mirror of
https://github.com/paperclipai/paperclip.git
synced 2026-10-21 00:05:49 +02:00
18e9ca7277
## Thinking Path
> - Paperclip is the open source app people use to manage AI agents for
work
> - Execution workspaces have a close-readiness check. The board uses it
before it archives or cleans up a workspace. The terminal-workspace
reaper and workspace hydration use the same delivery assessment.
> - The delivery assessment reads the source issue tree and the pull
request work products of the source issue.
> - A pooled database connection can become stale. Then postgres.js
rejects the query with `write CONNECTION_CLOSED`.
> - The issue-tree read had no retry. One stale socket failed the full
close-readiness request with a 500.
> - The database client does not replay disconnects globally, because a
write can commit before the disconnect (#14773). Callers that know a
lookup is read-only opt in with `retryIdempotentDatabaseOperation`
(#14873, #15248, #15300).
> - This pull request wraps the two read-only delivery lookups in that
helper.
> - The benefit is that close readiness, hydration and the reaper
recover from one stale pooled socket.
## Linked Issues or Issue Description
Refs #14773, #14873, #15248, #15300 (same retry pattern for other
read-only lookups).
**What happened?**
A hosted staging server returned an error for an execution workspace
close-readiness request. The issue-tree query in the delivery assessment
failed with `write CONNECTION_CLOSED` through a transaction-mode
database pooler.
**Expected behavior**
A read-only lookup recovers after one transient pooled disconnect.
Persistent failures still propagate.
**Steps to reproduce**
1. Use a pooled PostgreSQL endpoint that closes idle client sockets.
2. Let a pooled connection go stale.
3. Request close readiness for an execution workspace that has a source
issue.
4. The first query on the stale socket fails and the request fails.
**Paperclip version or commit**
`master` at 9966f31.
**Relevant logs or output**
`Error: Failed query: select "id", "status", "completed_at",
"cancelled_at", "updated_at" from "issues" where ... WITH RECURSIVE
issue_tree ...` caused by `write CONNECTION_CLOSED`.
## What Changed
- `listWorkspaceIssueTree` now runs inside
`retryIdempotentDatabaseOperation`. It is a plain recursive SELECT.
- `listDeliveryPullRequestProducts` now runs inside
`retryIdempotentDatabaseOperation`. It is a plain SELECT.
- Both functions capture `sourceIssueId` before the retry callback. Each
attempt builds a new query.
- Two regression tests in the embedded PostgreSQL `getCloseReadiness`
suite:
- One transient disconnect on each delivery read gives the same
readiness result as a healthy database.
- Three disconnects on the issue-tree read propagate the last error. The
pull request read does not run.
## Verification
- `pnpm exec vitest run
src/__tests__/execution-workspaces-service.test.ts` (in `server/`):
80/80 tests pass.
- The two new tests fail without the service change. The first fails
with `write CONNECTION_CLOSED`. The second records one attempt, not
three.
- I did not run the full server typecheck or the full suite locally. The
sandbox has a 4 GB memory limit and the native `tsc` process was killed.
CI runs both.
## Risks
- Low risk. The change wraps two read-only SELECTs. It does not wrap
writes or transactions.
- A persistent outage adds up to about 150 ms of retry pauses before the
error propagates.
> For core feature work, check [`ROADMAP.md`](ROADMAP.md) first and
discuss it in `#dev` before opening the PR. Feature PRs that overlap
with planned core work may need to be redirected — check the roadmap
first. See `CONTRIBUTING.md`.
## Model Used
- Anthropic Claude Opus 5.5 (`claude-opus-5-5`) in Claude Code, used as
an agent with tool use (shell, code editing, test execution) and
extended reasoning.
## Checklist
- [x] I have included a thinking path that traces from project context
to this change
- [x] I have specified the model used (with version and capability
details)
- [x] I have checked ROADMAP.md and confirmed this PR does not duplicate
planned core work
- [x] I have searched GitHub for duplicate or related PRs and linked
them above
- [x] I have either (a) linked existing issues with `Fixes: #` / `Closes
#` / `Refs #` OR (b) described the issue in-PR following the relevant
issue template
- [x] I have not referenced internal/instance-local Paperclip issues or
links (only public GitHub `#NNN` / `github.com/paperclipai/paperclip`
URLs)
- [x] My branch name describes the change (e.g. `docs/...`, `fix/...`)
and contains no internal Paperclip ticket id or instance-derived details
- [x] I have run tests locally and they pass
- [x] I have added or updated tests where applicable
- [x] I have updated relevant documentation to reflect my changes
- [x] I have considered and documented any risks above
- [ ] All Paperclip CI gates are green
- [ ] Greptile is 5/5 with no open P2s, recommendations, or follow-ups
- [x] I will address all Greptile and reviewer comments before
requesting merge
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-authored-by: devinfoley <139239+devinfoley@users.noreply.github.com>
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>