20:00:34 <jokke_> #startmeeting log_wg
20:00:35 <openstack> Meeting started Wed Jul 15 20:00:34 2015 UTC and is due to finish in 60 minutes.  The chair is jokke_. Information about MeetBot at http://wiki.debian.org/MeetBot.
20:00:36 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
20:00:39 <openstack> The meeting name has been set to 'log_wg'
20:00:40 <Rockyg> o/
20:00:43 <jokke_> #topic attendance
20:00:47 <jokke_> #chair Rockyg
20:00:48 <openstack> Current chairs: Rockyg jokke_
20:02:08 <jokke_> courtecy call dhellmann johnthetubaguy
20:02:14 <jokke_> someone else we're missing?
20:02:27 <dhellmann> o/
20:02:33 <dhellmann> johnthetubaguy: bknudson
20:02:52 <bknudson> I'm at the keystone mid-cycle this week
20:02:53 <Rockyg> yah. jokke_, keep the chair.  I'm in two meetings at once.  Blech.
20:03:26 <jokke_> do we have specific agenda points or do we go for open right away?
20:03:49 <Rockyg> fyi.  when we get there, I have a question/proposal that dhellmann can help decide on how to go on
20:03:58 <nkrinner> i have something for open discussion
20:04:16 <jokke_> #topic open discussion
20:04:22 <jokke_> nkrinner: stage is yours
20:04:47 <nkrinner> i was wondering if anybody submitted a talk about logging for the tokyo summit?
20:05:02 <nkrinner> i might do so if none of the regular contributers did so
20:05:07 <jokke_> nkrinner: I submitted one ref this wg
20:05:15 <nkrinner> jokke_: ok
20:05:24 <jokke_> if you have something specific in mind, please do so
20:05:25 <nkrinner> then it is yours ;-)
20:05:45 <Rockyg> nkrinner, go and add one if you have an idea...
20:05:54 <nkrinner> i do not have anything special, but i looked at the last meetings and did not see it covered
20:06:32 <jokke_> nkrinner: fair enough
20:06:41 <nkrinner> ok, i might still submitt something. let's see. but it is good to see that there is a chance that this topic is raised at the summit
20:06:41 <jokke_> Rockyg: you had something in your mind
20:07:31 <Rockyg> Yeah.  Here's the idea.  I think the spaceholders for optional log fields is less controversial and may be easier to get happening this cycle
20:08:00 <Rockyg> I think it can be done strictly through oslo, but we still would want to put together a spec to get operators feedback
20:08:19 <dhellmann> is that the thing where we include a '-' for a field that would be blank?
20:08:30 <Rockyg> if we made it an option in global config, then it's backward compatible
20:08:34 <Rockyg> Yup
20:08:47 <dhellmann> ok, I believe we're doing that so if we're not then it's a bug and doesn't need a spec
20:08:53 <dhellmann> are we not doing that somewhere?
20:09:18 <Rockyg> cool.  If we are doing it, we need to socialize how to get it set better.
20:09:37 <Rockyg> I don't remember seeing it, but most of the docs are dev docs, not ops docs.
20:09:42 <dhellmann> it's not something that needs to be set, it's just how the code should be working -- are you seeing logs with blanks in them?
20:09:50 <dhellmann> operators do not need to do anything to enable the behavior
20:10:03 <jokke_> Rockyg: is this about the syslog style structured data or just the '-' on empty field?
20:10:20 <Rockyg> just the - on empty fields.
20:10:42 <Rockyg> the ops have complained that from project to project the number of fields change in the logs
20:11:31 <jokke_> ok ... I'll try to do review for that in glance logs this week
20:11:49 <jokke_> see how they behave now
20:12:07 <jokke_> dhellmann: is that something that has been there for a while or recent change?
20:12:10 <Rockyg> Thanks!
20:12:22 <dhellmann> we need someone to sit down and compare the logs to see how they're different
20:12:46 <dhellmann> I'm not going to have time to do that for a while, so if there's some other resource who can do that we can turn the results into bugs and fix them
20:13:25 <dhellmann> jokke_: it has been our intent for it to work that way for a while. I thought it already was.
20:13:34 <jokke_> ++ I think biggest differencies probably comes from swift and keystone
20:13:35 <Rockyg> great.  If worse comes to worst, I can get this done at the ops midcycle.
20:14:00 <dhellmann> jokke_: ok, the swift logs are out of bounds for talking about these sorts of things -- the swift team is going to have to address any complaints themselves, because they don't use the oslo logs
20:14:03 <jokke_> dhellmann: on, I'll verify and if it's not there I try to find the time to log a bug and submit a review for oslo.log
20:14:12 <dhellmann> so we need someone to look at the other projects
20:14:31 <dhellmann> jokke_: ok
20:14:48 <Rockyg> we can see if bknudson can look at keystone
20:15:04 <jokke_> dhellmann: that was exactly my point swift is not using oslo logs and keystone is on apache which might give bit different outputs as well
20:15:19 <bknudson> keystone can work on eventlet or apache
20:15:28 <dhellmann> jokke_: right, we're not going to change apache logs but if keystone logs somewhere else besides apache then we can control those logs
20:15:28 <bknudson> but eventlet server is deprecated
20:15:54 <dhellmann> and swift has some different requirements, so let's just leave that one alone for now -- the low hanging fruit are the projects that are all supposed to be using the same library
20:16:28 <Rockyg> Yes.  I'll see if I can get input for the various projects.
20:16:33 <dhellmann> we need to start expressing these desires more concretely
20:16:37 <jokke_> dhellmann: has anyone made check if any projects are still using the incubator code or has everyone moved to the actual oslo logs?
20:16:50 <dhellmann> "everything is different" is not actionable because as a developer I wouldn't know which ones to change -- what format do you *want*
20:17:05 <dhellmann> jokke_: I have not done that, I've been tied up with release stuff
20:17:20 <Rockyg> I think there is still a project or two using incubator.  They are supposed to be moving.
20:17:25 <jokke_> dhellmann: fair enough ...
20:17:28 <dhellmann> saying "project X does not log using the standard format" is actionable
20:17:56 <jokke_> yes, but not necessarily a bug ;)
20:18:08 <dhellmann> true, but the action might be to say "yes, that's by design"
20:18:43 <jokke_> tru
20:19:53 <jokke_> #action jokke will check that empty fields are replaced with '-' on standard logging
20:20:30 <Rockyg> I can double check on which projects are using oslo.log
20:20:57 <jokke_> #action Rockyg makes a list of projects that are still using incubated logging code
20:21:23 <dhellmann> great, those are good steps
20:22:14 <jokke_> anyone going to be in ops midcycle this time? Rockyg?
20:22:14 <dhellmann> putting the notes in an etherpad will help us track the changes, if any
20:22:20 <dhellmann> I'll be there
20:22:31 <Rockyg> Yup.  I'm there.
20:22:57 <Rockyg> so, maybe we can make an etherpad of info I should collect or questions to ask, or info to pass on?
20:23:44 <jokke_> cool ... yes and specially please probe around what are the feelings
20:24:10 <Rockyg> Will do.
20:24:23 <jokke_> I do not want to be in the situation where we have no-one on the logging sessions next summit because they feel it being pointless
20:24:28 <Rockyg> It's in August, so we have time to plan.
20:25:01 <jokke_> yeah I just saw first mention about it yesterday or today so it's fresh on my mind
20:25:03 <Rockyg> I'm still reinstalling stuff on my new laptop  for instance have git and gerrit but can't clone a repo yet...
20:25:58 <Rockyg> and I've got booth duty at OSCon next week (what company sends senior engineers to stand around a booth?!?)
20:26:06 <jokke_> has anyone been able to follow up on the situation around those 2 specs over past couple of weeks?
20:26:21 <jokke_> Rockyg: Huawei, I assume? :P
20:26:46 <Rockyg> xproject meeting talked about request ids again.  The log has the etherpad link.
20:27:55 <Rockyg> they are converging on a solution for clients
20:28:05 <jokke_> argh ;)
20:28:49 <Rockyg> It's either option 1 or 3.  They keep bouncing back and forth.  1 is in the x-http field
20:28:59 <Rockyg> or whatever.  the x-nnn
20:30:23 <jokke_> how that is going to be implemented on client? :P
20:30:56 <jokke_> I mean wasn't the whole point to drop that convcersation for now and just focus the client support for current model?
20:31:28 <Rockyg> It's how it gets sent to the client.  What form
20:32:50 <jokke_> ^^
20:33:15 <Rockyg> I don't think they are talking about how to generate at the moment.  But, I need to look at etherpad.
20:33:49 <jokke_> no worries ... sorry but I haven't had cycles lately to follow that up
20:34:07 <jokke_> bit too tied up with our next Helion release
20:34:27 <Rockyg> Fun :P
20:35:28 <jokke_> one more thing, I just collected few poits to the agenda etherpad for next week
20:35:50 <jokke_> #link https://etherpad.openstack.org/p/log-wg-meeting-agenda
20:35:55 <Rockyg> Another thing we as agroup should be doing is watching bugs against oslo.log and discussing if we should raise awareness on some
20:36:27 <Rockyg> I likely won't make next week's meeting.  It's lunch at OSCon and that's when the floor is most busy.
20:37:05 <jokke_> Rockyg: ok, i'll make sure I'll be here then
20:37:21 <Rockyg> Thanks
20:38:05 <dhellmann> I should be here
20:38:33 <jokke_> last call for other items, or should we call this half way done?
20:39:05 <Rockyg> Sounds good to me.  I'm baked.
20:39:40 <jokke_> ok thanks everyone, we continue next week!
20:39:45 <Rockyg> Thanks!
20:39:46 <nkrinner> cu
20:39:47 <jokke_> #endmeeting