Cloud is effectively down right now - publishing appends output

[Clinton_Wood]

This remains unresolved in our cloud workspace.

I duplicated the flow and still see the same failures:

• Tables filters receive: “Invalid input: expected string, received undefined”
• Hard-coded filter values work
• Dynamic mappings from trigger/code-step outputs intermittently resolve undefined/empty
• Code-step inputs mapped from prior Tables steps sometimes arrive empty
• Duplicating the flow did not repair it

Please confirm specifically:

1. Has PR #13571 been merged?
2. Has the fix been deployed to cloud.activepieces.com?
3. Has the cloud publish bug that appends/nests [‘output’] into tokens been fixed?
4. Are existing affected flows automatically repaired, or do they require migration/export-import/manual token repair?
5. Can your team inspect and repair the serialized references in my affected flow?

At this point the issue is blocking production workflow execution.

[Clinton_Wood]

Following up because this remains production-blocking. I love activepieces and am unable to do any work on my workflows because of this issue. Please help!

I still have no response on:

Stability Issue - Not finding previously stable content/inputs

A webhook-triggered workflow is intermittently failing in Tables find-records filters with: “Invalid input: expected string, received undefined” The webhook body contains a valid body.run_id. I added a Code step immediately after the trigger to normalize body.run_id into a plain string output. Downstream Tables steps mapped to that normalized output still intermittently fail as if the mapped value is undefined. This began after previously stable runs. Please investigate dynamic value resoluti…

The issue appears aligned with PR #13571:
“Data mapping between steps is broken on main for many existing flows.”

Please confirm:

1. Has PR #13571 been merged?
2. Has it been deployed to cloud.activepieces.com?
3. Does the fix repair existing affected flows, or only newly published flows?
4. Is the cloud publish bug that appends/nests [‘output’] into tokens resolved?
5. Should affected users avoid publishing until this is fixed?
6. Can engineering inspect my affected flow, since duplicating the flow did not repair it?

This is currently blocking production execution because dynamic mappings from webhook/code/table steps intermittently resolve as undefined/empty while hard-coded values work.

[abuaboud]

Hi @Clinton_Wood

There was low percentage of the user traffic was redirect to our canary servers by bug causing the migration to run multiple times, the team has published the fix yesterday for new flows, Today they were able to fix all old flows as well.

They should be working right now, and the engineering are currently monitoring the situation.

Sorry for that, the team is looking now how to prevent similar bugs in the future.

Thanks,
Mo.

[Clinton_Wood]

Thank you so much! I appreciate your response and your platform!