[Bugs] [Bug 1356942] Problem with persistant mount on quorum fail when using SSL

[bugzilla at redhat.com]

https://bugzilla.redhat.com/show_bug.cgi?id=1356942



--- Comment #16 from Mohit Agrawal <moagrawa at redhat.com> ---
Hi,

  I am able to reproduce the issue in my vm environment without enable
SSL.After checked the logs I think it is defined behavior.As you also mentioned
issue is occurred only when mount is remaining from when quorum is lost.

Reproduction steps
  1) Create 1X3 cluster 
  2) Activate Quorum (Set cluster.quorum-type to auto)
  3) Mount the client
  4) Run below command on client to write a small line in a file
     cd /mnt;cnt=0;while [ 1 ]; do   echo "instance no. is $cnt";   echo "Start
instance is $cnt" >> f1;   cnt=`expr $cnt + 1 `; done
  5) kill glusterfsd  on two nodes
  6) Got "Read-only file system failure messages on client
  7) Press CTRL-C to kill while loop 
  8) After checked the contents of f1 file it shows after kill glusterfsd no
line was not updated in file.
  9) Restart glusterd to start glusterfsd on one node after regain quorum it
will show message like ("Connected to replicate-client-0) 


The logs(Read-only file system) are showing in mnt-03.log when quorum is not
met and client is trying to write a buffer on 
mount point continuously.

>>>>>>>>>>>>>>>

[2016-07-15 16:11:20.399760] W [fuse-bridge.c:1290:fuse_err_cbk]
0-glusterfs-fuse: 97: FLUSH() ERR => -1 (Read-only file system)
[2016-07-15 16:11:21.725593] W [fuse-bridge.c:1290:fuse_err_cbk]
0-glusterfs-fuse: 105: FLUSH() ERR => -1 (Read-only file system)
[2016-07-15 16:11:22.310991] W [fuse-bridge.c:1290:fuse_err_cbk]
0-glusterfs-fuse: 112: FLUSH() ERR => -1 (Read-only file system)
[2016-07-15 16:11:22.951805] W [fuse-bridge.c:1290:fuse_err_cbk]
0-glusterfs-fuse: 120: FLUSH() ERR => -1 (Read-only file system)
[2016-07-15 16:11:43.370784] W [fuse-bridge.c:1290:fuse_err_cbk]
0-glusterfs-fuse: 128: FLUSH() ERR => -1 (Read-only file system)
[2016-07-15 16:11:44.233100] W [fuse-bridge.c:1290:fuse_err_cbk]
0-glusterfs-fuse: 135: FLUSH() ERR => -1 (Read-only file system)
[2016-07-15 16:11:45.018485] W [fuse-bridge.c:1290:fuse_err_cbk]
0-glusterfs-fuse: 143: FLUSH() ERR => -1 (Read-only file system)
[2016-07-15 16:12:28.109760] W [fuse-bridge.c:1290:fuse_err_cbk]
0-glusterfs-fuse: 227: FLUSH() ERR => -1 (Read-only file system)

>>>>>>>>>>>>>>>

After regain the quorum it will show below kind of message in mnt.log

>>>>>>>>>>>>>>>>>>>>>>

2016-07-18 10:12:12.167800] I [MSGID: 114046]
[client-handshake.c:1222:client_setvolume_cbk] 0-replicate-client-1: Connected
to replicate-client-1, attached to remote volume '/dist1/brick2'.
[2016-07-18 10:12:12.167841] I [MSGID: 114047]
[client-handshake.c:1233:client_setvolume_cbk] 0-replicate-client-1: Server and
Client lk-version numbers are not same, reopening the fds
[2016-07-18 10:12:12.167863] I [MSGID: 114042]
[client-handshake.c:1053:client_post_handshake] 0-replicate-client-1: 1 fds
open - Delaying child_up until they are re-opened

>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>



As per documentation if cluster.quorum-type is auto It allows writes to the
file only if the percentage of active replicate
bricks is more than 50% of the total number of bricks that constitute that
replica so i think it is expected behavior.After stopped glusterfsd on one node
you will not get any error but after kill glusterfsd on other node you will get
this kind of error messages. 

Regards
Mohit Agrawal

-- 
You are receiving this mail because:
You are on the CC list for the bug.
Unsubscribe from this bug https://bugzilla.redhat.com/token.cgi?t=gFJxmI1X6m&a=cc_unsubscribe