Oracle Communications Policy Management
 

PLA-Support>Sun Systems>CommsGBU>Broadband Network Solutions>SN-SND: Tekelec Policy
 

Only on Cable systems with Managed Agents, and backplane replication connection, comcol is trying to resolve the un-routable backplane address of a different node's servers.     this is because a qpfeature has been turned off.  
commands to turn the feature on are the solution.

In this Document

Symptoms

Changes

Cause

Solution

References

Created from <SR 3-13054115971>

Applies to:

Symptoms

the Managed Agents are generating alarms:

     from RecentAlarms    # 'ra.stat -w'

part=0 srcNode=00000.000 eventNumber=Hostname Lookup Failed seqNum=18948494 timeStamp=07/21/2016 10:36:04.125 createTime=03/08/2016 20:56:43.476 firstSetTime=07/08/2016 20:56:43.476 lastSetTime=07/26/2016 10:36:04.212 severity=-* trapNumber=0 task=re.portmap instance=ma02-cl02-BP haScore=Normal autoClearSecs=300 errInfo=
GN_NOTFOUND/WRN getaddrinfo() lookup failed: Temporary failure in name resolution [PortMapServer.cxx:1249]

Server ma02-cl01   is trying to resolve the backplane address of a server in a different cluster. 

Changes

Cause

the /etc/hosts file of a cable configured server will have the non-routable backplane addresses of itself and its peer.
  
# BEGIN_QP_GENERATED_FIELDS
169.254.88.1  ma01-cl01-BP
169.254.88.2  ma02-cl01-BP
172.16.28.75  cmp01-cl01
172.16.28.76  cmp02-cl01

172.16.28.78  ma01-cl01
172.16.28.79  ma02-cl01
172.16.28.84  ma01-cl02
172.16.28.85  ma02-cl02

but in this case the Managed Agent is trying to find the BP address of a server is a different cluster.
----------------------
see bug 23148947 and bug 21936917
 

Solution

Correct steps to resolve this issue:

1. Run following command on the Active CMP in Primary CMP cluster as root:

  # qpfeature --set cmRepArb on

  Check comcol table LongParam using following command:

  # iqt -pE LongParam | grep "cm.rep.arbEnabled"

The output should be as followings:

[root@hostname ~]# iqt -pE LongParam | grep "cm.rep.arbEnabled"
allowPgmChg=Yes name=cm.rep.arbEnabled value=0 comment=Audit Arbitrator is enabled (value is 1) by default

2. Check comcol table LongParam on all other servers except active CMP using following command.
  # iqt -pE LongParam | grep "cm.rep.arbEnabled"
The output should be as followings:
[root@hostname admusr]# iqt -pE LongParam | grep "cm.rep.arbEnabled"
allowPgmChg=Yes name=cm.rep.arbEnabled value=0 comment=Audit Arbitrator is enabled (value is 1) by default

when the above is completed, the MA's will have to be restarted. this is safe to do as long as there is no scheduled task that is active.

the simplest way to restart is to reboot the servers, but the sequence of commands below will restart the database processes that need to be restarted to affect this change, without rebooting:

 # service qp_procmgr stop
 # prod.stop -i
 # prod.start -i
 # service qp_procmgr start

---------------------
  the command 'ra.stat-w' is use to show the active alarms on the current server. running it on the MA, which is the source of the 'unable to resolve hostname' alarm will show when it has cleared.
  running ra.stat -w on the CMP will show the alarms that are generated by the CMP itself, but not the alarms from all other servers which have been merged up to the CMP.
 

References

This solution has no attachment