[Github][CI] Remove premerge container

[boomanaiden154]

This patch removes the container from the premerge job. We are moving away from the kubernetes executor back to executing everything in the same container due to reliability issues. This patch updates everything in the premerge job to work.

This is part of a temp fix to llvm/llvm-zorg#362.

[boomanaiden154]

Mostly just putting this up so I can quickly land it when I update everything in llvm-zorg after the necessary infrastructure has gone up.

[llvmbot]

@llvm/pr-subscribers-github-workflow

Author: Aiden Grossman (boomanaiden154)

Changes

This patch removes the container from the premerge job. We are moving away from the kubernetes executor back to executing everything in the same container due to reliability issues. This patch updates everything in the premerge job to work.

---

Full diff: /~https://github.com/llvm/llvm-project/pull/123483.diff

1 Files Affected:

• (modified) .github/workflows/premerge.yaml (-5)

diff --git a/.github/workflows/premerge.yaml b/.github/workflows/premerge.yaml
index 261dc8bbb97e0a..ddef6206bcae89 100644
--- a/.github/workflows/premerge.yaml
+++ b/.github/workflows/premerge.yaml
@@ -18,11 +18,6 @@ jobs:
     concurrency:
       group: ${{ github.workflow }}-${{ github.event.pull_request.number || github.sha }}
       cancel-in-progress: true
-    container:
-      image: ghcr.io/llvm/ci-ubuntu-22.04:latest
-    defaults:
-      run:
-        shell: bash
     steps:
       - name: Checkout LLVM
         uses: actions/checkout@v4

[tstellar]

So will this be the container used on all the runners?

[boomanaiden154]

So will this be the container used on all the runners?

All the Linux runners, yes. Originally we used one of the official Github Actions images that would then spawn a new container where all the work was actually performed. We had reliability issues with that, so we're going back to running the runner/jobs in a single container. We will try and switch back to the original technique when we can cause it adds some nice flexibility, but we need to get the reliability problems fixed first.

[Keenuts]

Fine with the code change, but the PR description should reference an LLVM-zorg issue which explains the issues and switch, and this issue should link to the upstream konnectivity issue.
(Otherwise we will forget)