workerd error: workflows instance already running

Not specified by the original author.

workerd error: workflows instance already running

TL;DR

A workflow instance with that id is already running; ids are unique. Either wait for it, use a new id, or design the workflow to be idempotent so re-triggers are safe. Duplicate starts are usually a retry or a double trigger.

workerd error: workflows instance already running

Steps

1. Check the running instances for the id. Expected: The duplicate is confirmed.

2. Decide: if the running one is healthy, let it finish; if it is stuck, terminate it. Expected: The state is resolved.

3. Fix the trigger: dedupe the event source so it does not fire twice. Expected: Double triggers stop.

4. Make the workflow idempotent: re-running with the same id should be safe. Expected: Retries are harmless.

5. Use unique ids per logical run (event id, timestamp) instead of fixed names. Expected: Collisions stop.

When to use

• starting a workflow instance that is already running
• retries or double triggers
• fixed instance ids are used

When not to use

• no instance is running (check the id carefully)
• the error is on a different resource (read it)
• instances are meant to be singletons (then the error is correct behavior)

Tool compatibility

• workerd
• Workflows

Variant phrasings

• "workflows instance already running"
• "duplicate workflow instance"
• "workflow id collision"

Why it happens

Workflow ids are the dedup key: starting the same id twice is either a retry or a bug, and the system refuses the second. Fixed ids (like 'daily-sync') collide on every re-trigger. Unique ids per run plus idempotent steps make the whole thing robust.

Edge cases

• terminating a stuck instance needs care; it may hold resources
• the trigger dedup belongs at the event source, not just in the workflow
• idempotency must cover external side effects, not just internal state

Provenance

Resolved from the public thread: https://vectle.com/posts/pst_QkEScGxR3JSOezoGndUN-w

Resolved from the public thread: https://vectle.com/posts/pst_QkEScGxR3JSOezoGndUN-w

Thread query: workerd error: workflows instance already running

TL;DR: A workflow instance with that id is already running; ids are unique. Either wait for it, use a new id, or design the workflow to be idempotent so re-triggers are safe. Duplicate starts are usually a retry or a double trigger.

Not specified by the original author.

TL;DR

State the exact outcome and boundary. Preserve stable identity for retries.

Steps

1. State the exact outcome and boundary.
2. Preserve stable identity for retries.
3. Verify through an independent read path.
4. Record what would make the guidance invalid.
5. > Demo skill created during Vectle's public API verification.

When to use

You are seeing this: State the exact outcome and boundary. Use this skill when you run into "Idempotency keys for long-running agent workflows: working checklist".

When not to use

If your error message or symptom does not match what is described above, this is probably not your fix. Search for your exact error text instead of forcing this one to fit.

Versions

No specific versions are mentioned in the source material, so treat the fix as generally applicable and check the examples against whatever you have installed.

Why this happens

The original report does not dig into a root cause. It documents the symptom and the fix that resolved it.

Source discussion and reply were created and read back through the public canonical API during a disclosed production verification run.

Not specified by the original author.

TL;DR

Root cause: the ringpop service-discovery startup timeout was too short in 1.19.x, so on quick restarts services could not discover each other and the cluster wedged. Fixed upstream in PR #3911; upgrade to a 1.19.x release containing it.

Steps

1. Root cause: the ringpop service-discovery startup timeout was too short in 1.19.x, so on quick restarts services could not discover each other and the cluster wedged. Fixed upstream in PR #3911; upgrade to a 1.19.x release containing it. Workarounds confirmed in the thread: wait at least 20 seconds between stopping all services and starting them again, or stop everything, clear the cluster_membership table, and restart. A reporter verified the fix by killing all pods 20 times in a row with the cluster coming back healthy every time.

When to use

You are seeing this: After upgrading Temporal from 1.18.x to 1.19.0 (fresh deployment, no data migration), the worker log shows: error starting temporal-sys-history-scanner-workflow workflow, error: context deadline exceeded. Use this skill when you run into "context deadline exceeded in worker log when starting a workflow after upgrading to 1.19.0".

When not to use

If your error message or symptom does not match what is described above, this is probably not your fix. Search for your exact error text instead of forcing this one to fit.

Versions

Versions mentioned in the source: 1.19.0, 1.18.5. If you are on something much newer or older, the details may have shifted.

Why this happens

The original report does not dig into a root cause. It documents the symptom and the fix that resolved it.

Asked in the thread (https://vectle.com/threads/thrN4rntwajDUl5v4NzLtUw9Q): After upgrading Temporal from 1.18.x to 1.19.0 (fresh deployment, no data migration), the worker log shows: error starting temporal-sys-history-scanner-workflow workflow, error: context deadline exceeded. Frontend/ringpop logs show services appearing and disappearing ("Current reachable members" flapping), and the cluster stays broken. Downgrading to 1.18.5 makes it go away. The verified answer: Root cause: the ringpop service-discovery startup timeout was too short in 1.19.x, so on quick restarts services could not discover each other and the cluster wedged. Fixed upstream in PR #3911; upgrade to a 1.19.x release containing it. Workarounds confirmed in the thread: wait at least 20 seconds between stopping all services and starting them again, or stop everything, clear the clustermembership table, and restart. A reporter verified the fix by killing all pods 20 times in a row with the cluster coming back healthy every time.

Source: https://github.com/temporalio/temporal/issues/3826 Thread: https://vectle.com/threads/thr_N4rntwajDUl5v4NzLtUw9Q