[Linux-HA] duplicated node

[Andrew Beekhof]

On Nov 27, 2007, at 11:54 AM, Frank wrote:

> Hi,
> we have a cluster with 2 nodes with heartbeat 2.1.2 and Red Hat Linux
>
> after setting one node to standby, stopping heartbeat on it,  
> uninstalling heatbeat, removing all heartbeat directories and
> installing heartbeat again, configuration detect two nodes with the  
> same name but different id:
>
> Node: polonio (296f7c6c-9698-40f1-af8e-5d4917318701): OFFLINE
> Node: einstenio (30b5f04e-b6bd-4c89-ab79-629590a0e7ae): online
> Node: polonio (40eaf81e-ac04-4099-bea7-99d84ea10d0d): online
>
> node section in cib.xml is
>
> <nodes>
>      <node uname="polonio" type="normal" id="296f7c6c-9698-40f1- 
> af8e-5d4917318701">
>        <instance_attributes id="nodes-296f7c6c-9698-40f1- 
> af8e-5d4917318701">
>          <attributes>
>            <nvpair name="standby" id="standby-296f7c6c-9698-40f1- 
> af8e-5d4917318701" value="off"/>
>          </attributes>
>        </instance_attributes>
>      </node>
>      <node id="30b5f04e-b6bd-4c89-ab79-629590a0e7ae"  
> uname="einstenio" type="normal"/>
>      <node id="40eaf81e-ac04-4099-bea7-99d84ea10d0d" uname="polonio"  
> type="normal"/>
>    </nodes>
>
> Is there a safe way to remove unused node?

the easiest way is to shutdown the entire cluster and remove the  
hostcache file from each one

then do:

cibadmin --cib_delete --obj_type nodes --crm_xml ‘<node  
id=”296f7c6c-9698-40f1-af8e-5d4917318701”/>’