[Top][All Lists]

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

Re: Different startup time from polling time?

From: Martin Pala
Subject: Re: Different startup time from polling time?
Date: Thu, 26 Aug 2004 13:37:37 +0200

I think the discussion pointed by link bellow explains the problem.

It is not good solution to delay testing generaly on monit startup, because
this is just special workaround which is effective only on system startup
and affects all services. It is not good solution either to use common delay
for all services on their restart initiated by monit - some services may
start fast, some in few minutes and some in few hours (e.g. on db recovery).
Common limit for all services doesn't solve the problem (you need to set the
limit as high as is needed for the slowest service). This makes too large
window for fast starting service error detection beggining.

Correct solution could probably be to allow per-service timeout value on
startup. Monit in version 4.4 has 'initializing' monitoring mode value,
which is set to 'monitored' as soon as data are available and the service is
up. I think we can solve the problem via support of per-service
'initializing' mode timeout (which affects wait_start thread too).

Just my two cents ;)


----- Original Message ----- 
From: "Jan-Henrik Haukeland" <address@hidden>
To: "This is the general mailing list for monit" <address@hidden>
Sent: Thursday, August 26, 2004 11:34 AM
Subject: Re: Different startup time from polling time?

> On Aug 25, 2004, at 20:50, rory toma wrote:
> > The root problem is hard to fix...
> Yes it is.
> > How do you "fix" half a dozen external programs to "behave".
> I do not have a good answer to this, I wish I did, maybe others do? As
> I said we have had this discussion before and I think we decided that
> using a fixed time before poll is started does not really solve the
> problem. See e.g.
> If there is a strong support for re-adding this feature in monit I'm of
> course not going to oppose it. It would be interesting to know what
> Martin and Christian thinks and if they have changed their point of
> view on this feature.
> With Regards
> --
> Jan-Henrik Haukeland
> --
> To unsubscribe:

reply via email to

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