This page contains information to help setup an Evergreen availability monitoring system to monitor your Evergreen system.

Evergreen Availability Monitoring, Evergreen 2014 Conference, Michael Tate, Equinox Software - Focuses on Nagios but the general principles apply to all monitoring systems.

Equinox Evergreen Monitoring Scripts

Ideas on what to monitor to know how your Evergreen system is performing. This is roughly ordered from the top of the software stack to the bottom.

Monitor the status of your load balancer. Is it currently running, are all the backends considered active. How many redirects a minute is it handling.

1. Is the apache2 process running?
2. How many apache2 processes are running compared with the max allowed?
3. How much memory are the apache2 processes using?
4. Are there any runaway apache2 processes using too much CPU?
5. Are the http/https/websockets ports open and accessible?
6. How many requests per second? Bytes sent/received?
7. How many errors of various types were found in the logs?

Monitoring the response time for various web page loads will give you a good glimpse of how your system is performing for users.

Some ideas of areas to monitor.

• Main Opac Page
• Record Detail Page
• Search Results using various options
• Patron Login / account summary
• UNAPI results
• HTTP Opensrf gateway requests

1. Is the SIP Server process running?
2. How many processes are running compared with the max number allowed?
3. Is the SIP port open(TCP 6001) and responding to requests?

1. Is the Z39.50 Server process running?
2. How many processes are running compared with the max number allowed?
3. Is the Z39.50 port open(TCP 210) and responding to requests?

1. Are the opensrf router processes running?
2. Is there one listener process and at least one drone process for each service enabled?
3. Are the number of drone processes close to the max_children settings for that service?

1. Are there any stale action_trigger_runner.pl lock files?
2. How many pending events are there compared with your libraries normal level of pending events.