[Gluster-users] Bug - replace-brick commit fails after half a dozen calls / status reports failure, then success

Toby Corkindale toby.corkindale at strategicdata.com.au
Thu Aug 8 07:36:33 UTC 2013

Having built a fresh Gluster cluster, this time out of Ubuntu LTS with 
Gluster 3.3.2, we've found that the replace-brick command now seems to 
succeed. (Unlike our Debian 6 + 3.3.1 cluster before it)

I say "seems" to succeed, because it fails after about half a dozen 
volumes have been committed.

I have a shell script that was calling replace-brick .... commit over 
quite a lot of volumes. Every time I ran it, it would succeed on a few 
more volumes, then fail for all the remaining ones.

Waiting a minute and then re-running it would succeed for a few more 
down the list.

Anyway, I thought I'd report this.

As a minor annoyance, the "status" mode of the command initially had me 
filled with despair, because it was reporting operation failed at 
first.. however after some more time went by, it started reporting 
progress, than success.

I have no idea why that is, but I guess another bug?


More information about the Gluster-users mailing list