[
https://issues.apache.org/jira/browse/JCLOUDS-1379?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16469709#comment-16469709
]
Alex Heneveld commented on JCLOUDS-1379:
----------------------------------------
Discussion at [https://forums.aws.amazon.com/thread.jspa?threadID=98879]
suggests instance types in some cases can vary from day to day and from AZ to
AZ. And you can get this error even in the AWS UI.
But I think some instance types are never available, but I don't know if it's
documented which ones. (Also I get the sense some are more plentiful than
others.)
I'm not sure what is best. It seems wrong to hard-code what we think is the
case in jclouds. But equally it's not good if we're picking instance types
that fail.
I was hoping that the periodically updated pricing list might tell us but it
suggests all instance types are available in all regions :( --
[https://gist.github.com/ahgittin/1dee8b9f322252405879e1cf6d271932] .
Maybe the answer is to see what RI's are available for purchase (by region,
even by AZ, etc):
[https://docs.aws.amazon.com/cli/latest/reference/ec2/describe-reserved-instances-offerings.html]
– presumably if RI's are available for an instance type then you won't get an
insufficient capacity error. It might give us false negatives but we shouldn't
get false positives.
Possibly a simpler option is a heuristic, _if_ it's the case that this error is
mainly for older instance types where there just aren't many – in which case
maybe it could be solved by preferring modern instance types – eg in jclouds
scoring families and preferring more modern families. Worth recording instance
types + locations that are failing to see if we can figure this out.
> incompatible instance types selected in new AWS regions
> -------------------------------------------------------
>
> Key: JCLOUDS-1379
> URL: https://issues.apache.org/jira/browse/JCLOUDS-1379
> Project: jclouds
> Issue Type: Bug
> Components: jclouds-compute
> Affects Versions: 2.0.3
> Reporter: Alex Heneveld
> Priority: Major
> Labels: aws-ec2
>
> Using AWS EC2, if I specify minRam 4096M and minCores 4, I get a VM fine in
> some regions (eu-west-1) but errors in others (eu-west-2)
> AWSResponseException: request POST https://ec2.eu-west-2.amazonaws.com/
> HTTP/1.1 failed with code 400, error: AWSError\{requestId='...',
> requestToken='null', code='Unsupported', message='The requested configuration
> is currently not supported. Please check the documentation for supported
> configurations.' ...}
> I assume this is because some instance types are not available in some
> regions.
> Can we do further filtering on instance type selection so that
> minRam/minCores work as expected?
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
