[Gluster-devel] upstream: Symbolic link not getting healed

Venkatesh Somyajulu vsomyaju at redhat.com
Tue Dec 17 10:44:57 UTC 2013


For the upstream master branch, I found that symbolic link is not getting healed.

How I reproduced:
1. Created replicate volume with 2 bricks in a replica.
2. Created file from the mount point.
3. Killed one of the brick of replica.
4. Created symbolic link to that file from mount point and then brought the killed brick back up.

Tried to heal by both a) Mount Process and b) Self heal Daemon

a) When self heal daemon is off:
6. Gave ls at mount point.
   Observation: Rather than file getting healed, getting this output. 
               ls: cannot read symbolic link Link: No such file or directory
                File  Link

b) When self heal daemon is on:
6. "gluster volume heal volumename full" fails to heal and the output includes:
   [2013-12-17 10:21:49.863960] I [afr-self-heal-entry.c:1502:afr_sh_entry_impunge_readlink_sink_cbk] 0-volume1-replicate-0: readlink of /Link on volume1-client-1 failed (Stale file handle)

Still root causing the issue. Seems link ESTALE error needs to be handled properly.


More information about the Gluster-devel mailing list