[SCM] CTDB repository - branch 1.2.40 updated - ctdb-1.2.70-2-gf0416af

[Martin Schwenke]

On Sun, 8 Dec 2013 18:57:55 -0800, ronnie sahlberg
<ronniesahlberg at gmail.com> wrote:

> Are you really bouncing the lock manager still?
> You need to make sure that they enter a grace period across the
> cluster so you can have lock revocery working semi-reliably.

This is really just an incremental improvement to stop restarting NFS
after a failover and to only run statd-callout when IPs have actually
changed.

There's no change to statd-callout here.  This version is still only
synchronising client information across the cluster every 10 minutes
(just under 20 minutes worse case), so I don't think we're regarding
lock recovery as incredibly reliable anyway.  :-)

What should we be doing in statd-callout instead of bouncing the lock
manager?

Thanks...

peace & happiness,
martin