DM-42027: new stack segfault with libpq

[hsinfang]

No description provided.

[kfindeisen]

Hmm... looks like there's a bug in the GHA config; it's building the base container but not the service one. 😞

[hsinfang]

I thought it's a deliberate choice not to build a new service container with just the Dockerfile.main change?

[kfindeisen]

Change makes sense, but please do test on dev.

[kfindeisen]

Would it be possible to add a comment here that this is a workaround for DM-XXXXX?

I'm not familiar with mamba, so I don't know how this line will interact with Conda's rubin-env. I think it would be good to test this before merging.

[kfindeisen]

I thought it's a deliberate choice not to build a new service container with just the Dockerfile.main change?

Sorry, just saw your comment on Jira. You're right, you'd have to override BASE_TAG_LIST for a branch service build to be meaningful anyway.

[timj]

I think conda-forge have fixed the libpq problem now.

conda-forge/conda-forge-repodata-patches-feedstock#598

[hsinfang]

As of earlier today when I tested d_2023_12_05 on prompt processing dev it still crashed. It had openssl 3.2.0.

Not familiar with our whole stack building process well enough, and not finding a jira DM ticket... how do I know if/when the workaround is no longer necessary, except deploying to dev to test?

I'll definitely test in dev before touching prod.