Tracking user command activities

[lee212]

We have observed openstack user can not launch new instances due to quota limit of vcpu and instances. However, our framework does not catch this type of events and won't be able to provide information about errors.

Example of error message:
OverLimit: Over limit (HTTP Quota exceeded: code=InstanceLimitExceeded (HTTP 413) (Request-ID: req-...-...)

What we have found is that nova-api.log file generates the log entries that we are looking for as a WSGI server.

What we are going to do next is:

• watching nova-api.log
• collecting event messages (general and error)
• making backups of log files
• pouring the collected information into cloud metrics database
• displaying the error frequency on the web or on cm tools or else
• displaying a frequency of general user requests
• Documenting all processes
  This way allows us to provide rate of user (command) errors in a certain period.

BTW, the title might be misleading to this subject though.