Page MenuHomePhabricator

Identify candidate tools to test migrating to the Toolforge on bare metal POC
Open, HighPublic

Description

What the title says. We need to identify a subset of Toolforge tools that we understand to be representative of the total set of tools that we can test migrating to the POC. This will hopefully help de-risk a potential mass migration of tools if Toolforge was to be fully moved into the production infrastructure.

Event Timeline

CCiufo-WMF triaged this task as High priority.

I agree with @Andrew's assessment in T407296#11274680 that we won't benefit from simply selecting the "simplest" tools or the ones we think will be "easiest" to migrate. That would result in backloading all of the risk and potentially signaling that this whole effort will be easier than it really is. That said, we also want to avoid trying to make a POC so comprehensive that is actually just ends up being the entire end-product. It would be helpful to start by defining what constitutes a tool's "complexity" (i.e. is it compute needs? storage? build & deploy setup?). Maybe there is a dimension of complexity we can intentionally leave out of the POC if we know it won't be trivial but can be cleanly sliced out without affecting a majority of tools.

I maintain/co-maintain a number of tools that could be offered up for tribute here if desired:

"Legacy" tools use shared Docker images and NFS storage. Many (most? all?) of them could be updated to use build service containers. I have been holding off on moving things to components due to the lack of webservice support, but some could likely move without that.