world-postgres: startup recovery re-enqueues runs without a recorded namespace under the World's own prefix, not the registered flow handler's

Open
#5,442 0 comments 0 reactions 0 assignees View on GitHub

Maintainers usually reply within 1 day

Nobody has claimed this yet.

Assessment

Difficulty
3/5
Estimated time
1-2 days
Newbie friendliness
50/100
Issue type
Bug
Clarity
Clearly specified
Activity status
Active
Tech stack
postgresql, typescript
Domain
databases

Research direction

Start at the start() call in packages/world-postgres/src/index.ts, where the TODO points to this issue, and read reenqueueActiveRuns in @workflow/world to see how the fallback namespace is chosen. Find where the flow handler's in-process registration and its namespace are available before start(). Done when runs without a recorded namespace are re-enqueued under that handler's prefix, runs with a recorded namespace are unchanged, and the registration warning is still shown.

Written by the indexing model from the issue text.

Description

Summary

On start(), the Postgres World re-enqueues every active run it owns (reenqueueActiveRuns from @workflow/world, called from packages/world-postgres/src/index.ts). A run goes back to the queue namespace it recorded when it was created; a run that recorded none goes to the World's own namespace (the namespace option, or WORKFLOW_QUEUE_NAMESPACE in the World's process).

Since #5315 the World delivers the Workflow SDK's messages in process, to the flow handler the app registered. That handler is registered under the prefix the flow route was built with (WORKFLOW_QUEUE_NAMESPACE at build time). When that prefix differs from the World's own, the World warns once at registration and still delivers what the SDK publishes, but the runs it recovers at startup without a recorded namespace go to its own prefix, which no handler in the process takes: the runner claims the job and fails the delivery.

Intended fix

Recover such runs under the registered flow handler's prefix rather than the World's namespace option: when a flow handler (one carrying the in-process entry) is registered before start(), pass its namespace as the fallback to reenqueueActiveRuns. Runs that recorded a namespace are unaffected.

Current state

The limitation is noted in the Postgres World docs (Programmatic configuration) and in the registration warning, and a TODO at the start() call in packages/world-postgres/src/index.ts points here.

Related: #2978 covers the wider question of isolating applications that share one database; this issue is only about the fallback namespace of recovery inside one application.

Dominant language
TypeScript
Stars
2.5k
Forks
401
Avg merge
1d 21h
Merged PRs (30d)
201

Getting set up

  • No Dockerfile or Docker Compose file
  • Has a pull request template
  • No contributing guide

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

More from vercel/workflow

All issues in vercel/workflow

Similar issues

More TypeScript issues

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.