Created attachment 590233 [details] glusterfs log When trying to create several bricks, failure to create one brick will cause all bricks to fail. For example: * Create brick from host:/brick1. * Create 3 bricks from host:/brick1 host:/brick2 host:/brick3 * brick1 fails to create because it already exists. But also brick2 and brick3 fail to create, even though they should succeed.
can you elaborate what you mean by 'creating a brick' ? we can create a volume, not a brick.
(In reply to comment #2) > can you elaborate what you mean by 'creating a brick' ? we can create a > volume, not a brick. What I mean is adding more bricks to an existing volume. To make the process fail, I add a combination of existing and non-existing bricks. This is the command being run through vdsm: /usr/sbin/gluster --mode=script volume add-brick stripe1 stripe 4 10.35.97.47:/stripe3 10.35.97.47:/stripe4 10.35.97.47:/stripe5 10.35.97.47:/stripe6
This is an expected behaviour. As part of volume creation, if any of the brick creation fails then we fail the complete transaction and doesn't continue with creating partial bricks.