[heketi-devel] Need review of few patches

Raghavendra Talur rtalur at redhat.com
Thu Apr 6 01:33:11 UTC 2017


Hi,

1. There is a Gluster limitation that consecutive replace brick
operations performed can lead to data loss if self heal wasn't
completed in between each of the operation.

The PR to fix this is https://github.com/heketi/heketi/pull/718

2. We discovered another possibility of DB going into inconsistent
state due to remove device code. This can happen when volume creates
are performed in parallel to remove device and both operations use the
same new device. The state stored back into the disk might not have
bricks added by volume create.

The PR to fix this is: https://github.com/heketi/heketi/pull/736

3. I am also working on the refactoring of code, but it is failing a test
PR https://github.com/heketi/heketi/pull/737

Thanks,
Raghavendra Talur


More information about the heketi-devel mailing list