[Bugs] [Bug 1775461] protocol/client : After "Remove lock recovery logic from client and server protocol translators" patch commit, causing add fcntl lock on the file twice.
bugzilla at redhat.com
bugzilla at redhat.com
Tue Dec 10 10:32:31 UTC 2019
https://bugzilla.redhat.com/show_bug.cgi?id=1775461
--- Comment #15 from Anoop C S <anoopcs at redhat.com> ---
(In reply to yinkui from comment #14)
> Options enabled with v3.7.X.
>
> [root at node-1 ~]# gluster v get test all | grep features.lock-heal
> features.lock-heal on
>
> [root at node-1 ~]# gluster v get test all | grep features.grace-timeout
> features.grace-timeout 10
Ah.. then this behaviour is expected as these options(and corresponding
implementation) got removed with v4.0 due to its incomplete and/or incorrect
logic in various disconnect scenarios. It was decided to place lock-healing
logic in replication layer rather than on protocol-client. In case you are
interested refer the following doc for some notes around the internals of
lock-healing implementation and issues:
https://anoopcs.fedorapeople.org/Lock%20recovery%20in%20GlusterFS.txt
--
You are receiving this mail because:
You are on the CC list for the bug.
You are the assignee for the bug.
More information about the Bugs
mailing list