service guard errors when starting cluster

[Helen Macdonald]

Service Guard
I have created a cluster file
volume groups currently activated
Done a cmapplyconf -C cmclconf.ascii
cmruncl -v
Failed to start cluster which is what I expected as volume groups are not cluster aware yet.
deactivated volume groups then tried to make volume groups cluster aware.
Following error displayed
"cmtaped kernel tuneable st_ats_enabled disabled for ATS.

I tried setting st_ats_enabled to 1 and now the following error is displayed

"cmtaped there is no ATS device on this cluster"

Set kernel parameter back to 0

Tried to start cluster again same error as before.

The cmcld daemon will not start thus I am unable to make my volume groups cluster aware.

Any Ideas

[Dietmar Konermann]

Hi, Helen!

From my experience both of your cmtaped messages are only warnings. The cluster should still form.

So I assume your problem is something complete different. Any other messages? What SG revision?

During the (susequently failing) formation... are there already cmcld processes runnning on all nodes?

Regards...
Dietmar.

[Sanjay_6]

Hi Helen,

Try the SG Cookbook from ITRC,

http://us-support2.external.hp.com/iv/data/documents/DE_SW_UX_swrec_EN_01_E/ServiceGuard.pdf

And the SG docs from hp docs.

http://docs.hp.com/hpux/ha/index.html#ServiceGuard

Try the doc "Managing SG"

Hope this helps.

Regds

[Jeff Schussele]

Hi Helen,

Not knowing what OS or MC/SG versions you're running, I would suggest that you look at the following 11.0 patches

PHSS_26928

http://support2.itrc.hp.com/service/patch/patchDetail.do?patchid=PHSS_26928&context=hpux:800:11:00

or it superseding patch
PHSS_27087

http://support2.itrc.hp.com/service/patch/patchDetail.do?patchid=PHSS_26928&context=hpux:800:11:00

These will update MC/SG to 11.13 & will default st_ats_enabled to NOT enabled & not force the check for ATS devices.

The problem w/11.0 & 11i & ATS is they default to st_ats_enable & enforce the ATS device check.

If not 11.0 you'll have to look for corresponding patches.

Anyway - it's worth a look.

HTH,
Jeff

[Helen Macdonald]

Many thanks.

I have now resolved the problem. There was a problem with connecting to the other node.