I've encountered a case where if a user doesn't specify enough RAM to the STAR aligner and STAR has to use swap space, the task will get stuck running indefinitely. Is there a way to limit swap? Thanks.
@jgould I checked with the team and they suggest increasing RAM. If RAM stays the same and swap is decreased, there is a chance that you might run into a hard memory limit and crash. Please let us know if you have any further questions!
Yes, the issue is solved by increasing RAM. However when a user unknowingly does not allocate enough RAM for this task, it seems that STAR starts using swap memory which causes the job to continue running (costing $) without making any progress. It would be nice to be able to configure docker to not use swap (https://docs.docker.com/config/containers/resource_constraints/#prevent-a-container-from-using-swap). Thanks.
Hi @jgould, great question. Customization of instances in Terra/WDL is limited by what is possible in the Google Pipelines API product; I am not aware of swap space being customizable.