Retry failing nodes with wait (Try/Catch + loop) in KNIME 5.2.5

KNIME Analytics Platform
1

Hi everyone,

I’m using KNIME 5.2.5 and have a workflow that intermittently fails in two nodes:

  • Create DB Table (DB SQL Executor / Hive)

  • Call Workflow Service (custom step: SSH2GBucket)

image
image947×335 12.5 KB

I’d like to implement a retry mechanism:

  • Retry the failing node up to 10 times

  • Wait 5 minutes between retries

  • Continue as soon as the node succeeds

  • Fail the workflow only after all 10 attempts fail

I tried using Try (Variable Ports) / Catch Errors (Variable Ports) with a Wait… node, but I couldn’t figure out how to:

  • Loop back to retry

  • Count attempts

  • Exit early on success

What’s the recommended pattern in KNIME for this?
Should this be done with a Variable Condition Loop + Try/Catch, or is there a cleaner/best‑practice approach?

Any guidance or example workflows would be really appreciated. Thanks!

2

you will need a loop (variable loop) start with a variable condition loop end, within you can count your iterations, define wait times and when to abort.

within you add your try & catch and within that your actual task or failing parts.

see also here 250425_Retry Database Connection – KNIME Community Hub this might guide you

given you havent detailed why your nodes fail, maybe its already sufficient to increase the timeout limits? otherwise, Knime is a bad place if you want to have easy retries. depending on your case, wrapping the action into a java snippet with a quick LLM generated retry pattern might proof better.

(imagine Knime would finally add something like a retry-loop construct build in)

2 Likes
3

@m_rahman here you have another example of looping until a job is finished:

Medium: “Loop until a Job is Finished”