Page MenuHomePhabricator

Add monitoring to Kask service
Closed, InvalidPublic

Description

As an operations engineer, I would like to be able to use the test running in production to test if services are alive.

  • Needs support for retries

Priority: Optional

Acceptance Criteria:

  • Check the state of a running service
  • Detect that a service is not running

Event Timeline

@CCicalese_WMF: If you remember details and if this task is still valid, could you elaborate which "the test" codebase and/or team this relates to, and make the task title more specific? Thanks.

@daniel is there any benefit to leaving this task open at this point? If so, maybe you have more context?

@daniel: Could you please answer the last comment? Thanks in advance!

@daniel is there any benefit to leaving this task open at this point? If so, maybe you have more context?

From the tags, I gather that this task is about monitoring Kask. Who owns this servic ein production? Persistance? Service Ops? I suppose it would be up to them to decide.

MSantos subscribed.

@Aklapper this service is currently unowned, not sure how to move forward here. To reflect reality I'll remove MediaWiki-Engineering but happy to discuss it further to find a final solution.

I couldn't find a component board for Kask, maybe that's where it should be (?).

@Aklapper this service is currently unowned, not sure how to move forward here. To reflect reality I'll remove MediaWiki-Engineering but happy to discuss it further to find a final solution.

I think that I implicitly brought Kask over to Data-Persistence when I moved. That said, there might be an argument to be made that as a storage primitive, it belongs with DP anyway (read: explicitly). Longer term, that is probably something we should have a conversation about, as well we whether or not Kask as it exists is something we should perpetuate (versus —for example— using client-side k/v abstractions, or moving to a single multi-tenant k/v service).

TL;DR You can treat Data-Persistence as the owner, at least for the time being.

I couldn't find a component board for Kask, maybe that's where it should be (?).

Cassandra is probably the best choice for now.

@CCicalese_WMF: If you remember details and if this task is still valid, could you elaborate which "the test" codebase and/or team this relates to, and make the task title more specific? Thanks.

@daniel is there any benefit to leaving this task open at this point? If so, maybe you have more context?

@daniel is there any benefit to leaving this task open at this point? If so, maybe you have more context?

From the tags, I gather that this task is about monitoring Kask. Who owns this servic ein production? Persistance? Service Ops? I suppose it would be up to them to decide.

I don't know that I have enough context to understand this request...and it sounds like I might not be the only one. I'm going to be bold and close it, but if I've erred in doing so, then please re-open (preferably with a more detailed task description).