[Linux-HA] cluster fails over, but does not reboot the failed
johan.hoeke at uvt.nl
Fri Dec 21 05:28:14 MST 2007
Dejan Muhamedagic wrote:
>> Alas, I've added on_fail=fence everywhere but still the services
>> failover but the failed node doesn't reboot.
>> My newby logic thinks it's got something to do with the fact that the
>> resources don't actually fail, but that the switchover occurs because of
>> the constraint that has to do with the (hba)pingd. The constraint is
>> criptically called G_dloprd:connected in the attached cibadmin -Ql listings.
> Your logic is right. Sorry, didn't look closely enough.
>> I used the section from
>> "Quickstart - Only Run my_resource on Nodes with Access to at Least One
>> Ping Node" as as a template for my pingd constraint.
>> Any suggestions?
> None. The purpose of fencing is to establish, in case it is
> necessary, that a node released resources which were running on
> it. Now, I do understand your wish to reboot in case of ping
> failing, but there's no provision for that. You can file an
> enhancement bugzilla for that.
OK, thanks very much for taking a look! Your expert opinion is very much
I'm glad that i'm starting to get this stuff a little bit.
Just scratching the surface, but none the less...
Happy holidays to all!
Johan Hoeke <johan.hoeke at uvt.nl> – UNIX system administration
Universiteit van Tilburg – LIS • Postbus 90153, 5000 LE
Bezoekadres : Warandelaan 2 • 013 466 3374 • http://www.uvt.nl
-------------- next part --------------
A non-text attachment was scrubbed...
Size: 252 bytes
Desc: OpenPGP digital signature
Url : http://lists.community.tummy.com/pipermail/linux-ha/attachments/20071221/e10b7501/signature.pgp
More information about the Linux-HA