[Hcoop-discuss] Re: [Hcoop-announce] /var space problem fixed

[Chris Clearwater]

Chris Clearwater wrote:

> ntk at hcoop.net wrote:
>
>>> ntk at hcoop.net writes:
>>> I was going to recommend the logcheck package, but it's already
>>> installed.
>>>   
>>
>>
>> Looks useful, but not sufficient for this purpose.
>> When /var is actually out of space, it's too late--mail is already being
>> dropped.  We want an alert before this point (and when /var is 
>> completely
>> full, the logger may not be able to write to logfiles anyway, since they
>> are under /var/log).  Also if daemons crash catastrophically they may 
>> not
>> log anything, and in the worst case where fyodor goes down altogether
>> clearly logcheck won't do any good.  So what I propose is:
>>
>> 1. Script on fyodor, to monitor disk space and that the proper 
>> daemons are
>> running.
>> 2. Script on abu, to make sure that fyodor is up, and that mail and web
>> daemons are actually serving mail and webpages properly (because the
>> daemons may be running on fyodor, but unresponsive, or there may be
>> network or DNS problems).
>>
>> -ntk
>>  
>>
> I think http://www.nagios.org/ (formerly netsaint) does all of this 
> out of the box. There is even a debian package for it.

I would also like to restate that this solves a different set of 
problems than having a backup MX.