I want to catch a running process if it doesn't finish in 2 minutes or exceeds duration of 95 percent of executions by more than 2 times, for instance.
Hello Alex, thanks for logging in to ask - let me see if I can help.
Are you familiar with Kibana Alerting? See the docs here:
The Alerting Rules can send you notifications based on different types of thresholds and evaluations from the data ingested into your cluster.
The nuance being the data must be written into Elasticsearch to query against it... If you are capturing the process duration time metric, you can query against that field in the index. Perhaps, you could use the Metric Threshold Alert? See docs below:
Let us know if this doesn't meet your needs, please!
I checked the links you sent. There is general info.
I created the following alerts with an email connector: Latency anomaly and Latency threshold. Seems, these alerts aren't designed for my case. I also don't understand alerts with queries using field names. I'm not sure which field I should use in this case. Would be great to see examples.
Could you please provide how-to tips of creating an alert for a long process duration?
Apache, Apache Lucene, Apache Hadoop, Hadoop, HDFS and the yellow elephant
logo are trademarks of the
Apache Software Foundation
in the United States and/or other countries.