Hello,
 
I am trying to setup a number of active/active resource groups where the app runs simultaneously on both nodes.
 
The start/stop/monitor script is identical on both machines, and the install directory is identical on both nodes.
 
I have named my test nodes as sid and nancy.
 
This is the chain of events when I startup the apps on sid:
 
1) monitor action  takes place on primitve for group named_grp_sid
2) monitor action takes place on primitve for group named_grp_nancy
3) stop action takes place on primitve for group named_grp_sid
4) stop action takes place on primitve for group named_grp_nancy
5)start action takes place on primitve for group named_grp_sid
   ** all the above takes place on sid **
 
When I bring things up on nancy i get:
 
1) monitor action  takes place on primitve for group named_grp_sid
2) monitor action takes place on primitve for group named_grp_nancy
3) stop action takes place on primitve for group named_grp_sid  *** it seems to tell sid to stop this ***
4) stop action takes place on primitve for group named_grp_nancy
5)start action takes place on primitve for group named_grp_nancy
6) start action takes place on primitive for group named_grp_sid *** this action takes place on sid***
 
 
 
My questions now are:
 
1) how to I make it so that no actions for groups designated for node  nancy take place on sid and vice versa?
2) How do I keep nancy from telling sid to stop and restart the group associated with sid?
 
 
I am sure its something that I am missing in the way constraints work. If anyone can shed light on how I can achieve that, it would be greatly appreciated.
 
 
Below is a copy of my sample cib.xml file.
 
 
Many Thanks
 
Bob
 
 
 <cib>
   <configuration>
     <crm_config>
     </crm_config>
     <nodes/>
     <resources>
       <group id="named_grp_sid">
         <primitive id="named_res_sid" class="heartbeat" type="named">
           <operations>
             <op id="named_res_sid_monitor" name="monitor" interval="30s" timeout="10s">
             </op>
           </operations>
         </primitive>
       </group>
       <group id="named_grp_nancy">
         <primitive id="named_res_nancy" class="heartbeat" type="named">
           <operations>
             <op id="named_res_nancy_monitor" name="monitor" interval="30s" timeout="10s">
             </op>
           </operations>
         </primitive>
       </group>
     </resources>
     <constraints>
       <rsc_location id="run_named_grp_sid" rsc="named_res_sid">
         <rule id="pref_run_named_grp_sid_runonlyon" score="INFINITY">
           <_expression_ id="pref_run_named_grp_sidrunonlyon1" attribute="#uname" operation="eq" value="sid"/>
         </rule>
         <rule id="pref_run_named_grp_sid_donotrunon" score="-INFINITY">
           <_expression_ id="pref_run_named_grp_sid_donotrunon1" attribute="#uname" operation="ne" value="sid"/>
         </rule>
       </rsc_location>
       <rsc_location id="run_named_grp_nancy" rsc="named_res_nancy">
         <rule id="pref_run_named_grp_nancy_runonlyon" score="INFINITY">
           <_expression_ id="pref_run_named_grp_nancyrunonlyon1" attribute="#uname" operation="eq" value="nancy"/>
         </rule>
         <rule id="pref_run_named_grp_nancy_donotrunon" score="-INFINITY">
           <_expression_ id="pref_run_named_grp_nancy_donotrunon1" attribute="#uname" operation="ne" value="nancy"/>
         </rule>
       </rsc_location>
     </constraints>
   </configuration>
   <status/>
 </cib>
_______________________________________________________
Linux-HA-Dev: [email protected]
http://lists.linux-ha.org/mailman/listinfo/linux-ha-dev
Home Page: http://linux-ha.org/

Reply via email to