Today I was installing a web cluster with the usual heartbeat/drbd stuff. The customer also needs zope because a part of the web site requires it, so I went and built a fresh rpm of it (link to zope rpm).
I then added zope to the haresources file, only to find out that heartbeat will cowardly refuse to start it!
When heartbeat starts to acquire resources it checks the status first by running resource status.
The problem is that the zope rc script prints 'not running' when it cannot find any zope instance running and heartbeat resource manager greps the zope status output for [Rr]unning, to decide whether it must start zope or it is already started for whatever reason.
The solution should be already in heartbeat cvs for some time now and was proposed by Lars Ellemberg of the drbd project. See this link for details:
- ► 2015 (11)
- ► 2013 (10)
- ► 2012 (35)
- ► 2011 (18)
- ▼ March (3)