[Pacemaker] Migrate/run resource only if m/s resource is master/promoted on target node (follow-up)

[Martin Gombač]

Just in time, when i was about to implement STNITH on myself. :-)
Oh and thank you Lars for doing the bug report!

Meanwhile i noticed that with store 0 on ordering it didn't restart (i 
see you noticed that too in your bug report).

Now i just need to wait for repo i'm using to get new versions:
clusterlabs                                                      High 
Availability/Clustering server technologies 
(epel-5)                                                      enabled:    69

Regards,
M.

Lars Ellenberg wrote:
> On Sun, Jan 03, 2010 at 09:12:28PM -0700, hj lee wrote:
>   
>> On Sun, Jan 3, 2010 at 11:14 AM, Martin Gombač <martin at isg.si> wrote:
>>
>>     
>>> Hi,
>>>
>>> i removed colocation constraint per your suggestion.
>>> Please find other requested information attached.
>>> As logs show ibm1 restarted resource Hosting, when ibm2 was shut down,
>>> while leaving underlying drbd resource promoted. I wish the restart
>>> would never happen, as there is no need for it.
>>> Please let me know if i should provide any additional info.
>>> Thank you.
>>>
>>>       
>> I think the order constraint caused the restarting of XEN Hosting resource.
>> You can prove my assessment by removing order constraint. After removing
>> order constraint, try the same test. XEN Hosting resource should be OK.
>> Please see this document
>> http://www.clusterlabs.org/mediawiki/images/a/ae/Ordering_Explained_-_White.pdf.
>>
>>
>>  I think you have two choices.
>> 1. Try to use uni-directional ordering constraint by setting
>> symmetrical=false as explained in the above document. I never done this, so
>> not sure how to set it or it will work or not.
>> 2. Remove all drbd resources and start drbd by service. Configure drbd dual
>> primary in drbd.conf(I think you done this already) and enable drbd service
>> by chkconfig. So drbd will be started when the system is booting. Please
>> make sure drbd is started before heartbeat service.
>>     
>
>
> lf#2153 seems to be relevant, and has been fixed today.
> http://developerbugs.linux-foundation.org/show_bug.cgi?id=2153
>
>
>