FFAStrans forum

Hi,

I have a workflow that has a conditional fail on if a watermark.txt file doesn't exist.

If this workflow fails, it's set to retry but this retry and even a manual retry of the job from the webinterface doesn't seem to work.

I think I have traced this down to the %s_recursed_path% failing to populate when it's initiated on anything other than an original watch folder picking the file up.

Is this a known limitation? Any ideas on how to ensure that variables get re-populated when retrying a job from the status monitor?

Thank you,

Neil

Hi kimble,

Resubmit in the web interface is not the same as retry in the status monitor. Resubmit actually creates a whole new job with no prior knowledge of the initial job. That enables you to change your workflow and try the source once more with updated configuration.
Retry from the status monitor will retry the exact same job with exact same configuration. In this scenario %s_recursed_path% will work as expected but not in the first where you resubmit.
We will try to make this distinction more clear and also enable the option to retry from the web interface.

Thanks for reporting!

-steinar

Ok, this is strange as I am seeing the same behavior's in both web and client. It might be something specific to this workflow.

For some reason, if this job fails on the "Watermark Existence Check" regex check for an empty or missing string, it will never pass again. As if it's not reflowing through the populate vars. When I looked in the log I can see it fails to populate the %s_recursed_path% this time through. Highlighted in red here.

{
"created": "2023-10-17T15:50:09.537+01:00",
"node": {
"id": "20231010-1044-5303-802c-f18d4d7f69d5",
"type": "op_populate"
},
"split_id": "1-0-0",
"event": "pre_evaluation",
"type": "trace",
"context": "node execution",
"host": "redacted",
"data": {
"name": "s_my_toplevel_path",
"data": "_MyRegExpExt(\"\",\"^(?:[^\\\\]*\\\\)?(.*?)(?:\\\\|$)\")"
}
},
{
"created": "2023-10-17T15:50:09.553+01:00",
"node": {
"id": "20231010-1044-5303-802c-f18d4d7f69d5",
"type": "op_populate"
},
"split_id": "1-0-0",
"event": "populate_variable",
"type": "trace",
"context": "node execution",
"host": "redacted",
"data": {
"name": "s_my_toplevel_path",
"type": "string",
"data": ""
}
},

Interestingly, I have added a conditional node before the populate variable node and this has fixed the issue so perhaps this is related to the populate variable node being straight after a watch folder.

Sorry kimble, i downloaded your wf but didnt find the time to work into it.
Interesting solution, could you maybe share the working version too? It kind of sounds a little buggy to me...