[MITgcm-devel] forge is slow: too many "httpd" proc ?

[marshall-admin at techsquare.com]

it seems that a long-term solution may be
required. perhaps a 32bit box with 2G ram
isn't quite enough for all the traffic, etc.

jmc, are you around today (friday) ? 
perhaps we can consider some optimization
strategies...

[greg]


> Date: Thu, 6 May 2010 22:33:41 -0400
> From: Jean-Michel Campin <jmc at ocean.mit.edu>
> Cc: MITgcm-devel at mitgcm.org, Dimitris.Menemenlis at jpl.nasa.gov
> MIME-Version: 1.0
> 
> Hi Greg,
> 
> Seems that everything was fine until ~21. pm but by 21.50 it was 
> not so good. And it does not seem to get much better with time.
> It's not as bad as it has been, but there are still lot of httpd 
> proc running on forge (~130) and because it's using the swap,
> it's again quiet slow:. 
> 
> > forge{~}% top -n 1
> > top - 22:30:27 up 181 days, 14:09,  4 users,  load average: 139.30, 139.86, 139.55
> > Tasks: 390 total,   1 running, 322 sleeping,   0 stopped,  67 zombie
> > Cpu(s):  6.0%us,  2.0%sy,  0.0%ni, 89.5%id,  1.9%wa,  0.0%hi,  0.6%si,  0.0%st
> > Mem:   2062508k total,  2013364k used,    49144k free,     3288k buffers
> > Swap:  8191992k total,  2530888k used,  5661104k free,    24944k cached
> 
> Thanks,
> 
> Jean-Michel
> 
> On Wed, May 05, 2010 at 08:47:29AM -0400, marshall-admin at techsquare.com wrote:
> > nope. i am working on blocking the offending
> > ip-addresses at present.
> > 
> > [greg]
> > 
> > 
> > > Date: Wed, 5 May 2010 00:55:40 -0400
> > > From: Jean-Michel Campin <jmc at ocean.mit.edu>
> > > Cc: MITgcm Devel <MITgcm-devel at mitgcm.org>, 	Dimitris.Menemenlis at jpl.nasa.gov
> > > MIME-Version: 1.0
> > > 
> > > Hi Greg,
> > > 
> > > Thanks for fixing this earlier tonight.
> > > 
> > > It was working well for some time, but unfortunately,
> > > it looks like around mid-night things degrade and
> > > I have again problems to acces mitgcm.org
> > > (I managed to count 257 httpd processes at 12.50 am, not good)
> > > May be it will fix by itself ?
> > > 
> > > Thanks,
> > > Jean-Michel
> > > 
>