Solved: Apigee log file keywords to identify issue

rajeshkr · 05-22-2018 05:00 AM

Hi ,

we are having apigee on premises where we have monitoring system which will be checking the log files for any issue and generates the alerts. In order to generate alert it checks for the set of keywords / Strings matching in the log files. As per apigee documentation i see the file path's to the log files and only one example with the keyword "NoHeartBeat detected from/<MP_IP>:<PORT>MarkDown"

https://docs.apigee.com/private-cloud/v4.18.01/what-monitor

but i don't see any place in docs for the list of keywords to check in logs .Can you anyone help us here to give us the doc which has this info or set of keyword we need to monitor.

rmishra

Parsing Log files to generate unavailability alerts is generally a bad idea. Watching for keywords is a slippery slope because there is no contractual guarantee of the words remaining the same across upgrades or in all scenarios errors occur. Use Log files for debugging/triaging the issue(by feeding them to a log aggregator)

You should resort to log file based alerting for outage ONLY when you do not have better alternatives available.

In the case of Apigee, you have superior options. You can monitor the system over JMX , HTTP and OS Level. When used together (and correctly) they provide deeper and more reliable insights into the system.

For e.g. rely on configuring http://MP_IP:port/v1/servers/self/up in your monitoring system and listen for heartbeats to determine if the Message Processor is available.Your monitoring system can be configured to poll this URL every x seconds and report an outage only after y failed attempts. In my opinion, that would be a better way of approaching this problem rather than relying on log files.

View solution in original post

rmishra

Parsing Log files to generate unavailability alerts is generally a bad idea. Watching for keywords is a slippery slope because there is no contractual guarantee of the words remaining the same across upgrades or in all scenarios errors occur. Use Log files for debugging/triaging the issue(by feeding them to a log aggregator)

You should resort to log file based alerting for outage ONLY when you do not have better alternatives available.

In the case of Apigee, you have superior options. You can monitor the system over JMX , HTTP and OS Level. When used together (and correctly) they provide deeper and more reliable insights into the system.

For e.g. rely on configuring http://MP_IP:port/v1/servers/self/up in your monitoring system and listen for heartbeats to determine if the Message Processor is available.Your monitoring system can be configured to poll this URL every x seconds and report an outage only after y failed attempts. In my opinion, that would be a better way of approaching this problem rather than relying on log files.

rajeshkr

Thanks for your reply