Hello folks.
We had an instance a few times where our workflow jobs would start to get backed up to over 100 items stuck in the queue. Is there any type of alerting for this? I know we can log in and see the jobs and their status, but our users caught it before the admins did because we don't manually log in every day to look for this. We had some jobs that were stuck in the running status, and it caused a huge backlog.
Any ideas on how to get alerts on this?
Solved! Go to Solution.
I have a job that runs every 15 mins to pull relevant data from the MongoDB and store it in a SQL environment. Then, I have a SSIS job that queries the SQL environment (particularly, the completed datetimes of recently completed workflows). If the time between the max(completed datetime) and the time the SSIS job runs ever exceeds 30 minutes, then it emails our operations center and lets them know that the Alteryx Server is backed up and not completing jobs in timely fashion.
Thank you. Sounds like a good solution.
Sorry, one last question. When you say job that pulls MongoDB, do you use the MongoDB Connector BI tool? How do you pull the Mondo data(thur an Alteryx connector) or outside of that?
I use the MongoDB Input tool within Alteryx. I built my own version of the Alteyx Server Usage Report that incrementally loads new data that doesn't already exist in SQL. Taking this approach kept the run time of the ETL job very short (~10 seconds). The beauty of this approach is that your monitoring is reliant on the Server being performant. If the monitoring job gets queued for a long time, it won't update SQL, which triggers the alert from the SSIS job. It's a two-fold process, but it works for us. I also have tons of reporting dashboards built over the SQL sources, so there's added value to the integration.
You could very easily create the same alerting directly over the MongoDB by using the MongoDB BI Connector (relies on an enterprise version of MongoDB, which isn't included in the embedded configuration) or perhaps the Simba connector. There are other methods of connecting to Mongo, but I haven't really experimented with them.