Error output not working on Postgres Node

[pdondin]

Bug Description

Using any node setting (select, update... or sql queries) - when using a specific error management parameter - will return no result (empty result object).

To Reproduce

1. Create a step with Postgres Node
2. Add a basic select that should result in something
3. Set the parameter "On Error" to "Continue (using error output)"
4. Run the node
5. Should result in empty result
6. Try with the parameter "On Error" to "Stop workflow"
7. Should result in the expected result

Expected behavior

Returning the same result

Operating System

Alpine Linux v3.20

n8n Version

1.62.5

Node.js Version

v20.18.0

Database

PostgreSQL

Execution mode

main (default)

[Joffcom]

Hey @pdondin,

We have created an internal ticket to look into this which we will be tracking as "GHC-321"

[leesilverman]

Also just encountered this in 1.62.5. This problem is breaking many of my automations.

I this workflow, Postgres1 will return an item, Postgres2 will not. The only difference is that Postgres2 has an error output.

{ "meta": { "instanceId": "56a3b6e808c84fdadd2d80585289500a7b27625c2792fc6c214518fe7ce0a003" }, "nodes": [ { "parameters": {}, "id": "7285ed37-2d8a-4d66-b51a-08569fa9b7bd", "name": "When clicking \"Execute Workflow\"", "type": "n8n-nodes-base.manualTrigger", "typeVersion": 1, "position": [ 780, 880 ] }, { "parameters": { "operation": "executeQuery", "query": "Select NOW()::date ;", "options": {} }, "id": "572a255f-9e7a-4a83-b07b-0e32848aaa7f", "name": "Postgres1", "type": "n8n-nodes-base.postgres", "typeVersion": 2.3, "position": [ 1160, 880 ], "retryOnFail": true, "credentials": { "postgres": { "id": "VEr7bxWUKIQlYdtb", "name": "Postgres Fulfillment Database" } } }, { "parameters": { "operation": "executeQuery", "query": "Select NOW()::date ;", "options": {} }, "id": "c3107b8a-2344-405e-bebc-ee35c66b4f0a", "name": "Postgres2", "type": "n8n-nodes-base.postgres", "typeVersion": 2.5, "position": [ 1160, 1100 ], "credentials": { "postgres": { "id": "VEr7bxWUKIQlYdtb", "name": "Postgres Fulfillment Database" } }, "onError": "continueErrorOutput" } ], "connections": { "When clicking \"Execute Workflow\"": { "main": [ [ { "node": "Postgres2", "type": "main", "index": 0 }, { "node": "Postgres1", "type": "main", "index": 0 } ] ] } }, "pinData": {} }

[pdondin]

@leesilverman Looks like it happens on any node using the additional error output parameter :/ ive seen several issues being raised for this

[leesilverman]

I haven't tested any module besides postgres. But I use the error output extensively, especially on HTTP module, so if this is a broader problem it's going to be tough to manually override.

Tried downgrading to 1.61.0 but then I couldn't log in. Something must have changed in the n8n admin database structure.

Tried downgrading to 1.62.0 but still encounter this problem.

Confirmed that in 1.62.0 if I remove the error output from postgres module it works as expected.

[netroy]

We are working on getting this fixed in #11238, and will push out a patch release as soon as this is merged.

[leesilverman]

Preliminary testing shows this working in 1.62.6. Thanks for pushing out a fix so quickly!

[pdondin]

Hey Netroy thanks a lot!!

Just curious, I'm using a docker image to deploy my instance, it looks like those version are not available through a public image.

What's your way of deploying it @leesilverman ?

EDIT : nvm I found the docker image

[janober]

Fix got released with n8n@1.62.6

[netroy]

@pdondin The images finished building only about 15 minutes ago. Maybe there is some replication delay. I can see the images on docker hub and on ghcr. Can you please try again?

[pdondin]

yes nvm, worked out a minute after, idk if this was related to a wrong URL or the image not being published

edited my comment :')