Alerting on Workflow Error - best practices?

Question

Hello all, I am trying to improve visibility and action on errors in scheduled workflows in the Gallery.

Currently we do one of two things when a scheduled workflow errors:

* Have the workflow send an email after "Run with Errors" to an escalation mailbox
* We regularly collect the results of the run by sifting through the Gallery's MongoDB log and display the results in a visualization platform

#1 is meant to drive calls to action while #2 is meant to help understand trends of failures over time.

We'd like to get #1 even more real time and out of email by connecting to a Slack webhook or a Jira POST request or the like. However it doesn't appear you can easily pass information out of Alteryx via the Workflow Event architecture.

Are there other best practices users have found success with to help escalate workflow errors?

Scott_Snowman · Answer

Thanks @danilang -- to be more specific, rather than simply the workflow erroring out (and not finishing its run) I'm looking at a way to tap into the detail in the Error Log available to the native "After Run with Errors" event.

I'll look into the CReW macro suite. Based on the description at http://www.chaosreignswithin.com/2014/09/blog-macro-pack-2014-q3-release.html it looks like it will do exactly what we are looking for in that it outputs the results of the error log.

Thank you!

danilang · Answer

Hi @Scott_Snowman

Any suggestion built using only Alteryx is going to fail if the workflow errors before the log reporting part has run.  There won't be any results.  One thing you can look into is using the CReW Runner macro.   This macro runs the workflow out-of-process so the main workflow can fail without affecting the log reporting part.  The macro has two outputs, one that returns the results if the workflow is successful and the other if the workflow fails.   After the workflow called by the macro has run, you can log the results to Jira or Slack using their respective APIs.  If the called workflow has failed, you can issue an alert by either email or any other way that an Alteryx workflow can trigger.

Dan

cmcclellan · Answer

You can get the workflow name from [Engine.WorkflowFileName] and the error log might be possible (I used to just get the latest file in the log directory because we were only running 1 workflow at a time and it was simple to work out).  I'm sure you could pull the error log from somewhere 🙂