[Linux-HA] how to start/stop a resource and wait until success/fail by PROGRAM?

[Andrew Beekhof]

On 10/22/07, dxj_600 <dxj_600 at 126.com> wrote:
> Hi,
>
> I use heartbeat-2.1.2, consider the following configurations:
> <cib>
> <configuration>
> <crm_config> .... </crm_config>
> <nodes>
>   <node uname="node0" id="node0"/>
> </nodes>
> <resources>
>   <primative id="rsc_xxx" class="ocf" type="Stateful" provider="heartbeat" is_managed="true">
>   </primative>
> </resources>
> <constraints>
>   <rsc_location id="location_xxx" rsc="rsc_xxx">
>     <rule id="location_xxx_rule" score="-infinity"/>
>   </rsc_location>
> </constraints>
> </configuration>
> </cib>
>
> I want to code a program which can start/stop a resource, and I want the function to be synchronous.
>   static BOOL cib_updated=FALSE;
>   void on_cib_diff()
>   {
>     cib_updated=TRUE;
>   }
>   int start_xxx()
>   {
>      cib_t *cib=cib_new();
>      cib->cmds->signon();
>      cib_conn->cmds->add_notify_callback(cib, T_CIB_DIFF_NOTIFY, on_cib_diff);
>
>      code with effect: cibadmin -D -o constraints -X '<rsc_location id="location_xxx"/>'
>      while(cib_updated==FALSE) sleep(1);
>      /* Now, cib has been updated, and will trigger the pengine/tengine to start the resource
>         MY QUESTION IS:
>         WHEN and HOW can i determine that the resource xxx has been started successfully???????
>      */
>
>      cib->cmds->signoff();
>      cib=NULL;
>   }

your program should connect to the cib and subscribe to change
notifications (the TE does this, look there for how).

once you make your update, look for an lrm_rsc_op for (your resource +
action=start + rc=0)

once you see that, then you know it has started