Introduction

Version 2.0.0 (and newer) of opEvents provides a new mechanism for expiring old data from the database. The purging is totally optional and no old data will be removed unless you explicitely configure opEvents to do so.

What can be purged?

opEvents can expire four different types of old data independently:

• Summary reports
• Events and Event Actions
• Raw Logs
• Archve Logs

Configuration

Your desired purging policy is defined by setting one or more of the following four configuration properties in conf/opCommon.nmis. Here is the commented example from install/opCommon.nmis:

'opevents' => {
  # lots of other directives...
  # how long should things be kept in the db? format NN[dhm]MM[hm]..., plain N means N minutes. 0 or undef means no purging.
  'opevents_reports_purge_older_than' => undef,     # covers summary reports
  'opevents_events_purge_older_than' => undef,      # covers events and actions
  'opevents_rawlogs_purge_older_than' => undef,     # covers raw logs
  'opevents_archivelogs_purge_older_than' => undef, # covers archive logs

The configuration is pretty straightforward:

1. no value, the value 0 or the special value undef means no purging whatsoever.
2. a purely numeric value is interpreted to mean "purge entries that are older than so many minutes".
3. the system understands combinations of the units "d", "h" and "m", in any order and without any delimiting spaces.
   For example, 31d12h means "purge data older than 31 days and 12 hours" (as does  12h31d).

Activation

The expiration of old data is performed by opeventsd.pl if and when it is started with the argument act=purge. You can also instruct it to only tell you how many entries a purge run would remove (without removing anything) by giving the arguments act=purge simulate=true.

By default the installer for opEvents 2.0 will create a suitable cron schedule in /etc/cron.d/opevents which triggers this action once weekly, but you can of course modify this to your liking.