<div dir="ltr">I submitted a pull request that fixes this:<br><br><a href="https://github.com/golharam/StarCluster/commit/e4b776f2d4fd18827cb77d8e5422dc5dadcc6e31">https://github.com/golharam/StarCluster/commit/e4b776f2d4fd18827cb77d8e5422dc5dadcc6e31</a><br><br></div><div class="gmail_extra"><br><div class="gmail_quote">On Tue, Apr 21, 2015 at 10:37 PM, Ryan G <span dir="ltr"><<a href="mailto:ngsbioinformatics@gmail.com" target="_blank">ngsbioinformatics@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 dir="ltr"><div><div><div>I figured out what's going on. When StarCluster creates a new cluster, it reads the cluster template from the config file and knows the full path of the userdata_script. This gets fed to amazon when creating new nodes. This information is somehow encoded into a metatag (along with a whole bunch of other information) on the security-group of the cluster.<br><br>When adding nodes to an existing cluster, the config file is never referenced. The information about the cluster is encoded in a metatag on the security group for the cluster. StarCluster reads the metadata of the master node when creating additional nodes. In the metadata, the text (and name) of the script is loaded into a userdata variable which never gets propagated when creating new nodes. <br><br></div>If I feed just the body of script of the userdata of the master node, there is no path specified so it would fail. I think the fullpath of the userscript needs to be stored in the metadata of the securitygroup as the other information about the cluster. I'll look into applying this and make a patch. <br><br></div>If anyone more familiar with StarCluster knows more about this or can make recommendations on a better fix, please let me know. Perhaps storing the name of the cluster template then referencing the config file to determine the user script but this wouldn't go along with how the rest of StarCluster seems to work.<br><br></div><div>Ryan<span class="HOEnZb"><font color="#888888"><br></font></span></div><span class="HOEnZb"><font color="#888888"><div><br></div>Ryan<br></font></span></div><div class="HOEnZb"><div class="h5"><div class="gmail_extra"><br><div class="gmail_quote">On Thu, Apr 16, 2015 at 5:32 PM, Ryan G <span dir="ltr"><<a href="mailto:ngsbioinformatics@gmail.com" target="_blank">ngsbioinformatics@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 dir="ltr">Yes, I've since discovered USERDATA_SCRIPTS doesn't work when adding nodes. Perhaps I'll take a look since this is a big issue for me.<br><br></div><div class="gmail_extra"><br><div class="gmail_quote"><span>On Thu, Apr 16, 2015 at 1:42 PM, Cedar McKay <span dir="ltr"><<a href="mailto:cmckay@uw.edu" target="_blank">cmckay@uw.edu</a>></span> wrote:<br></span><div><div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">As an FYI, I’ve only been able to get USERDATA_SCRIPTS to work at cluster launch. Addnode and loadbalance don’t seem to run the script, therefore the feature is useless to me. I’d love to hear that I’m wrong.<br>
<div><div><br>
<br>
> On Apr 16, 2015, at 10:37 AM, Ryan G <<a href="mailto:ngsbioinformatics@gmail.com" target="_blank">ngsbioinformatics@gmail.com</a>> wrote:<br>
><br>
> Turns out this is supported by StarCluster in the config file using USERDATA_SCRIPTS<br>
<br>
<br>
</div></div></blockquote></div></div></div><br></div>
</blockquote></div><br></div>
</div></div></blockquote></div><br></div>