[Top][All Lists]

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

Re: Silent stderr patch

From: Christopher P. Lindsey
Subject: Re: Silent stderr patch
Date: Mon, 2 Oct 2006 10:25:52 -0500
User-agent: Mutt/1.5.10i

> We'll put the patch on hold then in case we find it useful  later.

Although it sounds like the patch is on hold, I'd like to make one more 
pitch/argument for its inclusion based on what's been said so far.

> >>Looks like a useful patch, what do others think? Another  
> >>alternative is of course to start monit with 2>/dev/null. Using  
> >>"set nostderr" is better though.

One thing to note is that not all operating systems support stderr
redirection on programs run via init.  Certain versions of AIX have
this limitation, for example.

> >Sounds useful. Although, wouldn't a "monit unmonitor all" have  
> >helped in this case? That's the first thing I do, when something  
> >has gone wrong with the system and I try to fix stuff.

Since the unmonitor features are only available to users running the 
monit Web interface, this isn't always a viable option.

Another issue is that it requires a monit-specific way to treat the
problem once it appears.  That's not something that scales well in a
large organization with a single system admin but lots of support staff
and different configurations.  In my workplace we have thousands of
machines with some managed by different groups (and different monitoring
software).  If a problem occurs after hours, the 24-hour support staff
should be able to log in with a root password and check things; they
can't have a list of different commands to run on different systems.
Suggesting that people use 'monit unmonitor all' implies a homogenous
working environment where monit is always the culprit.

Ultimately, the problem that I see the patch addressing is that some
operating systems and configurations have no standardized way to limit
stderr messages from scrolling across the console.

In all honesty, I don't think that monit should even send messages to
stderr and would have written the patch to make stderr output an opt-in
instead of opt-out process, but that would have broken functionality
with older systems and configuration files anyhow.  :)


reply via email to

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