call workflow service - Knime desktop

[GQRanalytics]

Hi there knimers,

I have recently started using the call workflow service nodes, but I see a problem that need to understand how do you handle it.

The orchestator workflow calls the child, but if due to any reason the child fails, I cannot go into the child for troubleshoot, the main reason its because the data stays there for about 60 seconds and then its gone.

in the call workflow nodes I see some time parameters but all of them are grayout, and also noticed a flow variable named “keepFailingJobs“ as boolean, where Im able to send true, false, but with both values the result is the same.

does somebody know how to handle this, is there a way to keep the child workflow with data for troubleshooting?

[GQRanalytics]

Hi All, as an update on this, I found the workarround, in case is usefull for you as well.

As stated in my original post, the callee data will disappear after 60 secs of background execution.

However, if you already know wich of the calles will fail, what you need to to is the following:

1. Execute the caller workflow.
2. Inmediatelly after (while running) go and open the callee workflow.
3. You will see it running and when it fails the data stays there and you can do all necessary troubleshooting.
4. save your changes, close the workflow.
5. execute again the caller.

its kind of anoying, but seems its the way it works, in case any of you find another way of handling this, please share

[k10shetty1]

@GQRanalytics Thanks for sharing the workaround! What you’re seeing is expected when using Call Workflow Service from the KNIME AP (Desktop). The called workflow runs as a temporary background job, which is cleaned up shortly after execution, so the intermediate data disappears after a short time.

The option you mentioned, “Retain job on failure”, only applies to remote execution. This works when the workflow is executed on a remote executor such as KNIME Hub, where failed jobs can be kept and inspected for debugging.