Pivotal Knowledge Base

Follow

Pivotal Command Center not displaying server system stats

Problem

In Pivotal Command Center under the Dashboard & Cluster Analysis tabs, no Server Stats are being displayed.

Cause

There may be several reasons for the stats not being displayed:

1. There is a known issue where the Cluster uses full hostnames but the system_history table stores the short hostname. The PCC nmon process sets the FQDN for hostnames and becuase these are different the system stats are not updated.

2. The nmon process is not running on the servers.

Solution

1. For the known issue above, this has been resolved in PCC 2.1.1. If you are running versions older than PCC 2.1.1, as a workaround you need to do the following:

a. Run the following queries in the gphdmgr DB to check if the hostnames returned for both queries are the same:

gphdmgr=# select distinct host from clusterhost;
gphdmgr=# select distinct hostname from system_history;

b. If the hostnames returned for the above queries are different then you need to edit the /etc/sysconfig/network file on each host and change the hostname to the FQDN name:

[gpadmin@hdc10 ~]$ cat /etc/sysconfig/network

NETWORKING=yes
HOSTNAME=hdc10.gphd.local    

c. Once the network file has been changed to the FQDN, you then need to restart the network services & nmon service on each host:

service network restart
service nmon restart

If you are not able to restart the network services at this time you can set the hostname on the server manually with the below command:

hostname hdc10.gphd.local

2. If you identify that the PCC service nmon is not running on a host, you will need to start nmon on that host:

service nmon start

If after following the above and the system stats are still not being displayed, you can follow the below suggestions to help identify what the issue is:

  • make sure nmon is running on master and all slaves
  • inspect nmon logs on master and slaves for any errors: /var/log/nmon/
  • make sure both "hostname" and "hostname -f" on master/slaves return hdc<XX>.gphd.local
  • make sure nmon cluster configuration on master and slaves uses correct hostnames in /etc/nmon/conf/nmon-site.xml
  • make sure DB connection is setup properly on master in /etc/nmon/conf/nmon-site.xml
  • try manually restarting nmon master and slaves

Comments

Powered by Zendesk