[Gllug] Identifying what processes are waiting on IO

[John Hearns]

2008/12/7 Stephen Nelson-Smith <sanelson at gmail.com>

>
>
> So far all I've done is use lsof and look at the major and minor device
> numbers and mount points - but this really doesn't tell me much.
>

Don't forget about 'fuser' also.

You should be looking at I/O elevator tuning I think - but as you say, you
should
be looking at finding what processes are responsible first.

Once you've identified the processes and disk devices which are causing the
problems,
then maybe a bit of lateral thought can help. As a for instance, one of the
applications I'm familiar with, Ganglia, uses a round-robin-tool database,
which is frequently updated with information from Beowulf cluster nodes - so
frequently that on a big cluster the didks are just continually busy. The
recognised fix is to write the data to a RAMdisk then flush to real disk
every so often.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.lug.org.uk/pipermail/gllug/attachments/20081207/2561fe4f/attachment.html>
-------------- next part --------------
-- 
Gllug mailing list  -  Gllug at gllug.org.uk
http://lists.gllug.org.uk/mailman/listinfo/gllug