submission monitor 20 per pass #2670
Conversation
| @@ -57,6 +57,8 @@ submissionmonitor { | |||
| trackDetailedSubmissionMetrics = true | |||
| attributeUpdatesPerWorkflow = 20000 | |||
| enableEmailNotifications = true | |||
| # how many workflows should the submission monitor attempt to process and update at once? | |||
| workflowsPerMonitorPass = 20 | |||
There was a problem hiding this comment.
what is a good initial value for this setting?
| statuses: WorkflowStatuses.WorkflowStatus* | ||
| ): ReadAction[Seq[WorkflowRecord]] = | ||
| findWorkflowsBySubmissionId(submissionId).filter(_.status inSetBind (statuses.map(_.toString))).result | ||
| findWorkflowsBySubmissionId(submissionId).filter(_.status inSetBind (statuses.map(_.toString))).take(limit).result |
There was a problem hiding this comment.
the .take(limit) here should result in only returning ${workflowsPerMonitorPass} workflows to the submission monitor actor for any given pass.
There was a problem hiding this comment.
The drawback to adding a limit here is that if the first limit rows from the db are still running, but limit+1, limit+2, limit+3 are done … the monitor won't complete the ones that are done. This isn't great.
There was a problem hiding this comment.
You could alleviate this by select … order by rand() limit 20; in mysql, to get a random selection of 20 rows that match the where clause. I'm not sure how to do that in Slick.
| def countWorkflowRecsForSubmissionAndStatuses(submissionId: UUID, | ||
| statuses: WorkflowStatuses.WorkflowStatus* | ||
| ): ReadAction[Int] = | ||
| findWorkflowsBySubmissionId(submissionId).filter(_.status inSetBind (statuses.map(_.toString))).length.result |
There was a problem hiding this comment.
previously, SubmissionMonitorActor.updateSubmissionStatus() called listWorkflowRecsForSubmissionAndStatuses(), returning all workflow records as a Seq, and then checked if that Seq was empty or not. This moves the logic to the db, returning just a count. I needed to change something here, because by adding a limit to listWorkflowRecsForSubmissionAndStatuses(), it would have been invalid for SubmissionMonitorActor.updateSubmissionStatus() - two birds with one stone?
Ticket:
PR checklist
model/, then you should publish a new officialrawls-modeland updaterawls-modelin Orchestration's dependencies.