Waiting Executions do not work correctly when changing time zone

[directpromo]

Describe the bug
With the default time zone GMT 0, everything works correctly.

For time zones from GMT - 1 to GMT - 11 workflow freezes. In waiting status.

For time zones from GMT +1 to GMT +14 workflow goes to status successful within about 1 minute. Even if you need to wait for an hour or 5 minutes.

To Reproduce

1. create a workflow for waiting
2. set the time zone in the workflow settings
3. run
4. see the result in the executions
5. repeat several times with different time zones

Environment:

• OS: Ubuntu Linux 22.04
• n8n Version 1.27.2
• Database POSTGRES
• EXECUTIONS_MODE=regular

Additional context
The bug also occurs if you set the timezone in .env via:
GENERIC_TIMEZONE=your time zone
TZ=your time zone

[Joffcom]

Hey @directpromo,

Just to check is this changing the timezone for a workflow that is already waiting or is this setting a timezone that is not GMT 0 then setting a wait?

[directpromo]

Hi @Joffcom,

I don't quite understand the question.

I made a screencast. It shows this bug.

2024-02-05_n8n_wait_timezone.mp4

[Joffcom]

Hey @directpromo,

I have done a bit of testing and I managed to reproduce it as my database was still using the UTC timezone, I have not done a lot of testing around starting a new database with the correct timezone configuration but I do wonder if that is linked.

For now I have created an internal ticket which we will be tracking as PAY-1341.

[stwonary]

I am experiencing a similar issue with my workflows in n8n, particularly around the handling of different time zones. It seems there is a discrepancy in how workflows proceed or freeze based on the time zone settings, which mirrors the problem described.

To reproduce this issue, I've been using the following Docker Compose file:

version: '3.1'

services:
  n8n:
    image: n8nio/n8n
    ports:
      - "5678:5678"
    environment:
      - N8N_PROTOCOL=http
      - N8N_HOST=localhost
      - N8N_PORT=5678
      - NODE_ENV=production
      - WEBHOOK_TUNNEL_URL=http://localhost:5678/
      - DB_TYPE=postgresdb
      - DB_POSTGRESDB_DATABASE=n8n
      - DB_POSTGRESDB_HOST=postgres
      - DB_POSTGRESDB_PORT=5432
      - DB_POSTGRESDB_USER=n8n
      - DB_POSTGRESDB_PASSWORD=password
    depends_on:
      - postgres

  postgres:
    image: postgres:16
    environment:
      - POSTGRES_DB=n8n
      - POSTGRES_USER=n8n
      - POSTGRES_PASSWORD=password
    ports:
      - "5432:5432"
    volumes:
      - postgres-data:/var/lib/postgresql/data

volumes:
  postgres-data:

And in my workflows, I've observed that a 3-minute Wait node is incorrectly storing timestamps in the database 5 hours in the future, potentially due to the default GENERIC_TIMEZONE setting, which is America/New_York (GMT-5).

I am running this with n8n version 1.29.1, utilizing a PostgreSQL database. The issue persists regardless of setting the time zone in the workflow settings or via the .env file using GENERIC_TIMEZONE and TZ variables.

Current time:

Workflow waitTill in database:

[stwonary]

@Joffcom, I hope you find my findings useful. I have traced the bug back to version 1.25.0, and it may be related to this PR: #8340.

[Joffcom]

@stwonary I believe we have actually fixed this in #8578 which is available in 1.29.0 which you can access now.

[stwonary]

You are correct. After conducting a test, I can confirm that everything is functioning perfectly with version 1.29.1. It appears I mistakenly believed I was using version 1.29.1 as the latest release on Docker Hub, when in fact, I was actually using version 1.27.3.

Thank you for your response!

[Joffcom]

Hey @stwonary,

No worries I will get this one marked as closed and update everyone else that had this issue as well. Thanks for sticking with us during this 👍🏻