fix(repo): pg-upgrade counts tables with a query PostgreSQL accepts (#1245) #1333

Merged

claude-bot merged 3 commits from fix/1245-pg-upgrade-count into main 2026-11-04 19:17:26 +00:00

Conversation 0 Commits 3 Files changed 2 +24 -1

claude-bot commented 2026-11-04 17:11:43 +00:00

Collaborator

Copy link

Found by the 4.7.0-rc1 rehearsal on dev (#1245), which is what the rehearsal is for.

What happened. make pg-upgrade stopped at step 4 with ERROR: collations are not supported by type integer. The count query ended order by 1 collate "C", and PostgreSQL reads 1 there as the integer literal, not as "the first column". The script failed safe as designed: the new volume was removed, the old volume was untouched, and the app was left stopped. Dev was back on 4.6.2 within about two minutes, with every service healthy and no data changed.

The fix. Order by table_name::text collate "C". Checked read-only against dev's live PostgreSQL 16 (115 tables counted) and against a throwaway PostgreSQL 18.

Why nothing caught it. The script's tests check the guard, the volumes and sh -n, but none ran the query on a database. A new test now pulls COUNT_SQL out of the script and runs it on the suite's own PostgreSQL. It passes with the fix (13 passed) and fails with the old query, with the same error dev saw. ruff is clean.

Next: merge, cut v4.7.0-rc2 (rc1's tag contains the broken script), and run the dev rehearsal again.

🤖 Generated with Claude Code

Found by the 4.7.0-rc1 rehearsal on dev (#1245), which is what the rehearsal is for. **What happened.** `make pg-upgrade` stopped at step 4 with `ERROR: collations are not supported by type integer`. The count query ended `order by 1 collate "C"`, and PostgreSQL reads `1` there as the integer literal, not as "the first column". The script failed safe as designed: the new volume was removed, the old volume was untouched, and the app was left stopped. **Dev was back on 4.6.2 within about two minutes, with every service healthy and no data changed.** **The fix.** Order by `table_name::text collate "C"`. Checked read-only against dev's live PostgreSQL 16 (115 tables counted) and against a throwaway PostgreSQL 18. **Why nothing caught it.** The script's tests check the guard, the volumes and `sh -n`, but none ran the query on a database. A new test now pulls `COUNT_SQL` out of the script and runs it on the suite's own PostgreSQL. It passes with the fix (13 passed) and fails with the old query, with the same error dev saw. ruff is clean. **Next:** merge, cut `v4.7.0-rc2` (rc1's tag contains the broken script), and run the dev rehearsal again. 🤖 Generated with [Claude Code](https://claude.com/claude-code)

claude-bot added 2 commits 2026-11-04 17:11:44 +00:00

fix(repo): pg-upgrade counts tables with a query PostgreSQL accepts (#1245) ... 5581d6d19d

The 4.7.0-rc1 rehearsal on dev stopped at step 4: 'order by 1 collate
"C"' reads 1 as an integer literal, which takes no collation, so the
old cluster's tables could not be counted. It failed safe (the new
volume was removed, the old one untouched) and dev went back to 4.6.2.
The query now orders by table_name::text collate "C", and a test runs
it on the suite's own PostgreSQL, which nothing did before.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

style: ruff format the pg-upgrade test (#1245) ... b9e682393d

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

claude-bot referenced this pull request 2026-11-04 17:11:47 +00:00

[Ops] Upgrade PostgreSQL 16 → 18: compose, client, CI, and a dump/restore upgrade for existing data #1245

claude-bot scheduled this pull request to auto merge when all checks succeed 2026-11-04 18:40:22 +00:00

rbrooks added 1 commit 2026-11-04 18:40:24 +00:00

ci: rerun after a network drop fetching the PGDG key (#1245) ... 4ad15ec59a

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

claude-bot merged commit 4d3bcd3815 into main 2026-11-04 19:17:26 +00:00

claude-bot referenced this pull request from a commit 2026-11-04 19:17:27 +00:00

Merge pull request 'fix(repo): pg-upgrade counts tables with a query PostgreSQL accepts (#1245)' (#1333)

rbrooks referenced this pull request from a commit 2026-11-04 19:18:37 +00:00

chore(release): v4.7.0-rc2

claude-bot referenced this pull request 2026-11-04 19:18:42 +00:00

chore(release): v4.7.0-rc2 #1334

claude-bot referenced this pull request 2026-11-04 20:44:59 +00:00

[Ops] Upgrade PostgreSQL 16 → 18: compose, client, CI, and a dump/restore upgrade for existing data #1245

claude-bot referenced this pull request 2026-11-05 19:55:12 +00:00

chore(deps): the open dependency updates and the axe 4.14 fixes, for 4.7.0 #1338

claude-bot referenced this pull request 2026-11-06 01:34:12 +00:00

CI: store the PGDG signing key in the repo, so image builds stop failing when its download drops #1346

claude-bot referenced this pull request 2026-11-06 08:06:50 +00:00

CI: the PGDG signing key is kept in the repo (#1346) #1356

Sign in to join this conversation.

Reviewers

No reviewers

Labels

Clear labels   

backend

Issue related to the backend

  

bot

Issue related to the bot

  

bug

Something is not working

  

duplicate

This issue or pull request already exists

  

enhancement

New feature

  

frontend

Issue related to the frontend

  

help wanted

Need some help

  

invalid

Something is wrong

  

question

More information is needed

  

recording

Issue related to the recording system

  

security

Security hardening / vulnerability work

  

wontfix

This won't be fixed

No labels

backend

bot

bug

duplicate

enhancement

frontend

help wanted

invalid

question

recording

security

wontfix

Milestone

Clear milestone

No items

No milestone

Projects

Clear projects

No items

No project

Assignees

Clear assignees

No assignees

2 participants

Notifications

Subscribe

Due date

The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set

Reference

rbrooks/Quest-Board!1333

No description provided.