[Gluster-users] How to force heal a disperse volume

[Ashish Pandey]

Hi Dan,

Please find my answers in reply to your questions/problem. 

 
Previously in our distributed+replicated setup, we could force heals
either with "gluster volume heal" commands.  This doesn't work with a
disperse setup.

>>>>>
There are 2 types of heal commands , index heal and full heal.

index heal - gluster v heal <volume name> 
full heal - gluster v heal <volume name> full

How did you find that heal does not work? Did you get any error messages? Or
Did you check the size of files on bricks and found different size on bricks? Or 
you used "gluster v heal <volume name> info" command which shows the list of files and directories to be healed?

In case you have not tried full heal, Please try it once. It might take long time (depends on the number of files) as it scan each and every file and decides if it requires heal or not.
In case you don't get success, please provide logs of /var/log/glusterfs/glustershd.log to debug it further.

>>>>>

Are there any tricks anyone can offer to manually trigger a heal?  Do
the old "find /blah -exec stat {}\;" or "ls -laR /blah" commands work
to force Gluster to do heals on files within disperse volumes?

>>>>>
If you access a file from mount point and if it requires heal, it will trigger heal for that file.

To trigger heal manually we use index heal or full heal commands.
index heal or full heal must work if a file needs to be healed.


Ashish





----- Original Message -----
From: "Dan Mons" <dmons at cuttingedge.com.au>
To: "gluster-users" <gluster-users at gluster.org>
Sent: Monday, November 23, 2015 4:38:25 AM
Subject: [Gluster-users] How to force heal a disperse volume

Hi there Gluster users,

We've recently upgraded our local Gluster install, and we're now
running 12 nodes in a disperse setup (10:2 data:redundancy).

Previously in our distributed+replicated setup, we could force heals
either with "gluster volume heal" commands.  This doesn't work with a
disperse setup.

Are there any tricks anyone can offer to manually trigger a heal?  Do
the old "find /blah -exec stat {}\;" or "ls -laR /blah" commands work
to force Gluster to do heals on files within disperse volumes?

Many thanks,

----------------
Dan Mons - Skunk Works
Cutting Edge
http://cuttingedge.com.au
_______________________________________________
Gluster-users mailing list
Gluster-users at gluster.org
http://www.gluster.org/mailman/listinfo/gluster-users