PDA

View Full Version : On 2-node Cluster: Resource restart when other node reboot



filipo33
25-Jun-2012, 15:41
Hello all,

Example of my problem:I build a 2-node cluster (web1 and web2), for a web server ( DRBD + Virtual IP + APACHE service )
Here web1 is DC, active/master node and start all resources.


Last updated: Mon Jun 25 16:25:13 2012
Last change: Thu Jun 21 18:03:33 2012 by root via crm_attribute on web1
Stack: openais
Current DC: web1 - partition with quorum
Version: 1.1.6-xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
2 Nodes configured, 2 expected votes
5 Resources configured.
============

Online: [ web2 web1 ]

fs (ocf::heartbeat:Filesystem): Started web1
Resource Group: webname
virtual-ip (ocf::heartbeat:IPaddr2): Started web1
lampp (lsb:my_service): Started web1
Master/Slave Set: ms_drbd [drbd]
Masters: [ web1 ]
Slaves: [ web2 ]


I have noticed, that when I stop web2 (node which is passive), all resources on web1 are OK.
When I re-start the web2 server (or cluster openais service), all resources on web1 are restarted ( so no web access during 1sec)

This "ressource restart" is unnecessary ....can I prevent it ? It's not for 1sec, but it's seems weird for me.

Thanks for future answers.
Filipo

jmozdzen
27-Jun-2012, 22:11
Hi Filipo,

are the "restarts" probably caused by resource moves from web1 to web2, once web2 goes online?

Regards,
Jens

filipo33
28-Jun-2012, 08:54
Hello,

No, all resources stays on web1 (preferred node).
I can check with "crm_mon", before and after: All is on web1 --> web2 never used.

I group my "virtual-ip" and my "service_lampp", and made web1 my favorite.


Thanks
Filipo

filipo33
05-Jul-2012, 14:33
[SOLVED]

It's because i started my resources with "primitive foo lsb:bar" and not "primitive foo ocf:heartbeat:bar".



Thanks
Filipo

jmozdzen
05-Jul-2012, 14:44
Hi Filipo,

thanks for the feedback - I'd never have guessed that cause!

Regards,
Jens