monit-general
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Config working on CentOS 5 stopped working on CentOS 6


From: Stas Oskin
Subject: Re: Config working on CentOS 5 stopped working on CentOS 6
Date: Mon, 21 Dec 2015 17:45:51 +0200

Strange, as on CentOS 5 it was enough to restart monit only.

Anyhow, thanks for helping to finally pinpoint the issue.

On Dec 21, 2015 4:46 PM, "Martin Pala" <address@hidden> wrote:
The monitoring state persistency is part of monit for very long time - i think most probably even monit <= 3.x worked like this.

As mentioned in the previous email, you can remove the timeout statement - if it won't be possible to recover the service and restart action will be called each cycle, there will be no limit on number of restarts, but you will be notified and can fix it manually (which is required in such case anyway).

Regards,
Martin


On 21 Dec 2015, at 12:09, Stas Oskin <address@hidden> wrote:

Alternatively, if no such mode is possible, are there any issues to just having the monit running in endless loop trying to recover the service?

On Mon, Dec 21, 2015 at 1:08 PM, Stas Oskin <address@hidden> wrote:
Hi,

Thanks for the clarification, it seems this was exactly the change between 4 and 5 that caused us the confusion.

Is there a way to have a previous mode of operation, where the monit will reset the state by restarting the monit itself (and not the server, as per your suggestion)?

Thanks.

On Wed, Dec 16, 2015 at 6:11 PM, Martin Pala <address@hidden> wrote:
Hi,

Monit will disable the process monitoring on excessive restart failures due to "if 5 restart within 5 cycles then timeout" statement in your configuration (the "timeout" action an alias for "unmonitor" and we switched the documentation in the past to "unmonitor" as it is more clear: https://mmonit.com/monit/documentation/monit.html#SERVICE-RESTART-LIMIT)

The monitoring state is persistent - the "timed out" service has usually some hard error which requires manual intervention (timeout statement prevents endless restart loop). When the problem is resolved, the monitoring needs to be enabled manually.

If you want to drop the state for example after reboot, place the statefile to tmpfs filesystem (you can use "set statefile <path>" statement to customize state file placement).

Regards,
Martin



On 16 Dec 2015, at 14:20, Stas Oskin <address@hidden> wrote:

Hi,

After some more digging, it occurred to me that it might that monit just stops monitoring the process after it unable to restart it.

So on monit 4.x it appears this state was cleared when just restarting monit, while on 5.x it seems you need actually to mark the check as active manually via the monit command.

Is this correct?

On Fri, Nov 20, 2015 at 5:37 PM, Martin Pala <address@hidden> wrote:
Hi,

please run monit in debug mode and send output:

monit -vI

Regards,
Martin



On 19 Nov 2015, at 20:39, Stas Oskin <address@hidden> wrote:

Hi,

The monit log shows only the general start-up information.

There is no messages about the processes going offline, it's like monit does not use the pid file to find the process anymore.

When I use HTTP port probing though it works just fine. Any idea what could it be?

Regards.

On Sun, Nov 15, 2015 at 9:16 PM, Martin Pala <address@hidden> wrote:
Hi,

please can you provide more details about the problem? (error messages and/or monit log).

Note that monit 5.9 includes fix for program execution for CentOS6/RHEL6, we recommend upgrade to latest monit version (5.15), you can get it here: https://mmonit.com/monit/#download. You can build rpm directly from the source code release: rpmbuild -tb monit-5.15.tar.gz. I think RHEL uses custom configuration file, official monit looks for /etc/monitrc, so you may need to rename the configuration file or create a link after upgrade.

Regards,
Martin


> On 14 Nov 2015, at 16:51, Stas Oskin <address@hidden> wrote:
>
> Hi,
>
> Monit has reliably served us through the years, and we are very happy of it.
>
> Unfortunately during scheduled migration to CentOS 6 due CentOS 5 EOL, we discovered it stopped monitoring the services pid files. HTTP monitoring works fine.
>
> The CentOS 6 version is:
> monit-5.1.1-4.el6.x86_64
>
> CentOS 5 version is:
> monit-4.10.1-8.el5
>
> An example config that not working anymore (but accepted by monit when starting):
> check process XXXX with pidfile /XXXX/pid/XXXXX.pid
>     start program  "/etc/init.d/xxxxx restart"
>     stop program  "/etc/init.d/xxxxxx stop"
>     if mem usage > 85% then restart
>     if 5 restarts within 5 cycles then timeout
>
> I guess something changed in configuration jump from 4 to 5, will appreciate any advice.
>
> Thanks!
> --
> To unsubscribe:
> https://lists.nongnu.org/mailman/listinfo/monit-general


--
To unsubscribe:
https://lists.nongnu.org/mailman/listinfo/monit-general

--
To unsubscribe:
https://lists.nongnu.org/mailman/listinfo/monit-general


--
To unsubscribe:
https://lists.nongnu.org/mailman/listinfo/monit-general

--
To unsubscribe:
https://lists.nongnu.org/mailman/listinfo/monit-general


--
To unsubscribe:
https://lists.nongnu.org/mailman/listinfo/monit-general


--
To unsubscribe:
https://lists.nongnu.org/mailman/listinfo/monit-general


--
To unsubscribe:
https://lists.nongnu.org/mailman/listinfo/monit-general

reply via email to

[Prev in Thread] Current Thread [Next in Thread]