Note: "permalinks" may not be as permanent as we would like,
direct links of old sources may well be a few messages off.
IMHO there are bug, but I don't know precise place.
In short:
after-resync-target "/sbin/drbdsetup $DRBD_MINOR primary";
- may cause long-time resource locking after initial sync.
Full background:
Gentoo, 2.6.34-rc7, tools from drbd 8.3.7, ocfs2 dual-primary.
I use next trick to resolve sb-2pri without double-reboot:
===
net {
after-sb-2pri violently-as0p;
rr-conflict call-pri-lost;
}
handlers {
pri-lost "echo b > /proc/sysrq-trigger ; reboot -f";
before-resync-target "/sbin/drbdsetup $DRBD_MINOR secondary";
## This place MAY cause resource locking:
after-resync-target "/sbin/drbdsetup $DRBD_MINOR primary";
}
===
(full commons:
http://raw.googlecode.com/svn/trunk/sys-cluster/drbd/files/global_common2.conf)
Yes, first goal - to reboot SyncTarget only required only "pri-lost" handler,
but idea to ensure no primary SyncTarget "before-resync-target" IMHO is good.
And it works. But "after-resync-target" may (not in any background, see
global_common2.conf) cause locking by pid. This pid contains "/sbin/drbdsetup
0 primary" and "kill" find no this pid and then lock is done - IMHO it is
actually zombie.