[Linux-cluster] Oracle 10G resource agent - status polling

[Lon Hohberger]

On Tue, 2008-06-24 at 17:35 +0000, Finnur Örn Guðmundsson - TM Software
wrote:
> Hi,
> 
> (Sorry for the top posting, i blame my email client!)
> 
> The service is started yes.
> 
> I was going for the cluster.conf way, IE, setting the status check to 1 minute.
> 
> I can see it should not be a problem with a script resource but since i use the Oracle Resource Agent it kinda runs around in my head.
> 
> Here is what i tried:
> 
>                 <service autostart="1" exclusive="0" name="oracle1" recovery="relocate">
>                         <ip __independent_subtree="1" ref="10.x.x.x">
>                                 <fs ref="ora1-data"/>
>                                 <fs ref="ora1-archlogs"/>
>                                 <oracledb home="/u01/app/oracle" name="oracle1" type="10g" user="oracle"/>
>                                 <action name="status" depth="*" interval="1m"/>
>                         </ip>
>                 </service>

I would do it this way (the __independent_subtree option in this context
doesn't do anything useful):

<service autostart="1" exclusive="0" name="oracle1" recovery="relocate">
  <ip ref="10.x.x.x" />
  <fs ref="ora1-data" />
  <fs ref="ora1-archlogs" />
  <oracledb home="/u01/app/oracle" name="oracle1" type="10g" user="oracle">
    <action name="status" depth="*" interval="1m"/>
  </oracledb>
</service>

(action looks like a resource here, but is handled "special-case" like
by rgmanager)

-- Lon