gpsmon agent is not running in one of the data nodes. How to start it without restarting DB?

[Gijo George]

gpmmon and gpsmon are running in mater as below.

gpadmin 548407 548398 0 Sep29 ?       00:10:01 /usr/local/GP-4.3.11.3/bin/gpmmon -D /data/master/gpseg-1/gpperfmon/conf/gpperfmon.conf -p 5432

gpadmin 565283     1 0 Sep29 ?       00:14:41 /usr/local/greenplum-db/./bin/gpsmon -m 0 -i -t 150 -l /data/master/gpseg-1/gpperfmon/logs -v 0 8888​

 

But gpsmon agent is not running in one of the segment node.

[gpadmin@sdw18 ~]$ ps -ef | grep gpsmon

gpadmin 310563 309821 0 07:17 pts/0   00:00:00 grep gpsmon

[gpadmin@sdw18 ~]$

 

History table in gpperfmon db are not having any information of the node sdw18. How we can restrta gpmmon agent in master without restarting DB

[Brendan Stephens]

The gpmon (the master process) will restart a missing worker (gpmon).

You might want to check /var/log/messages or dmesg to see if the worker has failed for some reason.

 

You may also be able to check the gpperfmon logs for the status of the worker on that host.

There is a gpperfmon directory for every segment on that host and a gpsmon log on the master.

[Faisal Ali]

Hello Gijo,

 

Unfortunately gpsmon doesn't have the level of flexibility as gpmon process. if the gpsmon is down only way to get it back up is via restart of the database.

 

You can have a look at the gpsmon architecture for answer to why we cant start separately.

 

https://github.com/greenplum-db/gpdb/wiki/Gpperfmon-Overview#gpsmon

 

Thanks

Faisal