Comments on: where did my bandwidth go http://baldric.net/2008/08/20/where-did-my-bandwidth-go/ another voice in the babble on the net Sat, 19 May 2012 20:25:02 +0000 hourly 1 http://wordpress.org/?v=3.3.1 By: Chris Samuel http://baldric.net/2008/08/20/where-did-my-bandwidth-go/comment-page-1/#comment-18 Chris Samuel Sat, 23 Aug 2008 13:08:38 +0000 http://baldric.net/2008/08/20/where-did-my-bandwidth-go/#comment-18 Nah, no LART, he had no reason to believe that changing that parameter would cause WRF to go ballistic. Now, LART'ing the authors of dodgy scientific codes, on the other hard, that can have merit! Why they all decide they need to invent their own configuration/build system is beyond me, especially when autotools work (well, much better than what they invent anyway).. Grr... :-) Nah, no LART, he had no reason to believe that changing that parameter would cause WRF to go ballistic. Now, LART’ing the authors of dodgy scientific codes, on the other hard, that can have merit!

Why they all decide they need to invent their own configuration/build system is beyond me, especially when autotools work (well, much better than what they invent anyway)..

Grr… :-)

]]> By: Mick http://baldric.net/2008/08/20/where-did-my-bandwidth-go/comment-page-1/#comment-17 Mick Fri, 22 Aug 2008 15:24:00 +0000 http://baldric.net/2008/08/20/where-did-my-bandwidth-go/#comment-17 Chris Cheers. I'd not heard of tcptrack before - it's a cool utility and I'll add it to my toolbox. I assume that you visited the user in question with a LART in hand. Mick Chris

Cheers. I’d not heard of tcptrack before – it’s a cool utility and I’ll add it to my toolbox.

I assume that you visited the user in question with a LART in hand.

Mick

]]> By: Chris Samuel http://baldric.net/2008/08/20/where-did-my-bandwidth-go/comment-page-1/#comment-16 Chris Samuel Fri, 22 Aug 2008 12:28:45 +0000 http://baldric.net/2008/08/20/where-did-my-bandwidth-go/#comment-16 Nice find Mick! It's packaged in Debian and Ubuntu which is handy. I've also used tcptrack on our NFS server to work out which nodes on which cluster are hammering it when we see high load averages and try and track back to users. Was very handy when someone changed a configuration parameter in their WRF weather model which caused every process in a 32 CPU job busy loop on an error message.. They managed to sustain 800Mb/s writes! Nice find Mick! It’s packaged in Debian and Ubuntu which is handy.

I’ve also used tcptrack on our NFS server to work out which nodes on which cluster are hammering it when we see high load averages and try and track back to users. Was very handy when someone changed a configuration parameter in their WRF weather model which caused every process in a 32 CPU job busy loop on an error message.. They managed to sustain 800Mb/s writes!

]]>