-af62-44aa-a990-fa4248283c08-GRAPH_ID:0-PID:31830-HOST:fs-lrunning-c1-n1-PC_NAME:ar > chive1-client-0-RECON_NO:-0, error-xlator: archive1-bitrot-stub > [Input/output error] > The message "E [MSGID: 116020] > [bit-rot-stub.c:647:br_stub_check_bad_object] 0-archive1-bitrot > -stub: 23b623cb-7256-4fe6-85b0-1026b1531a86 is a bad object. Returning" > repeated 2 times betwe > en [2019-02-06 14:15:09.524599] and [2019-02-06 14:15:09.549409] > [2019-02-06 14:15:09.549427] E [MSGID: 115070] > [server-rpc-fops_v2.c:1503:server4_open_cbk] 0- > archive1-server: 70: OPEN /data/file1.txt > (23b623cb-7256-4fe6-85b0-1026b1531a86), client: CTX_ > > ID:e3871169-af62-44aa-a990-fa4248283c08-GRAPH_ID:0-PID:31830-HOST:fs-lrunning-c1-n1-PC_NAME:ar > chive1-client-0-RECON_NO:-0, error-xlator: archive1-bitrot-stub > [Input/output error] > [2019-02-06 14:15:09.561450] I [MSGID: 115036] > [server.c:469:server_rpc_notify] 0-archive1-ser > ver: disconnecting connection from > CTX_ID:e3871169-af62-44aa-a990-fa4248283c08-GRAPH_ID:0-PID: > 31830-HOST:fs-lrunning-c1-n1-PC_NAME:archive1-client-0-RECON_NO:-0 > [2019-02-06 14:15:09.561568] I [MSGID: 101055] > [client_t.c:435:gf_client_unref] 0-archive1-ser > ver: Shutting down connection > CTX_ID:e3871169-af62-44aa-a990-fa4248283c08-GRAPH_ID:0-PID:31830-HOST:fs-lrunning-c1-n1-PC_NAME:archive1-client-0-RECON_NO:-0 > [2019-02-06 14:15:10.188406] I [glusterfsd-mgmt.c:58:mgmt_cbk_spec] > 0-mgmt: Volume file changed > [2019-02-06 14:15:10.201029] I [glusterfsd-mgmt.c:2005:mgmt_getspec_cbk] > 0-glusterfs: No change in volfile,continuing > [2019-02-06 14:15:10.514721] I [glusterfsd-mgmt.c:58:mgmt_cbk_spec] > 0-mgmt: Volume file changed > [2019-02-06 14:15:10.526216] I [glusterfsd-mgmt.c:2005:mgmt_getspec_cbk] > 0-glusterfs: No change in volfile,continuing > The message "E [MSGID: 101191] > [event-epoll.c:671:event_dispatch_epoll_worker] 0-epoll: Failed to dispatch > handler" repeated 79 times between [2019-02-06 14:15:09.499105] and > [2019-02-06 14:15:10.682592] > [2019-02-06 14:15:10.684204] E [MSGID: 116020] > [bit-rot-stub.c:647:br_stub_check_bad_object] 0-archive1-bitrot-stub: > 23b623cb-7256-4fe6-85b0-1026b1531a86 is a bad object. Returning > [2019-02-06 14:15:10.684262] E [MSGID: 115070] > [server-rpc-fops_v2.c:1503:server4_open_cbk] 0-archive1-server: 2146148: > OPEN /data/file1.txt (23b623cb-7256-4fe6-85b0-1026b1531a86), client: > CTX_ID:0545b52c-2843-4833-a5fc-b11e062a72d3-GRAPH_ID:0-PID:2458-HOST:fs-lrunning-c1-n1-PC_NAME:archive1-client-0-RECON_NO:-3, > error-xlator: archive1-bitrot-stub [Input/output error] > [2019-02-06 14:15:10.684949] E [MSGID: 116020] > [bit-rot-stub.c:647:br_stub_check_bad_object] 0-archive1-bitrot-stub: > 23b623cb-7256-4fe6-85b0-1026b1531a86 is a bad object. Returning > [2019-02-06 14:15:10.684982] E [MSGID: 115070] > [server-rpc-fops_v2.c:1503:server4_open_cbk] 0-archive1-server: 2146149: > OPEN /data/file1.txt (23b623cb-7256-4fe6-85b0-1026b1531a86), client: > CTX_ID:0545b52c-2843-4833-a5fc-b11e062a72d3-GRAPH_ID:0-PID:2458-HOST:fs-lrunning-c1-n1-PC_NAME:archive1-client-0-RECON_NO:-3, > error-xlator: archive1-bitrot-stub [Input/output error] > [2019-02-06 14:15:10.686566] E [MSGID: 116020] > [bit-rot-stub.c:647:br_stub_check_bad_object] 0-archive1-bitrot-stub: > 23b623cb-7256-4fe6-85b0-1026b1531a86 is a bad object. Returning > [2019-02-06 14:15:10.686600] E [MSGID: 115070] > [server-rpc-fops_v2.c:1503:server4_open_cbk] 0-archive1-server: 2146150: > OPEN /data/file1.txt (23b623cb-7256-4fe6-85b0-1026b1531a86), client: > CTX_ID:0545b52c-2843-4833-a5fc-b11e062a72d3-GRAPH_ID:0-PID:2458-HOST:fs-lrunning-c1-n1-PC_NAME:archive1-client-0-RECON_NO:-3, > error-xlator: archive1-bitrot-stub [Input/output error] > [2019-02-06 14:15:11.189361] I [glusterfsd-mgmt.c:58:mgmt_cbk_spec] > 0-mgmt: Volume file changed > [2019-02-06 14:15:11.207835] I [glusterfsd-mgmt.c:58:mgmt_cbk_spec] > 0-mgmt: Volume file changed > [2019-02-06 14:15:11.220763] I
Re: [Gluster-users] Corrupted File readable via FUSE?
Hi David, Do you have any bricks down? Can you please share the output of the following commands and also the logs of the server and the client nodes? 1) gluster volume info 2) gluster volume status 3) gluster volume bitrot scrub status Few more questions 1) How many copies of the file were corrupted? (All? Or Just one?) 2 things I am trying to understand A) IIUC, if only one copy is corrupted, then the replication module from the gluster client should serve the data from the remaining good copy B) If all the copies were corrupted (or say more than quorum copies were corrupted which means 2 in case of 3 way replication) then there will be an error to the application. But the error to be reported should 'Input/Output Error'. Not 'Transport endpoint not connected' 'Transport endpoint not connected' error usually comes when a brick where the operation is being directed to is not connected to the client. Regards, Raghavendra On Mon, Feb 4, 2019 at 6:02 AM David Spisla wrote: > Hello Amar, > sounds good. Until now this patch is only merged into master. I think it > should be part of the next v5.x patch release! > > Regards > David > > Am Mo., 4. Feb. 2019 um 09:58 Uhr schrieb Amar Tumballi Suryanarayan < > atumb...@redhat.com>: > >> Hi David, >> >> I guess https://review.gluster.org/#/c/glusterfs/+/21996/ helps to fix >> the issue. I will leave it to Raghavendra Bhat to reconfirm. >> >> Regards, >> Amar >> >> On Fri, Feb 1, 2019 at 8:45 PM David Spisla wrote: >> >>> Hello Gluster Community, >>> I have got a 4 Node Cluster with a Replica 4 Volume, so each node has a >>> brick with a copy of a file. Now I tried out the bitrot functionality and >>> corrupt the copy on the brick of node1. After this I scrub ondemand and the >>> file is marked correctly as corrupted. >>> >>> No I try to read that file from FUSE on node1 (with corrupt copy): >>> $ cat file1.txt >>> cat: file1.txt: Transport endpoint is not connected >>> FUSE log says: >>> >>> *[2019-02-01 15:02:19.191984] E [MSGID: 114031] >>> [client-rpc-fops_v2.c:281:client4_0_open_cbk] 0-archive1-client-0: remote >>> operation failed. Path: /data/file1.txt >>> (b432c1d6-ece2-42f2-8749-b11e058c4be3) [Input/output error]* >>> [2019-02-01 15:02:19.192269] W [dict.c:761:dict_ref] >>> (-->/usr/lib64/glusterfs/5.3/xlator/performance/quick-read.so(+0x7329) >>> [0x7fc642471329] >>> -->/usr/lib64/glusterfs/5.3/xlator/performance/io-cache.so(+0xaaf5) >>> [0x7fc642682af5] -->/usr/lib64/libglusterfs.so.0(dict_ref+0x58) >>> [0x7fc64a78d218] ) 0-dict: dict is NULL [Invalid argument] >>> [2019-02-01 15:02:19.192714] E [MSGID: 108009] >>> [afr-open.c:220:afr_openfd_fix_open_cbk] 0-archive1-replicate-0: Failed to >>> open /data/file1.txt on subvolume archive1-client-0 [Input/output error] >>> *[2019-02-01 15:02:19.193009] W [fuse-bridge.c:2371:fuse_readv_cbk] >>> 0-glusterfs-fuse: 147733: READ => -1 >>> gfid=b432c1d6-ece2-42f2-8749-b11e058c4be3 fd=0x7fc60408bbb8 (Transport >>> endpoint is not connected)* >>> [2019-02-01 15:02:19.193653] W [MSGID: 114028] >>> [client-lk.c:347:delete_granted_locks_owner] 0-archive1-client-0: fdctx not >>> valid [Invalid argument] >>> >>> And from FUSE on node2 (with heal copy): >>> $ cat file1.txt >>> file1 >>> >>> It seems to be that node1 wants to get the file from its own brick, but >>> the copy there is broken. Node2 gets the file from its own brick with a >>> heal copy, so reading the file succeed. >>> But I am wondering myself because sometimes reading the file from node1 >>> with the broken copy succeed >>> >>> What is the expected behaviour here? Is it possibly to read files with a >>> corrupted copy from any client access? >>> >>> Regards >>> David Spisla >>> >>> >>> ___ >>> Gluster-users mailing list >>> Gluster-users@gluster.org >>> https://lists.gluster.org/mailman/listinfo/gluster-users >> >> >> >> -- >> Amar Tumballi (amarts) >> > ___ Gluster-users mailing list Gluster-users@gluster.org https://lists.gluster.org/mailman/listinfo/gluster-users
Re: [Gluster-users] Corrupted File readable via FUSE?
Hello Amar, sounds good. Until now this patch is only merged into master. I think it should be part of the next v5.x patch release! Regards David Am Mo., 4. Feb. 2019 um 09:58 Uhr schrieb Amar Tumballi Suryanarayan < atumb...@redhat.com>: > Hi David, > > I guess https://review.gluster.org/#/c/glusterfs/+/21996/ helps to fix > the issue. I will leave it to Raghavendra Bhat to reconfirm. > > Regards, > Amar > > On Fri, Feb 1, 2019 at 8:45 PM David Spisla wrote: > >> Hello Gluster Community, >> I have got a 4 Node Cluster with a Replica 4 Volume, so each node has a >> brick with a copy of a file. Now I tried out the bitrot functionality and >> corrupt the copy on the brick of node1. After this I scrub ondemand and the >> file is marked correctly as corrupted. >> >> No I try to read that file from FUSE on node1 (with corrupt copy): >> $ cat file1.txt >> cat: file1.txt: Transport endpoint is not connected >> FUSE log says: >> >> *[2019-02-01 15:02:19.191984] E [MSGID: 114031] >> [client-rpc-fops_v2.c:281:client4_0_open_cbk] 0-archive1-client-0: remote >> operation failed. Path: /data/file1.txt >> (b432c1d6-ece2-42f2-8749-b11e058c4be3) [Input/output error]* >> [2019-02-01 15:02:19.192269] W [dict.c:761:dict_ref] >> (-->/usr/lib64/glusterfs/5.3/xlator/performance/quick-read.so(+0x7329) >> [0x7fc642471329] >> -->/usr/lib64/glusterfs/5.3/xlator/performance/io-cache.so(+0xaaf5) >> [0x7fc642682af5] -->/usr/lib64/libglusterfs.so.0(dict_ref+0x58) >> [0x7fc64a78d218] ) 0-dict: dict is NULL [Invalid argument] >> [2019-02-01 15:02:19.192714] E [MSGID: 108009] >> [afr-open.c:220:afr_openfd_fix_open_cbk] 0-archive1-replicate-0: Failed to >> open /data/file1.txt on subvolume archive1-client-0 [Input/output error] >> *[2019-02-01 15:02:19.193009] W [fuse-bridge.c:2371:fuse_readv_cbk] >> 0-glusterfs-fuse: 147733: READ => -1 >> gfid=b432c1d6-ece2-42f2-8749-b11e058c4be3 fd=0x7fc60408bbb8 (Transport >> endpoint is not connected)* >> [2019-02-01 15:02:19.193653] W [MSGID: 114028] >> [client-lk.c:347:delete_granted_locks_owner] 0-archive1-client-0: fdctx not >> valid [Invalid argument] >> >> And from FUSE on node2 (with heal copy): >> $ cat file1.txt >> file1 >> >> It seems to be that node1 wants to get the file from its own brick, but >> the copy there is broken. Node2 gets the file from its own brick with a >> heal copy, so reading the file succeed. >> But I am wondering myself because sometimes reading the file from node1 >> with the broken copy succeed >> >> What is the expected behaviour here? Is it possibly to read files with a >> corrupted copy from any client access? >> >> Regards >> David Spisla >> >> >> ___ >> Gluster-users mailing list >> Gluster-users@gluster.org >> https://lists.gluster.org/mailman/listinfo/gluster-users > > > > -- > Amar Tumballi (amarts) > ___ Gluster-users mailing list Gluster-users@gluster.org https://lists.gluster.org/mailman/listinfo/gluster-users
Re: [Gluster-users] Corrupted File readable via FUSE?
Hi David, I guess https://review.gluster.org/#/c/glusterfs/+/21996/ helps to fix the issue. I will leave it to Raghavendra Bhat to reconfirm. Regards, Amar On Fri, Feb 1, 2019 at 8:45 PM David Spisla wrote: > Hello Gluster Community, > I have got a 4 Node Cluster with a Replica 4 Volume, so each node has a > brick with a copy of a file. Now I tried out the bitrot functionality and > corrupt the copy on the brick of node1. After this I scrub ondemand and the > file is marked correctly as corrupted. > > No I try to read that file from FUSE on node1 (with corrupt copy): > $ cat file1.txt > cat: file1.txt: Transport endpoint is not connected > FUSE log says: > > *[2019-02-01 15:02:19.191984] E [MSGID: 114031] > [client-rpc-fops_v2.c:281:client4_0_open_cbk] 0-archive1-client-0: remote > operation failed. Path: /data/file1.txt > (b432c1d6-ece2-42f2-8749-b11e058c4be3) [Input/output error]* > [2019-02-01 15:02:19.192269] W [dict.c:761:dict_ref] > (-->/usr/lib64/glusterfs/5.3/xlator/performance/quick-read.so(+0x7329) > [0x7fc642471329] > -->/usr/lib64/glusterfs/5.3/xlator/performance/io-cache.so(+0xaaf5) > [0x7fc642682af5] -->/usr/lib64/libglusterfs.so.0(dict_ref+0x58) > [0x7fc64a78d218] ) 0-dict: dict is NULL [Invalid argument] > [2019-02-01 15:02:19.192714] E [MSGID: 108009] > [afr-open.c:220:afr_openfd_fix_open_cbk] 0-archive1-replicate-0: Failed to > open /data/file1.txt on subvolume archive1-client-0 [Input/output error] > *[2019-02-01 15:02:19.193009] W [fuse-bridge.c:2371:fuse_readv_cbk] > 0-glusterfs-fuse: 147733: READ => -1 > gfid=b432c1d6-ece2-42f2-8749-b11e058c4be3 fd=0x7fc60408bbb8 (Transport > endpoint is not connected)* > [2019-02-01 15:02:19.193653] W [MSGID: 114028] > [client-lk.c:347:delete_granted_locks_owner] 0-archive1-client-0: fdctx not > valid [Invalid argument] > > And from FUSE on node2 (with heal copy): > $ cat file1.txt > file1 > > It seems to be that node1 wants to get the file from its own brick, but > the copy there is broken. Node2 gets the file from its own brick with a > heal copy, so reading the file succeed. > But I am wondering myself because sometimes reading the file from node1 > with the broken copy succeed > > What is the expected behaviour here? Is it possibly to read files with a > corrupted copy from any client access? > > Regards > David Spisla > > > ___ > Gluster-users mailing list > Gluster-users@gluster.org > https://lists.gluster.org/mailman/listinfo/gluster-users -- Amar Tumballi (amarts) ___ Gluster-users mailing list Gluster-users@gluster.org https://lists.gluster.org/mailman/listinfo/gluster-users
[Gluster-users] Corrupted File readable via FUSE?
Hello Gluster Community, I have got a 4 Node Cluster with a Replica 4 Volume, so each node has a brick with a copy of a file. Now I tried out the bitrot functionality and corrupt the copy on the brick of node1. After this I scrub ondemand and the file is marked correctly as corrupted. No I try to read that file from FUSE on node1 (with corrupt copy): $ cat file1.txt cat: file1.txt: Transport endpoint is not connected FUSE log says: *[2019-02-01 15:02:19.191984] E [MSGID: 114031] [client-rpc-fops_v2.c:281:client4_0_open_cbk] 0-archive1-client-0: remote operation failed. Path: /data/file1.txt (b432c1d6-ece2-42f2-8749-b11e058c4be3) [Input/output error]* [2019-02-01 15:02:19.192269] W [dict.c:761:dict_ref] (-->/usr/lib64/glusterfs/5.3/xlator/performance/quick-read.so(+0x7329) [0x7fc642471329] -->/usr/lib64/glusterfs/5.3/xlator/performance/io-cache.so(+0xaaf5) [0x7fc642682af5] -->/usr/lib64/libglusterfs.so.0(dict_ref+0x58) [0x7fc64a78d218] ) 0-dict: dict is NULL [Invalid argument] [2019-02-01 15:02:19.192714] E [MSGID: 108009] [afr-open.c:220:afr_openfd_fix_open_cbk] 0-archive1-replicate-0: Failed to open /data/file1.txt on subvolume archive1-client-0 [Input/output error] *[2019-02-01 15:02:19.193009] W [fuse-bridge.c:2371:fuse_readv_cbk] 0-glusterfs-fuse: 147733: READ => -1 gfid=b432c1d6-ece2-42f2-8749-b11e058c4be3 fd=0x7fc60408bbb8 (Transport endpoint is not connected)* [2019-02-01 15:02:19.193653] W [MSGID: 114028] [client-lk.c:347:delete_granted_locks_owner] 0-archive1-client-0: fdctx not valid [Invalid argument] And from FUSE on node2 (with heal copy): $ cat file1.txt file1 It seems to be that node1 wants to get the file from its own brick, but the copy there is broken. Node2 gets the file from its own brick with a heal copy, so reading the file succeed. But I am wondering myself because sometimes reading the file from node1 with the broken copy succeed What is the expected behaviour here? Is it possibly to read files with a corrupted copy from any client access? Regards David Spisla ___ Gluster-users mailing list Gluster-users@gluster.org https://lists.gluster.org/mailman/listinfo/gluster-users
