[StarCluster] Multiple node cluster problems

Mayank Ahuja mahuja at qubole.com
Thu Dec 12 12:46:46 EST 2013


Hi Justin,

It was a bug in AWS us-east-1 region which was fixed after a couple of
days. The user data attribute of instances launched using one API call was
empty (on except the one with ami launch index = 0). As a result the
instances launched had no 'alias' tag.

It might be worth exploring if the dependence on user data attribute for
tag setting can be removed. Have seen a couple of times in the past when
fetch of user data attribute did not exceed (with few retries) due to may
be API latency issues at AWS end.

Thanks
Mayank

On Thu, Dec 12, 2013 at 10:30 PM, Justin Riley <jtriley at mit.edu> wrote:

> *major* issue if this change is indeed permanent. What region are you
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mailman.mit.edu/pipermail/starcluster/attachments/20131212/0e99fb30/attachment.htm


More information about the StarCluster mailing list