1543683 - crontabber node needs webapp configuration

Status: RESOLVED FIXED

(Socorro :: Infra, task, P1)

Description

[Will Kahn-Greene [:willkg] ET needinfo? me]

Previously, crontabber was a socorro configman app that ran using configman-style configuration. With bug #1493687, we're rewriting crontabber changing it from a socorro configman app into a Django command that runs other Django commands.

This inherently changes the context in which the code runs. Instead of running in a socorro configman app context, it needs to run in a webapp context. This came up with the updatesignatures rewrite which needs to use memcached which is only configured in the webapp.

We need a crontabber node. We're in an interim period where it needs crontabber AND webapp configuration. Soon it'll only need webapp configuration.

This bug covers figuring that out and making the required changes.

Comment 1

[Will Kahn-Greene [:willkg] ET needinfo? me]

Making this a P1 since it definitely blocks updatesignatures and probably blocks rewriting other jobs.

Miles: What should we do here short term (today) vs. long term (soon)? Should we copy all the webapp configuration into the crontabber node? Then long term, we'll take out the configman-style configuration from the crontabber node and continue forward with that?

Comment 2

[Will Kahn-Greene [:willkg] ET needinfo? me]

Miles is at a conference. I talked to Brian.

We're probably going to scrap all this when we move to GCP, so we just need to solve the short term. I have fears that today we're hitting problem with memcached_host, and tomorrow we'll hit something different. Brian said we should just copy all the bits over so the crontabber and webapp instance types have the same configuration.

I'll work on that next.

Comment 3

[Will Kahn-Greene [:willkg] ET needinfo? me]

Attachment: GitHub Pull Request

Comment 4

[Will Kahn-Greene [:willkg] ET needinfo? me]

The updatesignatures command kicked off and ran fine. I think we're good here. Marking as FIXED.