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.
On Sun, Dec 8, 2013 at 6:46 PM, Amitay Isaacs <[email protected]> wrote: > The branch, 1.2.40 has been updated > via f0416afa6dc776489e2c7b3f0ede3480a3035bdb (commit) > via bcb73f0b9c48fab6ac89e4ec0a297fd156beb6f6 (commit) > from 5620a403f3ed8dd1ec8dcb449de4cf8a840893a8 (commit) > > http://gitweb.samba.org/?p=ctdb.git;a=shortlog;h=1.2.40 > > > - Log ----------------------------------------------------------------- > commit f0416afa6dc776489e2c7b3f0ede3480a3035bdb > Author: Martin Schwenke <[email protected]> > Date: Mon Dec 9 12:02:47 2013 +1100 > > New version 1.2.71 > > Signed-off-by: Martin Schwenke <[email protected]> > > commit bcb73f0b9c48fab6ac89e4ec0a297fd156beb6f6 > Author: Martin Schwenke <[email protected]> > Date: Mon Dec 9 11:47:01 2013 +1100 > > eventscripts: Do not restart NFS on reconfigure, just do statd notify > > See also commit 2629de72e1f37b5e46772c2ef8d8d0012fc4ed37 in the master > branch. > > In this branch the NFS restart was accidentally re-added in > 30e05c8b14413f3e1150b01e856d4b51c675ea5c and formalised in > 3fd4da153703bc83cf102ad367dd20d873e938a0 when the mistake wasn't > realised. > > Signed-off-by: Martin Schwenke <[email protected]> > > ----------------------------------------------------------------------- > > Summary of changes: > config/events.d/60.nfs | 17 ++++++++++------- > packaging/RPM/ctdb.spec.in | 4 +++- > 2 files changed, 13 insertions(+), 8 deletions(-) > > > Changeset truncated at 500 lines: > > diff --git a/config/events.d/60.nfs b/config/events.d/60.nfs > index f567c82..13a2b04 100755 > --- a/config/events.d/60.nfs > +++ b/config/events.d/60.nfs > @@ -14,7 +14,16 @@ start_nfs() { > service_name="nfs" > service_start="start_nfs" > service_stop="startstop_nfs stop" > -service_reconfigure="startstop_nfs restart" > +service_reconfigure="reconfigure_nfs" > + > +reconfigure_nfs () > +{ > + # If IPs have been reallocated, we must restart the lock manager > + # across all nodes and notify statd clients. > + if [ -x "$CTDB_BASE/statd-callout" ] ; then > + "$CTDB_BASE/statd-callout" notify & > + fi >/dev/null 2>&1 > +} > > nfs_check_thread_count () > { > @@ -200,12 +209,6 @@ case "$1" in > ctdb_service_reconfigure > exit 0 > fi > - > - # if the ips have been reallocated, we must restart the lockmanager > - # across all nodes and ping all statd listeners > - [ -x $CTDB_BASE/statd-callout ] && { > - $CTDB_BASE/statd-callout notify & > - } >/dev/null 2>&1 > ;; > *) > ctdb_standard_event_handler "$@" > diff --git a/packaging/RPM/ctdb.spec.in b/packaging/RPM/ctdb.spec.in > index 2792b59..715a824 100644 > --- a/packaging/RPM/ctdb.spec.in > +++ b/packaging/RPM/ctdb.spec.in > @@ -3,7 +3,7 @@ Name: ctdb > Summary: Clustered TDB > Vendor: Samba Team > Packager: Samba Team <[email protected]> > -Version: 1.2.70 > +Version: 1.2.71 > Release: 1GITHASH > Epoch: 0 > License: GNU GPL version 3 > @@ -155,6 +155,8 @@ development libraries for ctdb > > %changelog > > +* Mon Dec 09 2013 : Version 1.2.71 > + - Don't restart NFS on failover, just do statd notification > * Mon Nov 25 2013 : Version 1.2.70 > - Just warn when recovery fails to update flags on inactive node > * Mon Nov 04 2013 : Version 1.2.69 > > > -- > CTDB repository
