Managing Greenplum Workload Manager Services
Greenplum Workload Manager installs and runs four services on all segment hosts in the Greenplum cluster:
-
agent
-
cfgmon
-
rabbitmq
-
rulesengine
The services can be managed using the INSTALLDIR/gp-wlm/bin/svc-mgr.sh
command. The command has the following syntax:
INSTALLDIR/gp-wlm/bin/svc-mgr.sh \
--service=SVCNAME \
--action=ACTION
SVCNAME
may be agent
, cfgmon
, rabbitmq
, rulesengine
, or all
. If SVCNAME
specifies an individual service, only that service is modified. Specify all
to manipulate all services.
The ACTION
parameter affects only the local system, unless it is prefixed with cluster-
, in which case it runs on all hosts in the cluster. The actions are:
-
start / cluster-start
– Start any of the Workload Manager services that are not running. -
stop / cluster-stop
– Stop any Workload Manager services that are running. -
status / cluster-status
– Determine if the services are running. -
restart / cluster-restart
– Restart the Workload Manager services. -
enable / cluster-enable
– Enable and start Workload Manager services. -
disable / cluster-disable
– Stop and disable Workload Manager services.
If you source the INSTALLDIR/gp-wlm/gp-wlm_path.sh
file in your shell, the Workload Manager scripts are in your path. Otherwise, you must provide the full path to the utility in the gp-wlm/bin
directory.
When a service is stopped, it will not be restarted until the start
action is invoked, or the local machine reboots, whichever comes first.
When a service is disabled, it will not be restarted until the enable
action is invoked. This is persistent across reboot.
The following example checks the status of all Workload Manager services on the local host:
[gpadmin@mdw ~]$ svc-mgr.sh --service=all --action=status
RabbitMQ is running out of the current installation. (PID=22541)
agent (pid 22732) is running...
cfgmon (pid 22858) is running...
rulesengine (pid 22921) is running...
Checking the Health of Greenplum Workload Manager Services
At any time, the health of Greenplum Workload Manager services can be verified across the cluster by invoking the cluster-health-check
utility. This tool confirms that all services are running across the cluster, and that messages are being received from each machine in the cluster. Following is the syntax for cluster-health-check
:
INSTALLDIR/gpwlm/bin/cluster-health-check --symlink=/absolute/path/to/installation/symlink
[--max-concurrency=N]
[--max-cluster-checks=N]
[--help]
Options:
-c
or --max-concurrency
The max-concurrency
option specifies the number of hosts to check at once. The default is a computed value based on the number of hosts in the cluster: 20 if there are fewer than 100 hosts, 50 if there are 100 to 199 hosts, and 75 if there are 200 or more hosts.
-m
or --max-cluster-checks
The number of times to check for a healthy cluster. The default is 1.
-s
or --symlink
The absolute path to the gp-wlm
directory linked to the installed Workload Manager release. Required.
-h
or --help
Display command usage information and exit.
If the command reports an error communicating with one or more services, the cluster may be restarted with this command:
INSTALLDIR/gp-wlm/bin/svc-mgr.sh --action=cluster-restart --service=all
This command stops and then restarts each of the Workload Manager services on each segment host.