mail statistic/graph

[ronni]

I think it would be a great thing if something like "mailgraph" was integrated in ispCP. I have looked on http://mailgraph.schweikert.ch/ and I will try to install it on my production server there is running ispCP RC2.

It would be a nice tool for administrators to monitor the load on the mail server I think. Please give me your opinion.

Best Regards
Ronni Poulsen

[ephigenie]

yeah that looks really nice.

It's just the question if those things shouldn't be done better by apps like munin, zabbix, nagios and so on.
Because that stats aren't domain-specific so its more a "how is the health/ busy state of the whole machine atm" thing.

I'm not quite sure if this has to be in there. Although its nice to know about it.
Perhaps we can find a way to integrate munin graphs somehow ?

[joximu]

I know this tool, it's nice - as Ephigenie said :-)

But we have to think about what is he main task for ispcp (omega). If we want a full service thing like plesk then we should integrate it or we should have the focus on a cool and flexible control panel with the possibility to integrate such tools (withour integrate them into ispCP). I vote for the second...
What we can do: have a link to mailgraph at the website somewhere... or when the project has grown - in the docu as a hint...

just my 2 cents...
/Joximu

[BeNe]

I already started a Thread about it --> http://www.isp-control.net/forum/showthread.php?tid=410

The good one - it is damn easy to install under Debian.
I wanted to test how it works. But i never get a answer.

It would be nice if we can include some that can more than only one thing.
Like munin, there is also a HowTo in the Forum from me.

The difference to zabbix or nagios is that they are monitoring and
not only create stats like munin or cacti. Is this needed ?

[ephigenie]

Yeah i think monitoring makes only sense if there's also a response about services running or not or running above some user-definable limits...

that maybe is a great option for the future of ispCP as well.
Because if we support multi-server solutions as well, than keeping a lot of servers in view ain't that easy as to keep one server monitored.

So i think admins will be glad about just defining limits for their servers and beeing notified if those are reached or if something else (service/server) not responding happens.

[BeNe]

For sure make monitoring sense, but not only on one host.
In some future version with multi-server is this option no question.

But when the monitoring and the services that should be monitored are on the same machine - it make this no sense for me.
Let the mailservice like Postfix go down on the host, you never will get a mail about this Problem

[ephigenie]

yeah but there're other ways to submit an alert if its urgent (like sms)
usually these got submitted via calling an url (can be done via bash-script or similar)
so as long as there's a network available you can send alerts.

[BeNe]

Yeah of course - there are many ways.
But in my eyes it make only _really_ sense when a Monitoringsystem is extern from the system that is monitored.

Ok - lets test it, why not!?
I mean that Nagios has something like a self helping system that can start or reboot the service after it goes down.

[soringo]

I vote for munin...I'm useing for my production server and I'm pleased with it....munin is more for admins and not for end users (clients). I guess statistics from CP is enough for clients.

Hac!..my opinion.