After upgrading to 3.6.2, “End If” nodes are not being executed when their top branch is inactive and “Execute all executable nodes” or executing a downstream node is performed. Execution gets stopped with the node just prior to the “End If” node. Simply rerunning “Execute all executable nodes” or executing a downstream node again resumes execution and everything runs to completion. Any idea what is causing this?
Sry for a late response. Is there a possibility for you to share a workflow with us? We have encountered similar behavior and are working on it.
Is this issue fixed in the meantime? I am using version 3.7.1 and have the same kind of issue.
We are sorry you are experiencing this issue. Can you share a workflow because I couldn’t reproduce this exact behavior?
I will be at the KNIME Summit in Berlin. Would it be an option that I show it to someone there and set aside some time to go through the workflow?
yes this will work
Feel free to find me or come join us at our Demo Booth, there are always a lot of KNIMErs and we can check this together.
Hi @petra_benz !
Just to follow up on this one. Have you managed to solve it?
My workflow likes the following:
chunk loop start※A ->
try ->csv reader->catch->
check error in the table by filter error message->
empty table switch※B->
top->empty table switch-> …chunk loop start->…redshift…->loop end->end if->↓
bottom--------------------------------------------------------------------------------------------->end if※B->loop end※A
「end if※B」 is inactive, when 「empty table switch※B」's bottom branch is active, after clicking execute button again,everything runs to completion.
In the above workflow’s 5th line, I have added try/catch in the redshift SQL Executor:
～～～～～～～～～～～↑write error log↓
chunk loop start->…try->redshift ->catch …->loop end
If I remove the try/catch, the problem disappeared.
But we need the try/catch to handle the redshift exception, how can I solve the problem?
Thanks in advance.
welcome to KNIME Community Forum!
Not sure if this is solved or if there is a workaround for it but we can give it another go
So, although your workflow representation is pretty nice a print screen of it would be more helpful in order to build workflow that reproduces this kind of behavior.
Also is there a possibility you can share a workflow where this can be occurred? You can replace Redshift with SQLite and add some dummy data.
Thank you for the quick response.
About this problem, I think there is nothing to do with redshift , if there are try / catch nodes in the inactive branch, the “end if” node will be always inactive even the active branch finished.
because the workflow is developed for our customer who bought the KNIME Server license, if you want to check the workflow, we need to contact the customer to get permission to show the workflow , and then contact your company’s agency in our country.
No need for workflow then. I will try to reproduce it and get back to you.
Hi there @laughsmile,
Tried but failed in reproducing it. Can you take a look at workflow and modify it in order to have behavior you described?
Here is workflow:
Thank you for the reproducing.
I also tried but failed to reproduce it by using your workflow,
The real reason may be another one.
※ My workflow is very long.
Fortunately I find one workaround to solve my problem without removing the try / catch node.
beside my try / catch , there is a metanode, I move the nodes in the metanode to the main workflow, and remove the metanode, the problem is solved.
glad you found a workaround. There have been some issues with (Wrapped) Metanodes in Try/Catch sequence. Hopefully it will be resolved soon.
Hi all -
Test.knwf (94.2 KB) @ipazin you may find attached an example of the issue. Issue present when input gives empty table switch and takes empty path where try to manually configure a “dummy cell” so it wont affect the extraction of SAP information.
and welcome to KNIME Community!
I have taken a look at your workflow example and have seen you have Chunk Loop Start node in your Input Dates Sorting metanode (the upper one that stays inactive) and no matching loop end node. That is the reason why execution is stopped. Close your loop and workflow should run as expected.