| Another option to consider is a standard SNMP tool like LibreNMS. We use Monit and LibreNMS to monitor our infrastructure - both Linux and Windows servers - with great success. LibreNMs has traditional alerting (email, Slack, Pager Duty, etc), and you can create custom alert rules if needed (host has been down for x days).
Just food for thought…
-Ron
On Nov 23, 2023, at 5:23 AM, Graham Smith <gsmitheidw@gmail.com> wrote:
Thanks to you both - unfortunately the clients are all Windows desktops.
I thought about WSL or even quemu etc but it sounds very messy at scale and
I'd need to ensure that software was reliable too.
Our student users are all developers who need Virtualbox, certainly WSL2 would likely conflict. It's a shame there's no native Windows port of Monit. Maybe there
could be a Cygwin solution.
Timetabling complexity means sometimes small groups may be in larger capacity rooms
but then if there's an exam, student turnout is close to full and nearly everything is needed.
One or two machines down is something we can absorb, but I'm trying to avoid surprises of multiple outages at the same time.
I think a central Monit instance with a higher than 64 cycle limit would work for me.
Maybe I can change it in the source and recompile.. although that may have other implications!
Many thanks,
Graham
Worst case, use WSL on windows to run monit.
-----Original Message-----
From: monit-general-bounces+rory=trs80.net@nongnu.org <monit-general-bounces+rory=trs80.net@nongnu.org> On Behalf Of Lutz Mader
Sent: Thursday, November 23, 2023 4:37 AM
To: This is the general mailing list for monit <monit-general@nongnu.org>
Subject: Re: Monitoring Desktops
Hello Graham,
and MacOS systems are supported also.
Lutz
is M/Monit to monitor all systems centrally.
Am 23.11.23 um 09:28 schrieb Lutz Mader:
> Hello Graham,
> as long as the desktop systems are Linux systems, the simple answer
> here is M/Monit to monitor all systems centrally.
>
>> Any ideas or strategies?
>
> On the other hand, I use a similar approach to yours to monitor
> applications on other systems (I check the availability of a port) to
> start central applications. But I need an alert after 20 minutes, or
> wait only 50 minutes for the other systems.
>
>> I'm finding the 64 cycle limit a bit of a struggle. What I'm trying
>> to achieve is to get an alert if a machine hasn't been seen
>> responding to pings for about 3 to 4 days.
>
> Nice to see that you have enough systems to go several days without one.
>
> Sorry, I use a similar approach,
> Lutz
>
>
> Am 22.11.23 um 18:50 schrieb Graham Smith:
>> Hi,
>>
>> Has anybody successfully used Monit to alert on desktop outages? We
>> have a number of desktops in student labs (University) on a campus.
>> Although used heavily, we could legitimately find one pc in the
>> corner of a room which simply has been unplugged for a day or two but
>> is actually fully functional.
>>
>> I'm finding the 64 cycle limit a bit of a struggle. What I'm trying
>> to achieve is to get an alert if a machine hasn't been seen
>> responding to pings for about 3 to 4 days.
>> After that, it's probably a machine that warrants some investigation
>> for a potential failure.
>>
>> Are there any strategies I could consider?
>>
>> Currently I'm just using lines in a file in the monit conf.d similar to:
>>
>> check host Room4-36 with address 10.10.4.36
>>
>> if failed icmp type echo count 1 with timeout 1 second for 64
>> cycles then alert
>> else if succeeded for 64 cycles then alert
>>
>> The main monitrc cycle I'd want to check every 5 to 10 mins or so...
>> but because of the cycles 64 limit I'm having to increase that to
>> more than I'd like.
>> If I increase it to say an hour, a machine could be turned on the off
>> again within that window and I'd not be aware.
>>
>> Any ideas or strategies?
>>
>> Kind regards,
>>
>> Graham Smith
>>
>
|