[Linux-HA] heartbeat 2.0.8: unmanaged to managed

[greno at verizon.net]

>From: greno at verizon.net
>Somehow one of my group resources became unmanaged after I restarted heartbeat and now it refuses to become managed again.  I've tried issuing:
>crm_resource -S is_managed -v true -r myresource -t primitive
>
>and also creating an nvpair name="is_managed" value="true" in the cib.xml
>
>Neither of these made any difference. 
>
><group...
>  <primitive 1
>  <primitive 2 
>  <primitive 3
>  <primitive 4 
>  <primitive 5 ended up unmanaged
></group...
>
>The resource when it is unmanaged tries to start too early.  It needs to wait for a filesystem to be mounted by primitive 4 before it runs or it will fail.  When it was managed it would wait for the filesystem to be mounted but now it is not waiting.  How do you fix this problem?
>

I've tried deleting and recreating this resource and it still insists on being unmanaged.  This started when for some reason it was trying to run over on the standby node.  Rather weird since all it's resources were on the active node and I didn't tell it to run on he standby.  I thought all group resources were colocated but this one isn't.  An errant one for sure.  I've read all the manual recovery tips but still no luck.  What's weird is that I take it out of the cib altogether and then put it back later and it still behaves strangely.  It's like there's a memory of it that can't be erased.