Skip to content

Troubleshooting topology already exist on cluster

Why do that

When you abruptly restart some processes of your punch platform (production or standalone), you may have an inconsistency between what is effectively running in the various processing engines, in particular the storm cluster, and what is known to be running by the punch administration store.

If you see that error; it means a punchline is running whereas the punch admin thinks it is stopped.

What to do

  • Check the status of your channel:
channelctl status --channel <channel>
  • Check the name of the target cluster in the channel_structure.json (located in tenants/<tenant>/channels/<channel>/)
  • Go to the Web Interface of the stream processing cluster (for example the storm UI).
  • Kill the running punchline(s) of your off channel. Storm and spark UIs both provide a 'kill' button.
  • restart your channel