glustershd needs to be restarted for volume start/stop operations
atinmu opened this issue · 1 comments
Based on code reading, it was found that glustershd wasn't restarted on volume start/stop operations which was resulting shd process to not having the latest volfile and it was continuing to attempt to connect to the stale bricks and GD2 logs were flooding with
time="2019-01-03 15:43:09.300668" level=info msg="client disconnected" address="172.17.0.4:996" server=sunrpc source="[server.go:109:sunrpc.(*SunRPC).pruneConn]"
time="2019-01-03 15:43:09.815217" level=info msg="client connected" address="172.17.0.3:996" server=sunrpc source="[server.go:148:sunrpc.(*SunRPC).acceptLoop]" transport=tcp
time="2019-01-03 15:43:09.818169" level=error msg="registry.SearchByBrickPath() failed for brick" brick=/tmp/b1 error="SearchByBrickPath: port for brick /tmp/b1 not found" source="[rpc_prog.go:104:pmap.(*GfPortmap).PortByBrick]"
glustershd will not be enabled with GCS/1.0. client side self heal can still heal the files in case there's a catch up to do when such a file under question is accessed from the mount point.