Hi !

Currently, I perform some test on XVS appliance powered by VMware ESX 3.5. The nodes are configured in dual-primary mode, and there is no problem when I have to shutdown properly one of them. Sync Ok !

But when I simulate a network problem (disconnect the NIC) or when I
made a snapshot of my XVS VM, the both node can't resync' and still
detected a Split-brain.

--root@lvs-node1 ~--# cat /proc/drbd
version: 8.2.5 (api:88/proto:86-88)
GIT-hash: 9faf052fdae5ef0c61b4d03890e2d2eab550610c build by root@lvs-node2.xtravirt.com, 2008-04-14 16:57:48

• 0: cs:StandAlone st:Primary/Unknown ds:UpToDate/Outdated r---*

ns:0 nr:0 dw:20970844 dr:20970844 al:0 bm:2560 lo:0 pe:0 ua:0 ap:0
resync: used:0/31 hits:2618796 misses:2560 starving:0 dirty:0 changed:2560
act_log: used:0/127 hits:0 misses:0 starving:0 dirty:0 changed:0

--root@lvs-node2 ~--# cat /proc/drbd
version: 8.2.5 (api:88/proto:86-88)
GIT-hash: 9faf052fdae5ef0c61b4d03890e2d2eab550610c build by root@lvs-node2.xtravirt.com, 2008-04-14 16:57:48

• 0: cs:StandAlone st:Primary/Unknown ds:UpToDate/Outdated r---*

ns:0 nr:0 dw:20970844 dr:20970844 al:0 bm:2560 lo:0 pe:0 ua:0 ap:0
resync: used:0/31 hits:2618796 misses:2560 starving:0 dirty:0 changed:2560
act_log: used:0/127 hits:0 misses:0 starving:0 dirty:0 changed:0

This is my drbd.conf:

root@lvs-node1 ~# cat /etc/drbd.conf
global {
+ usage-count no;+
}

common {
+ syncer {+
+ rate 100M;+
+ }+

+ handlers {+
+
outdate-peer "/usr/lib/heartbeat/outdate-peer.sh";+
+
split-brain "/usr/lib/drbd/outdate-peer.sh";+
+ }+
}

resource vmfs-0 {
+ protocol C;+

+ startup {+
+ become-primary-on both;+
+ degr-wfc-timeout 120;+
+ }+

+ net {+
+ after-sb-0pri discard-zero-changes;+
+ after-sb-1pri consensus;+
+ after-sb-2pri disconnect;+
+ allow-two-primaries;+
+ }+

+ disk {+
+ on-io-error pass_on;+
# fencing resource-only;
+ }+

+ on lvs-node1.xtravirt.com {+
+
device /dev/drbd0;+
+
disk /dev/sdb;+
+
address 10.1.14.130:7788;+
+ meta-disk internal;+
+ }+

+ on lvs-node2.xtravirt.com {+
+
device /dev/drbd0;+
+
disk /dev/sdb;+
+
address 10.1.14.132:7788;+
+ meta-disk internal;+
+ }+
}

I try all of the basics commands from drbdadm, like
connect all, up all, outdate all, invalidate all but still failed with
exiting code or it say as the device is held open by someone it can't
change state... How can I resolve Split Brain automatically? I want that the both nodes will come back in a dual-primary mode, once the split brain is resolved.

Best Regards!