[Linux-cluster] Service Failovers are slow
mockey.chen at nsn.com
Fri Feb 13 01:15:56 UTC 2009
ext Burton Simonds wrote:
> We are testing our RHCS with several services, for example httpd. In
> the httpd service there is a script resource and an ip resource. To
> test failover, we do a kill -9 <httpd pid>, and wait. Then we wait
> some more. It takes about 30 seconds for the service to fail over to
> the other node.
> it is a 2 node cluster running on RHEL 5.2
> Is there a way to adjust the health checks of the services?
Sure, you can add child-tag 'action' to change the interval, for example:
<script file="/opt/cm/bin/ucs_service" name="ucs_service">
<action name="status" interval="10" depth="*"/>
More information about the Linux-cluster