[cbartz]

What this PR does

Bumps the pinned GARM workload commit to the merge of cloudbase/garm#926.

Why we need it

We regularly see runners stuck in a state, most likely because of SQL queries that hang. #926 addresses three causes in GARM's database layer:

• Every PostgreSQL query now has a timeout (statement 60s, lock 30s, connect 10s by default). Migrations are exempt.
• Runners left in creating are cleaned up when recording them as running fails.
• Instance and scale set updates no longer write each other's rows, which removes an ABBA deadlock between them.

Checklist

• Changes comply with the project's coding standards and guidelines (see CONTRIBUTING.md and STYLE.md)
• CONTRIBUTING.md has been updated upon changes to the contribution/development process (e.g. changes to the way tests are run)
• Technical author has been assigned to review the PR in case of documentation changes (usually *.md files)
• I updated docs/changelog.md with user-relevant changes
• I used AI to assist with preparing this PR
• I added or updated tests as needed (unit and integration)
• If integration test modules are used: I updated the workflow configuration
  (e.g., in .github/workflows/integration_tests.yaml, ensure the modules list is correct)
• If this PR involves a Grafana dashboard: I added a screenshot of the dashboard
• If this PR involves Terraform: terraform fmt passes and tflint reports no errors
• If this PR involves Rockcraft: I updated the version
• If this PR adds/removes a charm, or changes a charm's base class, conventions, tooling, or repo structure: I updated the relevant AGENTS.md
• If this PR changes .copilot-collections.yaml or .github/instructions/: I re-checked whether the AGENTS.md "12-factor divergences" guidance still matches the upstream copilot-collections guidance

Not applicable or not ticked:

• No new test here. Hung queries and the deadlock can't be reproduced reliably in the integration suite. Upstream #926 adds tests for them.
• The rock version stays 0.2; it isn't bumped per workload pin (see fix(garm): bump pinned commit to fix scale set pool ID tag length #351).
• CONTRIBUTING.md, Grafana, Terraform, AGENTS.md and copilot-collections are untouched.

Test plan

• Built the rock with ./build-garm-rock.sh (Go 1.26 snap). The garm binary is statically linked and garm version reports 1ad3680.
• CI integration run (garm) on the rebuilt rock.

Review focus

• The pin is the #926 merge commit, not upstream main. main has three more commits, two of which change scale set deletion behaviour; they're left for a separate bump.
• The new timeouts use GARM's defaults. The charm doesn't set the new statement_timeout_secs / lock_timeout_secs / connect_timeout_secs options, so no config change is needed.
• Besides #926, the pin brings in one more functional commit (scale set jobs are listed in the job listing again) and two webapp dependency bumps.

AI-generated message.

[Copilot]

🔵 Needs a closer look

It changes how PostgreSQL queries and runner cleanup behave in a production workload, and the integration test run that validates it is still pending.

1 open issue

• Correct changelog claims about runner cleanup and query timeouts · New

What changed in this PR

This PR bumps the pinned GARM workload commit in garm-rockcraft.yaml to the merge of cloudbase/garm#926. That change targets runners that get stuck, which the PR links to hung PostgreSQL queries and a database deadlock. The changelog gets a dated entry describing the user-facing effect.

Changes:

• Moves the GARM source pin to commit 1ad3680, and documents the reason in the pin comment.
• Adds query timeouts for PostgreSQL (migrations are exempt), cleanup for runners left in creating, and a fix for an instance and scale set update deadlock.
• Adds a 2026-10-25 changelog entry for the change.

File	Description
garm-rockcraft.yaml	Moves the GARM pin to the #926 merge commit and adds a comment explaining the fixes it brings in.
docs/​changelog.md	Adds a dated entry describing the stuck-runner fixes for users.

_{Review effort: Lite}