Symptoms
After POA upgrade, clustered CP becomes unavailable from cluster IP. /var/log/messages on load balancer contains following lines:
lvs[3342]: nanny died! shutting down lvs
lvs[3342]: shutting down virtual service cluster:1
Preceded by list of upgraded packages.
ps fax shows pulse working but its child is in zombie state:
3249 ? Ss 0:00 pulse
3342 ? Zs 0:00 \_ [lvsd] <defunct>
Cause
Pulse malfunction during packages' upgrade due to POA-83207.
Resolution
Restart pulse service:
# service pulse stop && service pulse start