Open Stack

Hi Doug,

Thanks for the quick response and clarification on process.  Sounds like there are two key issues:

1) /dev/log is not portable
2) a code change was made that broke use_syslog=True

From the paper trail, it looks like the fix for use_syslog=True by reconnecting /dev/log was down-voted and abandoned.  I just resubmitted the change and am hopeful that we can treat these two issues separately, and get back to a working, if not portable, configuration setting.

Best regards,
John


> On Feb 26, 2015, at 9:55:57, Doug Hellmann <doug at doughellmann.com> wrote:
> 
> 
> 
> On Thu, Feb 26, 2015, at 11:22 AM, John Stanford wrote:
>> Hi,
>> 
>> I’m trying to understand the path of a series of issues I’ve been
>> interested in for some time.  Hopefully, someone can spend a few minutes
>> to educate me on the state of affairs, how/when incubation code might
>> make it to the main project.
>> 
>> The bugs in question are:
>> 
>> https://bugs.launchpad.net/oslo.log/+bug/1076466
>> <https://bugs.launchpad.net/oslo.log/+bug/1076466> - this is the bug that
>> identifies that restarting syslog breaks openstack APIs.  It appears to
>> be rooted in an eventlet issue, and the commentary suggests (rightly)
>> that a fix should be sought in the eventlet code rather than in oslo.log.
>> Unfortunately, it looks like no such fix is forthcoming.
>> https://bugs.launchpad.net/oslo.log/+bug/1385295
>> <https://bugs.launchpad.net/oslo.log/+bug/1385295> - this was a bug
>> introduced in Juno that completely broke syslog messaging from Openstack.
>> It has a long series of attempts at resolution and is recently active,
>> but to date, has no fix has been accepted.
>> Finally, in oslo-incubator, there is an accepted and merged change to
>> revert the commit that broke syslog: 
>> https://review.openstack.org/#/c/133638/
>> <https://review.openstack.org/#/c/133638/>
>> 
>> My questions are:
>> Why wasn’t the commit that closes 1385295 by reverting the original
>> change submitted against oslo.log?
> 
> It may have had to do with timing. If no one was using the library, yet,
> we may have decided to make a different set of changes there to support
> both behaviors. It looks like that work stalled.
> 
> This change looks like it is related:
> https://review.openstack.org/#/c/136842/ but it doesn't directly address
> the issue of logging to /dev/log instead of the daemon.
> 
>> Is it possible to make Juno work today with use_syslog=True, or is Juno a
>> lost cause in that respect?
> 
> Are you asking whether the code already works, or whether patches would
> be accepted to make it work?
> 
>> How/when does oslo-incubator code get merged to the main oslo codebase? 
> 
> Typically it does not. The incubator is for code *before* it becomes a
> library. After graduation, the incubated version is meant to be deleted.
> The log library graduation got stuck on some other work early this
> cycle, and so that part of the process was delayed. For more on how the
> incubator works, see
> http://specs.openstack.org/openstack/oslo-specs/specs/policy/incubator.html
> 
>> Specifically, when could we expect to see the 133638 commit patched into
>> oslo.log?
> 
> I don't see a patch submitted to make that change.
> 
> Doug
> 
>> 
>> Thanks in advance...
>> 
>> Regards,
>> John    
>> 
>> 
>> 
>> __________________________________________________________________________
>> OpenStack Development Mailing List (not for usage questions)
>> Unsubscribe:
>> OpenStack-dev-request at lists.openstack.org?subject:unsubscribe
>> http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
> 
> __________________________________________________________________________
> OpenStack Development Mailing List (not for usage questions)
> Unsubscribe: OpenStack-dev-request at lists.openstack.org?subject:unsubscribe
> http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev