What about cycling the power using a PDU?<br><br><div class="gmail_quote">On Tue, Feb 28, 2012 at 2:43 AM, Daniel Fernando Coimbra <span dir="ltr"><<a href="mailto:danielfcoimbra@gmail.com">danielfcoimbra@gmail.com</a>></span> wrote:<br>
<blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">I assume that by "turning off" you mean actually power down the node. I<br>
am just curious on how do you intend to power it up again later. I<br>
suppose you could use something like WakeUp on Lan, but I never actually<br>
got to test this kind of thing and don't know how it would behave on a<br>
high traffic network (I suppose the network card doesn't keep it's IP<br>
once it's in such state).<br>
<div class="im"><br>
On 02/26/2012 08:24 PM, Arka Aloke Bhattacharya wrote:<br>
> Hi everyone,<br>
><br>
> I am a PhD student at UC Berkeley, and I wanted to add a "turning off<br>
> idle/underutilized servers" feature to our 100 server torque+maui<br>
> deployment. However, I want to implement this feature using only<br>
</div>> existing torque+ maui interfaces and extensions ( i,e _without<br>
> modifying_ the torque or maui source code in any way ).<br>
<div class="im">><br>
> My proposed way is to<br>
> 1. monitor the maui queue length , and estimate the number of servers<br>
> I can switch off.<br>
> 2. I would then use "pbsnodes -o <nodename>" command to render a<br>
> certain number of servers offline for scheduling.<br>
> 3. A bash script would turn the servers off.<br>
><br>
> The servers would be turned back on (and added to the torque nodes<br>
> list) when the queue length increases beyond a certain threshold.<br>
><br>
> I had two questions :<br>
><br>
> 1. Is there any existing open source code which already implements the<br>
> "turning off idle servers" functionality in torque ?<br>
> 2. Are there complications that would arise if I implemented the<br>
> "turning-off idle servers" feature in my proposed way ? [ e.g - Is it<br>
> possible that after turning off servers, they would lose some state<br>
> and hence would not get added to the torque <nodes_list> when turned<br>
> back on? Are there long lived TCP connections which need to be<br>
> restarted separately ? , etc ]<br>
><br>
> It would be great if anyone could help.<br>
><br>
> Thanks a lot,<br>
> Arka.<br>
><br>
><br>
</div>> _______________________________________________<br>
> torqueusers mailing list<br>
> <a href="mailto:torqueusers@supercluster.org">torqueusers@supercluster.org</a><br>
> <a href="http://www.supercluster.org/mailman/listinfo/torqueusers" target="_blank">http://www.supercluster.org/mailman/listinfo/torqueusers</a><br>
<br>
_______________________________________________<br>
torqueusers mailing list<br>
<a href="mailto:torqueusers@supercluster.org">torqueusers@supercluster.org</a><br>
<a href="http://www.supercluster.org/mailman/listinfo/torqueusers" target="_blank">http://www.supercluster.org/mailman/listinfo/torqueusers</a><br>
</blockquote></div><br>