Downtime due to halt script failure

[Rui Vilao]

Greetings,

I have another question concerning MS/SG...
I am currently tunning my Cluster configuration in order to make in more
robust. One of my packages uses products like DCE,OTS and HP-OV DM.

Sometimes package switch fails. I have identified two main reasons:

- umount of DCE related FS fails;
- halt script timeout;

The consequence is that no package switch occurs and thus unplanned downtime!

The expected behaviour could be that the system TOCs in those situations.
I don t think I can achieve this using SERVICE_HALT_TIMEOUT or
HALT_SCRIPT_TIMEOUT.

Thanks in advance for your help,

Kind Regards,

Rui.

[Stefan Farrelly]

You need to supply some more info on your package failures. How about listing the errors from the package logfile ? then you can get some good suggestions as to how to code your package stop scripts to prevent them.

[James R. Ferguson]

Rui:

A good discussion, which may be helpful, is in document #UXSGKBAN00000102.

http://us-support.external.hp.com/cki/bin/doc.pl/sid=ecf5abe60bb957d939/screen=ckiDisplayDocument?docId=200000047385458

...JRF...

[Anthony Goonetilleke · Accepted solution]

Are'nt DCE related stuff in /opt????
What exactly are you trying to unmount?

[Alan Riggs]

We definitely need more information. At what point are the halt scripts failing? What FS are you attempting to umount that hangs? Is the timeout a consequence of teh umount problem or does it occur independently? If independent, what command(s) is hanging?

[melvyn burnard]

due to the fact that swagentd uses some DCE stuff (not sure what as I do not know DCE) you may have to add a workaround into your halt script to first stop swagentd with swagebtd -k, then try the umounts, then restart the swagentd, possibly with -r option.