Intermittent Connectivity Problem [was Re:linux problem]

[Alex Beamish]

On 12/12/05, Robert Brockway <rbrockway-wgAaPJgzrDxH4x6Dk/4f9A at public.gmane.org> wrote:
>
> On Mon, 12 Dec 2005, Jerome Macaranas wrote:
>
> > hi all,
> >
> >       I have a system that serves HTTP and PGSQL ( serves locally only
> contents for
> > dynamic HTTP conntents )... it runs on Redhat 9 with 2G memory..... for
> some
> > reason the box just suddenly became unavailable... the server is
> pingable..
> > but services running on it is unaccessible .. HTTP, SSH... Ive already
> > checked the logs... but didn't saw anything unusual..


In addition to all of the good things Rob mentioned, have you VACUUMed your
Postgres database recently?

I experienced a 'knee' of performance on one of my database servers
recently, and did a VACUUM that took 45 minutes to complete. After that,
performance was almost two orders of magnitude better. There used to be a
daily VACUUM on that server but due to some other changes it was disabled.
There is now a daily VACUUM on all of my database servers.

And of course, when all else fails, check what the log files are saying. If
you use Nagios to monitor the server (from another box, obviously) you
should get some kind of warning that the system is degrading.

Looks like you've got some fun investigative work ahead of you.

Alex

If the box becomes unavailable how are you getting in to checkout the
> logs?  If it is after a reboot then the horse has bolted - you need to
> capture important data right after the system has the problem.
>
> Can you checkout the console when this is happening?  If it is not easy to
> get to the keyboard & monitor consider setting up a serial console - this
> is great for capturing console output for later.  Also checkout dmesg.
>
> Turn all the logs you can to debug.  This will add load to the server.  If
> the box is in production watch the system load.
>
> How long does the system last before it suffers this problem?  Minutes,
> days, weeks?
>
> Google for cross references to your hardware, kernel, filesystems, etc and
> watch for others reporting similar problems.
>
> Also, it's better to be more specific with the title.  Some people may not
> checkout the thread based on the title, or you may get the wrong people
> doing so (ie, those without enough experience).  If you used a subject
> line of something like "Intermittent Connectivity Problem" (which I have)
> you may tend to draw in the right audience more easily.
>
> Cheers,
>
> Rob
>
> --
> Robert Brockway B.Sc.           Phone:  +1-416-669-3073
> Senior Technical Consultant     Email:  support-wgAaPJgzrDxH4x6Dk/4f9A at public.gmane.org
> OpenTrend Solutions Ltd.        Web:    www.opentrend.net
> We are open 24x365 for technical support.  Call us in a crisis.
> --
> The Toronto Linux Users Group.      Meetings: http://tlug.ss.org
> TLUG requests: Linux topics, No HTML, wrap text below 80 columns
> How to UNSUBSCRIBE: http://tlug.ss.org/subscribe.shtml
>



--
----------
Linux, Firefox and GMail .. what a combination.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://gtalug.org/pipermail/legacy/attachments/20051212/052d1c67/attachment.html>