Maniphest T137442

Metrics for TimeoutError
Closed, ResolvedPublic
Actions

Assigned To

Authored By

	Halfak
	Jun 9 2016, 5:12 PM

Tags

Referenced Files

None

Subscribers

Description

See https://github.com/wiki-ai/ores/blob/master/ores/score_processors/score_processor.py#L96 We could catch a TimeoutError here report metrics on it.

Event Timeline

Halfak created this task.Jun 9 2016, 5:12 PM

Restricted Application added subscribers: Zppix, Aklapper. · View Herald TranscriptJun 9 2016, 5:12 PM

Halfak moved this task from Unsorted to New development on the Machine-Learning-Team board.Jun 14 2016, 5:28 PM

Does this task entail just implementing the timeout error, or integrating it into grafana as well?

We have a TimeoutError. We'll need a statsd metric and then a grafana graph too.

• schana claimed this task.Jun 30 2016, 10:50 AM

• schana edited projects, added Machine-Learning-Team (Active Tasks); removed Machine-Learning-Team.

• schana moved this task from Parked to Backlog on the Machine-Learning-Team (Active Tasks) board.

@Halfak could you explain the purpose of the count parameter found here and elsewhere?

• schana moved this task from Backlog to Review on the Machine-Learning-Team (Active Tasks) board.Jul 11 2016, 8:36 AM

Yes. count represents the number of scores that errored. See an implementation of this interface for an example of how it is used. E.g. https://github.com/wiki-ai/ores/blob/master/ores/metrics_collectors/logger.py#L42 or https://github.com/wiki-ai/ores/blob/master/ores/metrics_collectors/statsd.py#L104

I don't see where it is actually used though. In everything I can find, it just uses the default value.

You're right. It looks like the code that uses metrics_collector longer needs count.

Halfak moved this task from Review to Completed on the Machine-Learning-Team (Active Tasks) board.Jul 14 2016, 2:41 PM

Ladsgroup closed this task as Resolved.Jul 18 2016, 9:09 PM