Inconsistent timestamp formats in UDP messages

Bug #1370607 reported by Daniele Venzano
6
This bug affects 1 person
Affects Status Importance Assigned to Milestone
Ceilometer
Invalid
Undecided
Unassigned

Bug Description

The messages generated by cinder-volume-usage-audit and neutron-metering-agent contain timestamps in a different format than the other messages received through UDP from ceilometer-agent-notification. This creates unnecessary troubles for whoever is trying to decode the messages and do something useful with them.

I particular, up to now, I found out about these fields:
- timestamp in bandwidth, volume and volume.size messages
- resource_metadata.created_at in volume and volume.size messages
- resource_metadata.launched_at in volume and volume.size messages
- resource_metadata.audit_period_* in volume and volume.size messages

They contain UTC dates (I hope), but there is no Z at the end, and they contain a space instead of a T between date and time. In short, they are not in ISO8601 as the timestamps in the other messages. I found out about them because elasticsearch tries to parse them and fails, throwing away the message.

I don't know if it is appropriate to file the bug here, or if the issue comes from neutron/cinder.

Revision history for this message
Nejc Saje (nejc-saje) wrote :

The issue is with Cinder & Neutron, namely in files https://github.com/openstack/neutron/blob/master/neutron/services/metering/agents/metering_agent.py and https://github.com/openstack/cinder/blob/master/bin/cinder-volume-usage-audit , which are both part of their respective projects. Can you please file bugs against those projects instead?

Thanks!

Changed in ceilometer:
status: New → Invalid
To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.