[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Problem with monit reload ("design" note)
From: |
Martin Pala |
Subject: |
Problem with monit reload ("design" note) |
Date: |
Tue, 11 Feb 2003 11:58:34 +0100 |
User-agent: |
Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.2.1) Gecko/20021226 Debian/1.2.1-9 |
Hi,
there's little problem with monit reload. Because monit is stateless in
the configuration reload, it could cause mess after it.
Oliver's patch for monit restart (p.y):
/* if process is already running... set validate true */
if (is_process_running(p))
p->do_validate= TRUE;
suits for most situations, however it isn't "clear" solution for it. The
problem with configuration reload is:
In the case of manual mode, monit should start monitoring the service
ONLY if it was instructed to do so. Usualy if the 'manual' mode process
runs on the node, it is correct to enable monitoring after reload again,
but not in the case that:
- monit stopped process monitoring because of timeout
- the process shoudn't run on that node
The problem with timed out process monitoring affects all monitoring
modes => with above patch monit will start monitoring again if the
process is running, regardless the state before configuration reload.
Probably this behavior is better then previous, but the final solution
will be to replace this patch with statefull configuration (we discussed
it on the list some time ago). This is part of external configuration
sources support as well, so maybe i can start with its implementation
after i will finish device monitoring stuff.
Cheers,
Martin
- Problem with monit reload ("design" note),
Martin Pala <=