Open Stack

The main motivation for this over something like rsyslog is realtime integration with external systems.
If this information is available via notifications, someone can turn of the amqp notification driver, and have external monitoring, metering, usage, etc systems use that data.

(at the same time, if you just want to log, and don't need the queueing, etc from amqp, , you can just turn on the logging notification driver, which will output to syslog. (it would be nice to have the log notification driver support structured syslog too, eventually) )

From: Joe Gordon <jogo at cloudscaling.com<mailto:jogo at cloudscaling.com>>
Reply-To: OpenStack Development Mailing List <openstack-dev at lists.openstack.org<mailto:openstack-dev at lists.openstack.org>>
Date: Monday, May 6, 2013 12:59 PM
To: OpenStack Development Mailing List <openstack-dev at lists.openstack.org<mailto:openstack-dev at lists.openstack.org>>
Subject: Re: [openstack-dev] Unified Notifications ...




On Thu, May 2, 2013 at 12:16 PM, Sandy Walsh <sandy.walsh at rackspace.com<mailto:sandy.walsh at rackspace.com>> wrote:
Hey y'all!

At the summit we had a session about unifying the various ways we record
important events within OpenStack (Nova specifically).

Currently we have the following approaches:
1. The event notification system,
2. The InstanceFault DB table,
3. The ServerAction DB table
4. Exception handling decorators,
5. and, to a lesser extent, logging

With five different ways of doing things not all important events are
being reported to all channels.

We did come up with a plan-of-attack. Specifically:

1. Drop InstanceFault and roll it into ServerAction
2. Change any logging messages greater than DEBUG (like INFO and ERROR)
into Notifications.
3. Default the Logging Notification Driver to ON, so we don't lose any
existing functionality. Users can opt to enable the AMQP driver in order
to expose these notifications to external systems.

What is the motivation for this over, a tool that collects log files, like rsyslog?


The etherpad is here: https://etherpad.openstack.org/unified-notifications

Also, by using the notification mechanism we have a route into things
like Structured Logging (via json) vs the unstructured strings we have
today. There is also a routing notification driver which permits you to
only send certain messages to other notification drivers.

Debug messages would continue to be used as they are today.

Before we start down this road I just wanted to get a taste of the
sentiment out there.

Thanks!
-S

_______________________________________________
OpenStack-dev mailing list
OpenStack-dev at lists.openstack.org<mailto:OpenStack-dev at lists.openstack.org>
http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstack.org/pipermail/openstack-dev/attachments/20130506/ebd9004f/attachment.html>