[Bugs] [Bug 1741783] volume heal info show nothing, while visiting from mount point blame "no such entry"

bugzilla at redhat.com bugzilla at redhat.com
Mon Aug 19 12:48:16 UTC 2019


Karthik U S <ksubrahm at redhat.com> changed:

           What    |Removed                     |Added
              Flags|                            |needinfo?(zz.sh.cynthia at gma
                   |                            |il.com)

--- Comment #9 from Karthik U S <ksubrahm at redhat.com> ---
>From the outputs provided, the file
"mn-1__dbim-redis.service__database-nosql-cmredis.sync_state.tmp" is present
only on brick "mn-0.local:/mnt/bricks/services/brick" without the gfid-link
under .glusterfs dir. On the other two bricks the file itself is not present.
By looking at the state of the file one possible scenario that I can think of
which can lead to this is:
- File creation succeeded on 1st brick
- Gfid assignment succeeds but gfid-link creation fails on that brick
- On the 2nd & 3rd brick file creation itself will fail
Are there frequent disconnects of the bricks?

I do not see the following message in the source code in version 3.12.15.
[2019-08-18 09:04:45.976644] I [MSGID: 108026]
[afr-self-heald.c:446:afr_shd_index_heal] 0-services-replicate-0: ret =
-2,purge gfid:5f2dc6c6-3fc4-41ba-9800-a560b483de13
Is this something added by you or anyone from your team? Is this a source
install or rpm install?
Give the outputs of "gluster --version" and "rpm -qa | grep gluster".
This is an INFO log according to the log level in the message. So how did you
come to the conclusion that entry purge failed? Was this file present on the
brick even after this message was logged?
SHD caught this entry during heal, so most probably this file or its parent had
some pending marker on them. Or did a full heal was run on this volume?

You are receiving this mail because:
You are on the CC list for the bug.

More information about the Bugs mailing list