[Gluster-users] More Troubles with 3.2.2

Georg Höllrigl georg.hoellrigl at xidras.com
Mon Aug 1 10:47:49 UTC 2011


Due to previous errors we had deaktivated the replication to the second glusterfs brick. The first 
try to activate the second brick went terribly wrong, ending in lots of old data.

For our new try, we removed all the old data before adding the second brick. This looked quiet good, 
and I tried to do the complete selfheal with find | xargs stat. But this didn't work as expected. 
The second brick by now conatins only about 10 - 30 % of the data instead of a full replicate! I 
also was getting errors about files that couldn't be accessed (but was found by the find command).

Because of this I removed the second brick again.

But this still is not the end of the story! Now the Data seems to be there. But some files are 
missing. Even while they are not on one of the two bricks, the clients see them via a samba share! 
Also on the mountpoint of the samba share different files in compairison to the backend are visible! 
Partly there are only files which were stored 14 - 20 days ago, but no newer files.

Now there are a few questions:

Is there any possible way to add an replicated brick safley?
Where can such different views from server and client come from?
Where can the files be found, which I don't see on the backend? Even after remounting the 
filesystem, the files stay the same, but vanish when restarting the server?


More information about the Gluster-users mailing list