[Linux-HA] Doesn't Heartbeat stop resources at system shutdown?

[Alan Robertson]

Shallyee Shang wrote:
> Hello, 
> 
> I am running heartbeat-1.0.4 on a two nodes cluster. I noticed that heartbeat 
> doesn't stop resource while system going down.
> 
> [ha-log]
> heartbeat: 2004/06/23_20:49:39 info: Heartbeat shutdown in progress. (1240)
> heartbeat: 2004/06/23_20:49:39 info: Client pid 1397 died (input)
> heartbeat: 2004/06/23_20:49:39 info: api_send_client: client 1397 died
> heartbeat: 2004/06/23_20:49:39 info: Giving up all HA resources.
> heartbeat: 2004/06/23_20:49:39 info: Core process 1239 exited. 11 remaining
> heartbeat: 2004/06/23_20:49:39 info: Core process 1238 exited. 10 remaining
> heartbeat: 2004/06/23_20:49:39 info: Core process 1237 exited. 9 remaining
> heartbeat: 2004/06/23_20:49:39 info: Core process 1236 exited. 8 remaining
> heartbeat: 2004/06/23_20:49:39 info: Core process 1235 exited. 7 remaining
> heartbeat: 2004/06/23_20:49:39 info: Core process 1234 exited. 6 remaining
> heartbeat: 2004/06/23_20:49:39 info: Core process 1233 exited. 5 remaining
> heartbeat: 2004/06/23_20:49:39 info: Core process 1232 exited. 4 remaining
> heartbeat: 2004/06/23_20:49:39 info: Core process 1231 exited. 3 remaining
> heartbeat: 2004/06/23_20:49:39 info: Core process 1230 exited. 2 remaining
> heartbeat: 2004/06/23_20:49:39 info: killing /usr/local/lib/heartbeat/ipfail process group 1397 with signal 15
> heartbeat: 2004/06/23_20:49:39 info: All HA resources relinquished.
> heartbeat: 2004/06/23_20:49:40 info: control process Received SIGQUIT
> heartbeat: 2004/06/23_20:49:40 info: Core process 1240 exited. 1 remaining
> heartbeat: 2004/06/23_20:49:40 info: Heartbeat shutdown complete.
> 
> This happened only at system shutdown, if I stop heartbeat by typing
> "service heartbeat stop", resources are released as expect.
> 
> heartbeat: 2004/06/21_09:56:06 info: Heartbeat shutdown in progress. (2341)
> heartbeat: 2004/06/21_09:56:06 info: Giving up all HA resources.
> heartbeat: 2004/06/21_09:56:06 info: Releasing resource group: harpro03pt datadisk::drbd0 datadisk::drbd1 datadisk::drbd
> 2 oracle oraas ipmasq 43.17.43.30 192.168.1.10
> heartbeat: 2004/06/21_09:56:06 info: Running /etc/ha.d/resource.d/IPaddr 192.168.1.10 stop
> ......
> 
> Resource rc scripts are disabled because they are expected start/stop by heartbeat.
> Is this normal behavior? Please advise me to solve this. Thank you.


Heartbeat stops all resources which report themselves as running when it 
shuts down.

But, if you kill all its processes with -15, instead of shutting it down, 
it might not stop all resources...

So one of two things comes to mind:
	There is no "Kxxheartbeat" script in your rc.d directory
	One of your resources is not reporting that it's running when it is
	Your system somehow shut down that resource before it stopped
		heartbeat

The first one seems most likely.

-- 
     Alan Robertson <alanr at unix.sh>

"Openness is the foundation and preservative of friendship...  Let me claim 
from you at all times your undisguised opinions." - William Wilberforce