AWS backend for Coclobas #94
Labels
Comments
smondet
added a commit
that referenced
this issue
Apr 17, 2017
The notion of cluster is an AWS Batch “job-queue,” Coclobas does not create it but it checks that it's there; and fails at start-up if it isn't “`VALID`”. The `get_logs` function, does not et the logs, it instead generates a URL to visit on <https://console.aws.amazon.com/cloudwatch>. The tests now depend on (and use) `ppx_deriving_cmdliner`. The plugin is for now very basic: it only runs commands and does not understand `Program.t` values since there is no easy way of “mounting data” on the containers. When we go over-size for the command AWS-Batch fails with: > An error occurred (ClientException) when calling the > RegisterJobDefinition operation: Error executing request, Exception : > Limits Error: JobDefinitionsize is above the limits 20 KB For now, downstream users of the plugin should use other means to get bigger scripts to run (e.g. use S3 or mount an EFS filer). Killing jobs appears to work but it is somewhat fragile because AWS does not really “confirm” the killing cf. issue [`aws/aws-cli#2551`](aws/aws-cli#2551) (also, jobs take quite a while to change status in case of killing).
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Cf. #68
The text was updated successfully, but these errors were encountered: