Log as ERROR only for conditions requiring operator attention

Registered by Maru Newby

There is a tempest blueprint (https://blueprints.launchpad.net/tempest/+spec/fail-gate-on-log-errors) that proposes to start failing gate jobs if ERROR appears in the logs unexpectedly. There is a whitelist so that expected error conditions for a given project don't trigger failure, but at present Neutron logs so many things as ERROR that whitelisting is not feasible. A cleanup effort is required to reduce the incidence of logging as ERROR to cases that would conceivably require operator attention. Ideally the number of such conditions would be small enough that whitelisting wouldn't mask actual problems.

Tempest is going to start outputting all ERROR log lines in the gate jobs for Neutron so we can start eliminating the unnecessary logging to the point where whitelisting is possible. Once we're satisfied with the whitelisting, Tempest can start failing jobs that produce errors that aren't whitelisted.

Blueprint information

Status:
Complete
Approver:
None
Priority:
Undefined
Drafter:
None
Direction:
Needs approval
Assignee:
None
Definition:
Obsolete
Series goal:
None
Implementation:
Unknown
Milestone target:
None
Completed by
Armando Migliaccio

Related branches

Sprints

Whiteboard

This was scrapped by Tempest long time ago (today is Oct 2015).

Tempest maintains a list of process names ignored for the purpose of log checking here: https://github.com/openstack/tempest/blob/master/tools/check_logs.py#L35

<mangelajo> I think I can automate this indexing passed logs, and our source code, and then automatically looking for fragments of the errors in our source code, and changing any .error into .warning.

(?)

Work Items

This blueprint contains Public information 
Everyone can see this information.