wilfred-s commented on PR #986: URL: https://github.com/apache/yunikorn-core/pull/986#issuecomment-2426469406
This e2e test is failing due to the fact that the list of requests has a mixture of allocated and unallocated entries and the merging of the Ask and Allocation objects. The fix for the timeout tracking does not filter the requests lists for already allocated requests and needs to do that. We seem to be adding that same filter everywhere now to compensate for not removing the allocation from the request list. This causes two issue: * We send allocations to the shim twice to be released. Once from the request list once from the allocations list. * Asks and allocations are processed the same way and both are returned to the core which can cause double counting them as timed out. The other problem detected on log analysis is that the shim tries to clean up the same placeholders multiple times. First based on the core request then based on internal logic as part of the placeholder code. All cleanup in the placeholder code fails as the core has already done it. It does trigger more release messages to be sent to the core which then get ignored as the work is already done. That needs a cleanup in a follow up jira. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
