StarCluster - Mailing List Archive

USERDATA_SCRIPTS not running with addnode

From: Cedar McKay <no email>
Date: Wed, 14 May 2014 16:30:41 -0700

Sorry to bump myself, but this is really the biggest difficulty I'm having with starcluster at the moment. It makes it difficult to use the addnode function, and even harder (impossible) to use the Elastic Load Balancer.

As a general question, is there a place other than the USERDATA_SCRIPTS that I could use to execute scripts on each node's boot? I'm thinking of /etc/init.d on the boot image, or something like that, but I've been reluctant to modify areas where I suspect some starcluster magic might be happening.

> Hello all.
> I'm using the USERDATA_SCRIPTS setting to run a script on cluster start. It is mounting an s3fs volume if that matters, though I've also tested it with simply touching a file. As an aside, is there a better place to mount a volume? I didn't dare touch fstab because I figured it might be generate/changed by starcluster depending on the cluster config.
> In any case, it works as expected when starting a cluster, but if I use addnode, the new nodes don't execute the script. Is this a bug, or behavior that I can change? It seems to me that new nodes should be started in the exact same way as other nodes.
> Thanks!
> Cedar
Received on Wed May 14 2014 - 19:30:45 EDT
This archive was generated by hypermail 2.3.0.


Sort all by: