nfs started ,after some time(few seconds ) on brick where "volume start" command is issued. 10.192.134.144#gluster volume start beta Starting volume beta has been successful 10.192.134.144#showmount -e localhost showmount: RPC: Program not registered 10.192.134.144#showmount -e localhost showmount: RPC: Program not registered 10.192.134.144#showmount -e localhost showmount: RPC: Program not registered 10.192.134.144#showmount -e localhost Export list for localhost: /beta *
This is expected behaviour from 3.1beta. For a distributed config, NFS will now wait till all bricks have connected before making them visible to nfs clients. I agree there is some confusion here whether the gnfs init failed or whether the bricks have not connected yet. I'll see if this can be made more definite.
This behaviour will change now because distribute supports doing a fresh lookup by itself. Changing to 3.1.1. Only requires removing code and I want to do this carefully.
Not relevant anymore since nfs init does not block on all child-ups from dht.
Does this bug needs verfication from QA side? If yes, Does the steps mentioned to introduce this bug are enough, else please provide test plan.
(In reply to comment #4) > Does this bug needs verfication from QA side? > > If yes, > > Does the steps mentioned to introduce this bug are enough, else please provide > test plan. No. It does not neet verification.