Hide Forgot
Description of problem: ======================== Consider the case where the storage node is stopped and upon restart the storage node gets a new ip_address or hostname. When the node comes back online, the bricks on the storage node is no more valid. This is because currently we recognize each brick by "<IP/Hostname>:<Brick_Directory>". Since the storage node gets a new hostname/ip , the brick paths should also be changed once the node comes back online. Version-Release number of selected component (if applicable): ============================================================ glusterfs 3.4.0.44.1u2rhs built on Nov 25 2013 08:17:39 How reproducible: =============== Often Steps to Reproduce: =================== 1. Create a 1 x 2 replicate volume . Start the volume. 2. Stop one of the storage nodes (shutdown). Restart the node. (Restart should change the IP/hostname of the node) 3. The glusterd of the storage node remains the same. The brick paths should be changed since there is change in hostname/ip.
This shall be addressed in glusterd2. Here's the upstream issue: https://github.com/gluster/glusterd2/issues/356
As Atin has already mentioned that glusterd is not able to resolve the bricks on glusterd restart when network interface is down. We have a similar kind of bug 1472267, which has been addressed and below is the upstream patch for the same. Upstream patch: https://review.gluster.org/#/c/17813/
Gaurav - Why is this BZ reopened? Please note fix for BZ 1472267 doesn't address this RFE and we'll consider this request only for GlusterD 2.0.