Alerting on no log entires

Hi All,

I am putting this under Logstash as I am reasonably sure (but not totally sure) I will need Logstash to achieve this.

We have around 170 servers all running the same application. On each of these servers there are about 5 log files we need to monitor. We need to trigger an alert if any **ONE **of these log files stops receiving data for a period of time (say 1 minute for example) as this indicates the application has failed.

I can get Filebeat running on them to monitor the files and send to Elasticsearch, but I can't get my head around what to do next (I'm a server and infrastructure guy not a developer :-))

I am thinking I will have to send logs to Logstash instead of ES, then use a pipeline to write each hostname/logfile combination to it's own index in ES and from there look at a watcher (or ML Job?) to alert if the number of lines in the file hasn't changed over an interval.

Any suggestions?


I think you can have filebeat write directly to elasticsearch. Use ML to monitor the number of lines for each hostname/logfile and generate alerts based on that.

It's been a while since I ran es/ML so I cannot provide detail.

This topic was automatically closed 28 days after the last reply. New replies are no longer allowed.