Re: [SSI-users] 2-node OpenSSI-1.2.2 cluster hangs on "top" and "ps"

[Roger T. <rog...@gm...>]

Hi John,

We have had people compile the entire kernel on the cluster without a
glitch.  I think you may be having network/hardware problems.  If you dig
the mailing list you will find that Maurice had a similar problem with
period hangs and such while migrating large processes (can be similar to
rsync activity).  We eventually nailed it down to a network driver problem.

Roger


On 12/9/05, John David <jo...@na...> wrote:
>
> I am running FC2 + OpenSSI 1.2.2 set up successfully with a 2-node i686
> cluster in a testing environment.  The installation is fresh and stock,
> with no patches or modifications.  I have limited experience with
> OpenSSI, but it seems to be the solution we've been waiting for.
>
> I have been testing it lately before using it in production.  During my
> tests, I tried to use "rsync" to save copies of important files to
> another machine.  Rsync would, at times, hang hard -- something I have
> experienced with rsync before, so I did not attribute it to OpenSSI.
> During this time, I could (from another terminal or physically at the
> node) type "top" and the machine would hang.  Additionally, trying "ps
> -aux" would list a few processes, then hang, list another, hang, etc.
> for a very long time -- I don't think I ever got a complete list of
> processes.
>
> Thinking this to all be rsync-specific, I recently decided to omit rsync
> and instead opt for tar.  So I began to build the production cluster to
> be used as our server.
>
> However, in the process, as I was compiling certain programs I had the
> same exact symptoms.  This really concerns me!  Is OpenSSI 1.2.2 stable
> enough for production use?  I would think that, regardless of how bad a
> crash is on a process, I would be able to "top" or "ps" to find out the
> status.  I have stripped the cluster down to its most basic form, but am
> still running into this problem.  Could this be hardware- or
> network-related?  Would it be possible to locate the problem through
> logs?  Any insight would be very helpful in deciding which route to take
> (i.e. whether to use OpenSSI!)
>
> TIA,
> John David
>
>
> -------------------------------------------------------
> This SF.net email is sponsored by: Splunk Inc. Do you grep through log
> files
> for problems?  Stop!  Download the new AJAX search engine that makes
> searching your log files as easy as surfing the  web.  DOWNLOAD SPLUNK!
> http://ads.osdn.com/?ad_id=3D7637&alloc_id=3D16865&op=3Dclick
> _______________________________________________
> Ssic-linux-users mailing list
> Ssi...@li...
> https://lists.sourceforge.net/lists/listinfo/ssic-linux-users
>