Suppress Monit message 'Monit instance changed' on stop/start

Question

I have monit installed on some user laptops/desktop so it can make sure a couple processes are always running/get restarted as needed. But since the devices are regularly shutdown/restarted I frequently receive a message like below which I do not want or care about, since they are telling me that the system was restarted. I want to suppress the messages that seem to automatically happen when monit starts/stops.

Subject: monit alert -- Monit instance changed

Service     - system_slaveone
Event       - Monit instance changed
Action      - start
Date        - Thu, 18 Apr 2013 07:53:51
Host        - slaveone.example.local (slaveone)
Description - Monit started.

But once the system is up, I do want to get some alerts about a couple services/resources that I do want to be notified about, so simply disabling email isn't want I want to do.

How can I suppress the message on startup/restart without completely disabling email?

Answer 1

Oh, sure...

To suppress Monit alerts for a particular check, you need to use the noalert directive.

For instance, I can have a check for the cron daemon, which restarts weekly. Maybe I don't want that email in my inbox every Sunday...

check process cron
    noalert [email protected]
    with pidfile "/var/run/crond.pid"
    start program = "/sbin/service crond start"
    stop program = "/sbin/service crond stop"

This works for any Monit check, so to handle the instance alarms for the system, you can suppress by adding the noalert under the check system localhost entry:

  check system localhost
    noalert [email protected]

Try it. This still leaves any real services/daemons under Monit protection, but reduces the chatter. Great for misbehaving applications...

check process nslcd
        with pidfile "/var/run/nslcd/nslcd.pid" every 2 cycles
        noalert ewwhite@bra**ers.com
        start program = "/sbin/service nslcd start"
        stop program = "/sbin/service nslcd stop"
        if 10 restarts within 11 cycles then timeout
        if cpu usage > 95% for 11 cycles then restart
        if totalmemory > 128 MB then restart

Answer 2

To particularly filter out the e-mails when monit starts/stops, you can use the instance event filter:

set alert [email protected] not on { instance }

As in the documentation:

Event:     | Failure state:            | Success state:              
---------------------------------------------------------------------
...
INSTANCE   | "Monit instance changed"  | "Monit instance changed not"
...