[Top][All Lists]

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

Re: monit 4.3, what event for generic protocol failure?

From: Tom O'Brien
Subject: Re: monit 4.3, what event for generic protocol failure?
Date: 18 May 2004 10:01:53 -0400

I think your second response captured the problem. I'll more fully
explain the situation:

I wrote a test server that does a listen() on port 1033. The server
NEVER calls accept()--I want to see whether monit would detect the case
where the socket connected successfully (in the ESTABLISHED backlog),
but never responded to a protocol string.

The string of events that I saw was this:
1) Start monit with configuration shown
2) Connection attempt was made--connection failed, alert sent to tobrien
3) Server started on port 1033
4) Connection attempt was made, succeeded, but protocol test failed, no
alert sent to tobrien
5) Modify configuration, send HUP to monit
6) Connection attempt succeeded, but protocol test failed, alert sent to

So, as monit saw a previous connection failure, it didn't send another
alert in Step 4.

thanks for the explaination

On Mon, 2004-05-17 at 22:22, Jan-Henrik Haukeland wrote:
> > "Tom O'Brien" <address@hidden> writes:
> >
> >> What is the event definition that triggers the alert?
> Just to answer this one also, since that was your actual question :)
> The event that triggers an alert if a connection fails or if a
> protocol test fails is "connection". So when you correctly use:
>  alert address@hidden { timeout connection checksum nonexist }
>   or 
>  alert address@hidden
> you should get an alert when the protocol test fails, but remember
> that monit only sends 1 (one) failed alert for each service and only
> send a new failed alert if a service gets up and *then* fails
> again. So when you did not get an alert it may be because the
> connection to port 1033 failed earlier?
> If this is not the case and you didn't not get an alert the first time
> your send/expect test failed it's a bug in monit and we need to fix
> it. However, doing a fast test I can't really reproduce the problem on
> my machine.

reply via email to

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