<div dir="ltr">Thanks for kicking this around with me, Nik. <div><br></div><div>Since SGE isn't installed on the new node til the sge plugin runs, and so the exechost doesn't exist in the config anywhere, I'm pretty sure I can't modify the node's exechost complex_values until after sge plugin runs. (I should probably test this hypothesis before discounting it....)</div>
<div><br></div><div>I've already got a local OGS configurator plugin that runs on cluster start, but it doesn't have an on_add_node method to it yet, as it's mostly about global config settings.<div><br></div>
<div>Good to know about scriptrunner.py; may come in handy!<br></div><div><br></div><div>Thanks again,</div><div>Lyn</div></div></div><div class="gmail_extra"><br><br><div class="gmail_quote">On Mon, Jan 6, 2014 at 3:02 PM, Nik Krumm <span dir="ltr"><<a href="mailto:nkrumm@gmail.com" target="_blank">nkrumm@gmail.com</a>></span> wrote:<br>
<blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><div style="word-wrap:break-word"><div class="im">Hi Lyn,<div><br></div><div>How about using the SGE plugin (this one [1] not the baked-in setup) and a simple script-runner plugin (see [2] for the one I use), but having the SGE plugin last in the order. That way the setup script should run first, and then SGE will be configured second.</div>
<div>I have not tested this approach but in principal it will work.</div><div><br></div><div>So a sample config snippet:</div><div><br></div><div><font face="Courier">[cluster smallcluster]</font></div><div><font face="Courier"> # don't use baked-in SGE<br>
disable_queue = True</font></div><div><font face="Courier"> # the order matters here</font></div><div><font face="Courier">plugins = scriptrunner, sge</font></div><div><font face="Courier"><br></font></div><div><font face="Courier">[plugin sge]<br>
setup_class = starcluster.plugins.sge.SGEPlugin<br>master_is_exec_host = False</font></div><div><font face="Courier"><br></font></div><div><font face="Courier">[plugin scriptrunner]<br>setup_class = scriptrunner.ScriptRunner<br>
# note this is the path on the cluster itself, e.g., NFS drive.</font></div><div><font face="Courier">node_startup_script = /path/to/script</font></div><div><br><div><div>Good luck,</div><div>Nik</div><div><br></div><div>
[1] <a href="http://star.mit.edu/cluster/docs/latest/plugins/sge.html" target="_blank">http://star.mit.edu/cluster/docs/latest/plugins/sge.html</a></div><div>[2] <a href="https://gist.github.com/nkrumm/8292905" target="_blank">https://gist.github.com/nkrumm/8292905</a></div>
</div></div><div><br></div></div><div><div class="im"><div>On Jan 6, 2014, at 10:14 AM, Lyn Gerner wrote:</div><br></div><blockquote type="cite"><div><div class="h5"><div dir="ltr">Hi Folks,<div><br></div><div>I need to make node-resource-specific SGE config changes everytime I add a new node. To avoid race conditions, I need to bring the node up in a disabled state, make the changes, then enable it for running jobs.</div>
<div><br></div><div>I don't see a way to do this with StarCluster commands or config settings. Maybe I missed something.</div><div><br></div><div>Could I get a best practice suggestion for how to achieve that set of interactions? I don't want to hack the SGE plugin if there's a better way.<br>
</div><div><br></div><div>Thanks,</div><div>Lyn</div></div></div></div><div class="im">
_______________________________________________<br>StarCluster mailing list<br><a href="mailto:StarCluster@mit.edu" target="_blank">StarCluster@mit.edu</a><br><a href="http://mailman.mit.edu/mailman/listinfo/starcluster" target="_blank">http://mailman.mit.edu/mailman/listinfo/starcluster</a><br>
</div></blockquote></div><br></div><br>_______________________________________________<br>
StarCluster mailing list<br>
<a href="mailto:StarCluster@mit.edu">StarCluster@mit.edu</a><br>
<a href="http://mailman.mit.edu/mailman/listinfo/starcluster" target="_blank">http://mailman.mit.edu/mailman/listinfo/starcluster</a><br>
<br></blockquote></div><br></div>