error on /var/log/messages --> oom problem?
Robert Brockway
rbrockway-wgAaPJgzrDxH4x6Dk/4f9A at public.gmane.org
Fri May 6 09:34:18 UTC 2005
On Fri, 6 May 2005, Jerome Macaranas wrote:
> hmm.. i havent tried rebooting the box.. but im thinking of a long term
> solution without rebooting.. would adding addtional memory modules work?
If you hit an OOM condition then adding more memory is fine as long as the
new amount is sufficient to allow the box to run without hitting another
OOM later (ie, ideally you'd want to know how far short of the mark memory
went). Adding swap (which can be done on the fly, with nothing more than
a prepared file) is a good short term fix to an approaching OOM (assuming
you don't want to start killing processes yourself). You may wish to add
lots of swap and monitor swap usage over time to see how much additional
memory may be needed to run the system well).
Before anyone says "but swap is slow Rob!" consider some points:
1. I'm suggesting swap be used to avoid an OOM until such time as the
actual memory requirements can be properly assessed.
2. Swap allows the system to free up real memory for use in caching.
Many processes running on a long lived system rarely hit the cpu.
Allowing them to be swapped is a big win in performance.
3. The system overheads of having more swap than you need are really
minimal. OOM conditions can ruin your day. Think of it as an
insurance policy.
Rob
--
Robert Brockway B.Sc.
Senior Technical Consultant, OpenTrend Solutions Ltd.
Phone: +1-416-669-3073 Email: rbrockway-wgAaPJgzrDxH4x6Dk/4f9A at public.gmane.org http://www.opentrend.net
OpenTrend Solutions: Reliable, secure solutions to real world problems.
Contributing Member of Software in the Public Interest (http://www.spi-inc.org)
--
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
More information about the Legacy
mailing list