Perl 6 - the future is here, just unevenly distributed

IRC log for #puppet-openstack, 2016-10-18

| Channels | #puppet-openstack index | Today | | Search | Google Search | Plain-Text | summary

All times shown according to UTC.

Time Nick Message
00:01 markvoelker joined #puppet-openstack
00:08 hogepodge joined #puppet-openstack
00:11 xarses joined #puppet-openstack
00:21 hogepodge joined #puppet-openstack
00:29 yolanda joined #puppet-openstack
00:53 ducttape_ joined #puppet-openstack
00:55 _ducttape_ joined #puppet-openstack
01:09 michapma_alt joined #puppet-openstack
01:23 yolanda joined #puppet-openstack
01:35 hogepodge joined #puppet-openstack
02:03 markvoelker joined #puppet-openstack
02:23 AndyWojo joined #puppet-openstack
02:30 johnmilton joined #puppet-openstack
03:04 hogepodge joined #puppet-openstack
04:04 markvoelker joined #puppet-openstack
04:33 spredzy_ joined #puppet-openstack
04:41 markvoelker joined #puppet-openstack
04:42 markvoelker_ joined #puppet-openstack
04:56 rwsu joined #puppet-openstack
05:13 markvoelker joined #puppet-openstack
05:14 markvoelker_ joined #puppet-openstack
05:18 jaosorior joined #puppet-openstack
05:57 skramaja joined #puppet-openstack
06:01 rcernin joined #puppet-openstack
06:13 jtomasek_ joined #puppet-openstack
06:18 openstackgerrit Dan Trainor proposed openstack/puppet-tripleo: Enable communication between UI and the Undercloud by making HAProxy proxy for the UI  https://review.openstack.org/386817
06:42 mkarpin1 joined #puppet-openstack
06:56 fxpester joined #puppet-openstack
07:01 dgurtner joined #puppet-openstack
07:01 tdb joined #puppet-openstack
07:03 tesseract joined #puppet-openstack
07:05 ccamacho joined #puppet-openstack
07:08 shardy joined #puppet-openstack
07:19 clayton joined #puppet-openstack
07:24 hjensas joined #puppet-openstack
07:24 hjensas joined #puppet-openstack
07:39 jpich joined #puppet-openstack
07:40 hogepodge joined #puppet-openstack
07:46 Kvisle My proposal got a minus score due to the CI problems yesterday, do I have to do anything to get that fixed?  Do I need to resubmit it? https://review.openstack.org/#/c/387263/
07:52 jaosorior Kvisle: rechecked your proposal, it was an un-related issue.
07:53 Kvisle jaosorior: thanks - unrelated to what exactly?
07:54 jaosorior Kvisle: meaning, your patch didn't cause the error in CI
07:54 Kvisle thanks :)
07:54 jaosorior Kvisle: If you want to recheck your patches, it's a matter of doing a comment with the content 'recheck' in it. I already did so for your patch.
07:55 Kvisle ah, didn't know that (this is my first proposal)
07:55 jaosorior Kvisle: no biggie, that's why I mentioned it to you, hope it's useful.
07:56 Kvisle what does recheck trigger, exactly? does it trigger new CI runs, or is it simply a flag that reviewers go by?
07:56 jaosorior Kvisle: it triggers a new CI run
07:56 openstackgerrit Michele Baldessari proposed openstack/puppet-tripleo: Fix broken rabbitmqctl commands when using ipv6  https://review.openstack.org/386921
07:57 Kvisle cool
08:02 asilenkov joined #puppet-openstack
08:03 fxpester joined #puppet-openstack
08:05 openstackgerrit joined #puppet-openstack
08:08 openstackgerrit Merged openstack/puppet-tempest: Enable Keystone v3 support for providers  https://review.openstack.org/387365
08:16 paramite joined #puppet-openstack
08:19 tdb joined #puppet-openstack
08:48 derekh joined #puppet-openstack
08:57 markvoelker joined #puppet-openstack
09:04 openstackgerrit joined #puppet-openstack
09:06 rmart04 joined #puppet-openstack
09:09 Kvisle jaosorior: should I report the (new) failed CI run somewhere? safe to say it's not related to anything I did.
09:12 jaosorior Kvisle: might want to ask around if someone is familiar with that failure
09:16 shardy joined #puppet-openstack
09:18 openstackgerrit Juan Antonio Osorio Robles proposed openstack/puppet-openstacklib: Add tls options to mysql user creation  https://review.openstack.org/387880
09:34 gfidente joined #puppet-openstack
09:38 markvoelker joined #puppet-openstack
09:39 hjensas joined #puppet-openstack
09:42 jaosorior joined #puppet-openstack
09:51 openstackgerrit Merged openstack/puppet-tripleo: Remove faulty migration logic to stop nova-api  https://review.openstack.org/387189
10:09 richm joined #puppet-openstack
10:21 EmilienM hi
10:28 openstackgerrit Merged openstack/puppet-neutron: Add option to enable Neutron's SSL middleware  https://review.openstack.org/387174
10:29 openstackgerrit Babu Shanmugam proposed openstack/puppet-tripleo: WIP - Fix bug in ovn plugin  https://review.openstack.org/387366
10:52 degorenko joined #puppet-openstack
10:53 degorenko hi :)
10:53 degorenko_ joined #puppet-openstack
10:55 johnmilton joined #puppet-openstack
11:07 spredzy joined #puppet-openstack
11:10 markvoelker joined #puppet-openstack
11:14 dprince joined #puppet-openstack
11:17 EmilienM degorenko, iurygregory, mwhahaha: fyi I'm not sure to attend whole puppet meeting, I need to go at airport and take my flight to BCN
11:17 openstackgerrit Steven Hardy proposed openstack/puppet-tripleo: WIP Split ovn plugin and northd configuration  https://review.openstack.org/387939
11:17 Kvisle so my proposal has +1 code review point, +1 ci point ... it still needs workflow; should I try to find someone who can make final approval, or will someone eventually get around to it?
11:18 Kvisle ( https://review.openstack.org/#/c/387263/ )
11:18 degorenko EmilienM, ack :)
11:47 openstackgerrit Emilien Macchi proposed openstack/puppet-openstack-integration: Add puppet-qdr and puppet-cloudkitty  https://review.openstack.org/387970
11:52 openstackgerrit Emilien Macchi proposed openstack/puppet-keystone: Introduce memcached_ips and memcached_port  https://review.openstack.org/387974
11:53 EmilienM iurygregory:  ^
11:55 dprince joined #puppet-openstack
11:55 openstackgerrit Emilien Macchi proposed openstack/puppet-keystone: Introduce memcached_ips and memcached_port  https://review.openstack.org/387974
11:57 openstackgerrit Dan Prince proposed openstack/puppet-tripleo: Set memcached_servers for nova API  https://review.openstack.org/387624
11:59 openstackgerrit Emilien Macchi proposed openstack/puppet-keystone: Introduce memcached_ips and memcached_port  https://review.openstack.org/387974
12:12 markvoelker joined #puppet-openstack
12:18 iurygregory EmilienM, ack =)
12:29 cdelatte joined #puppet-openstack
12:39 fultonj joined #puppet-openstack
12:47 rmart04 joined #puppet-openstack
12:55 paramite joined #puppet-openstack
13:01 ayoung joined #puppet-openstack
13:07 StraubTW joined #puppet-openstack
13:09 davideagle joined #puppet-openstack
13:15 davideagle Hi all, I'm using puppet-neutron 7.1.0 module to manage openstack liberty on ubuntu 14.04. I'm using the neutron::server class to configure neutron-server on a controller node, but there seems to be a bug in the module where it adds admin_user, auth_port etc to the filter:authtoken section which in turn prevents neutron-server to run  https://github.com/openstack/puppet-neut​ron/blob/7.1.0/manifests/server.pp#L424 any ideas for a workar
13:15 davideagle ound?
13:17 jperry joined #puppet-openstack
13:18 rmart04_ joined #puppet-openstack
13:19 markvoelker joined #puppet-openstack
13:19 iurygregory davideagle, are you setting auth_uri and identity_uri?
13:20 iurygregory to have auth_port you are setting a value but it's deprecated and you should not use i think
13:20 davideagle no only auth_passwor, auth_tennant, database_connection, db_sync
13:21 iurygregory omg auth_port is deprecated but have a default value
13:21 iurygregory D:
13:22 iurygregory can you try to set https://github.com/openstack/puppet-neutro​n/blob/7.1.0/manifests/server.pp#L287-L289 to undef?
13:22 iurygregory and run?
13:22 davideagle ahh shii that must be the reason
13:23 iurygregory yeah
13:23 iurygregory it should be undef
13:23 iurygregory but we missed
13:23 davideagle totally missed that
13:23 iurygregory sorry =(
13:23 davideagle no worries as long as I can solve it :)
13:24 davideagle have been banging my head against the wall for some time now haha
13:24 iurygregory you can always ask here =D
13:24 davideagle thank you
13:24 iurygregory np
13:26 shardy EmilienM, degorenko: Hey I wanted some further advice on https://review.openstack.org/#/c/3​86495/1/manifests/keystone/domain.pp
13:26 shardy I abandoned that and passed the default values as you suggest, but it looks like there's dependencies which mean even with service default values we try to start heat-engine:
13:27 shardy http://paste.openstack.org/show/586192/
13:27 shardy I don't want that, I just want to create the domain etc on the node running keystone (it should not run heat here at all)
13:27 shardy does that mean I need to restore my conditional change?
13:27 degorenko shardy hey, where i can see your tripleo patch? or changes?
13:28 shardy degorenko: https://review.openstack.org/#/c/386115/
13:28 shardy degorenko: keystone.pp and heat.pp won't necessarily get run on the same node
13:29 EmilienM shardy: yeah heat_config relies on heat-common package
13:29 EmilienM but it shouldn't be a problem
13:29 EmilienM or maybe? maybe it requires ::heat class
13:29 EmilienM yeah that's it
13:29 EmilienM in https://review.openstack.org/#/c/3861​15/5/manifests/profile/base/keystone.pp
13:30 EmilienM we might need to include the base profile for heat
13:30 EmilienM if we don't want that, I guess shardy was right and we need a boolean to whether or not include the heat_config thing
13:32 shardy https://github.com/openstack/puppet-he​at/blob/master/manifests/config.pp#L36
13:33 shardy it seems like there's a bunch of logic I really don't want running on a node that's not running heat
13:33 EmilienM right
13:33 degorenko is there any way to disable autorequire? :)
13:33 shardy I guess it's the heat::deps include which is trying to start the service?
13:34 EmilienM autorequire is not forcing the resource to be here
13:34 shardy https://github.com/openstack/puppet​-heat/blob/master/manifests/deps.pp
13:34 EmilienM shardy: did it fail somewhere in your tests?
13:34 EmilienM or is it working but you see stuffs we shouldn't do?
13:34 degorenko we can also create some dummy resource for heat-common
13:34 shardy EmilienM: Yes, I'm testing with isolated keystone/heat roles and hit a failure because it's trying to start heat-engine on the keystone node
13:35 shardy To me just adding back the conditional seems like a simple solution, but I'm open to other ideas :)
13:35 EmilienM what? heat engine?
13:35 shardy Service[heat-engine]/ensure: ensure changed 'stopped' to 'running'
13:36 shardy Let me re-test and see if it actually starts
13:36 EmilienM it's weird, the heat::keystone::domain class doesn't do it
13:38 EmilienM shardy: i'm ok with conditional...
13:38 shardy EmilienM: unless I'm mistaken, heat::keystone::domain runs heat_config, which includes heat::deps, which has a bunch of dependencies which ends up starting the engine?
13:38 EmilienM shardy: heat::deps has dependencies if resources are declared
13:38 EmilienM in tripleo, we have profiles that will start heat engine only if declared.
13:38 morazi joined #puppet-openstack
13:38 EmilienM shardy: can i see your roles data?
13:39 davideagle iurygregory, this did not work if you take a look at https://github.com/openstack/puppet-neutro​n/blob/7.1.0/manifests/server.pp#L426-L440 too me it seems that if i have auth_type set as keystone and I set auth_password I will always get filter:authtoken settings in my api-paste.ini
13:40 shardy EmilienM: http://paste.fedoraproject.org/454978/97995147
13:40 shardy EmilienM: I'll re-test with the conditional and see if that resolves it
13:40 shardy I suspect I have other issues related to network isolation, but it seemed wrong that heat-engine was trying to start
13:40 EmilienM shardy: heat engine should not start
13:41 EmilienM I'm confused
13:41 shardy Heat::Engine/Service[heat-engine]/ensure: ensure changed 'stopped' to 'running'
13:41 shardy what does that mean then?
13:41 shardy That's from overcloud.AllNodesDeploySteps.H​eatControllerDeployment_Step3.0
13:41 shardy oops
13:41 shardy that's actually the heat role
13:42 shardy doh
13:42 iurygregory davideagle, yeah right the condition when auth_type == keystone and you have auth_password you will have information in filterauthtoken
13:42 mjblack EmilienM: got a question, with regards to p-o-i, why is ssl disabled for ubuntu?
13:42 shardy Ok, sorry for the noise folks
13:42 EmilienM shardy: taking it on #tripleo
13:42 EmilienM mjblack: I never managed to make it work
13:43 mjblack ah ok
13:43 davideagle <iurygregory> according to my neutron-server that logic is wrong
13:43 mjblack hmmm, not sure if its related to the issues with the p-o-i barbican tests
13:44 iurygregory davideagle, yeah i agree, if you set information in keystone_authtoken in neutron.conf you don't need any information in api-paste...
13:45 davideagle <iurygregory> are my options just to clone the module and make the changes needed?
13:47 iurygregory davideagle, i think soo, i'm looking at stable/mitaka we still set keystone_authtoken and filter:authtoken.. this should not happen
13:48 Kvisle mjblack: I apologize if I ended up sending you a bunch of emails by saving the same comment in gerrit multiple times --- new to this..
13:48 jtriley joined #puppet-openstack
13:48 Kvisle mjblack: and I'm sorry for highlighting you, you are not the person I meant to highlight
13:50 davideagle <iurygregory> I noticed that too, but I came across that it seems to have been fixed  somewhere along the way https://github.com/openstack/puppet-neutron​/blob/master/manifests/server.pp#L515-L526
13:51 iurygregory davideagle, yeah but this is only in newton the new authtoken schema for all modules
13:52 davideagle <iurygregory> you guys do not make patches to older versions do you?
13:53 iurygregory we make when possible XD, but Liberty is EOL not sure we can do a patch maybe for mitaka
13:53 iurygregory can you open a bug?
13:53 openstackgerrit Emilien Macchi proposed openstack/puppet-keystone: Introduce memcached_ips and memcached_port  https://review.openstack.org/387974
13:54 iurygregory maybe we can discuss in our meeting that will start in 1 hour (15 UTC in #openstack-meeting-4)
13:54 davideagle <iurygregory> okay, I'll try to get the bug report in by then
13:55 iurygregory davideagle, tks :D
13:57 iurygregory ansiwen, http://logs.openstack.org/70/387970/1/ch​eck/gate-puppet-openstack-integration-4-​scenario002-tempest-centos-7/443ccce/con​sole.html#_2016-10-18_13_35_43_017199
13:58 morazi joined #puppet-openstack
13:59 ansiwen iurygregory: what's that? I mean, I see the failure, but what make it break?
14:00 rpittau joined #puppet-openstack
14:00 med_ joined #puppet-openstack
14:00 iurygregory can you ask the guy in ec2-api if they merged something, i'll only have time to debug at nigh =(
14:01 rmart04 joined #puppet-openstack
14:06 openstackstatus NOTICE: We are away of pycparser failures in the gate and working to address the issue.
14:08 davideagle <iurygregory> bug report is in https://bugs.launchpad.net/​puppet-neutron/+bug/1634529
14:08 openstack Launchpad bug 1634529 in puppet-neutron "api-paste.ini filter:authtoken bug" [Undecided,New]
14:10 iurygregory davideagle, can you join the meeting when it starts?
14:10 davideagle sure
14:10 iurygregory and also add the bug to https://etherpad.openstack.org/p/pupp​et-openstack-weekly-meeting-20161018
14:11 davideagle done
14:17 rpittau hi all
14:18 rpittau hope someone will have a couple of minutes to unrevel a doubt
14:19 iurygregory go ahead and we can try to see if we can help ^^
14:19 rpittau eheh thanks :)
14:19 rpittau so it's about the puppet module for trove
14:19 rpittau There is no default value for the Trove taskmanager_manager setting in the code and this setting must be provided in trove-taskmanager.conf
14:19 rpittau so I put this in: https://review.openstack.org/#/c/386587/
14:20 rpittau because in that module there was no entry for taskmanager_manager
14:20 iurygregory so taskmanager_manager is like a driver information?
14:21 rpittau basically
14:21 rpittau the interesting thing is that for example devstack configures it automatically
14:21 rpittau but there's no os default
14:21 iurygregory :O
14:21 rpittau because there's no default in trove code
14:22 iurygregory https://github.com/openstack/trove/blob/mas​ter/etc/trove/trove-taskmanager.conf.sample
14:22 rpittau yeah, it's in the sample but not in the code
14:22 iurygregory well oslo generate i think =)
14:23 rpittau I guess
14:23 iurygregory rpittau, well since is in the configuration with default
14:23 iurygregory you should change your patch to $::os_service_default
14:24 rpittau ok, thanks, I will give it at try then
14:24 openstackgerrit Mykyta Karpin proposed openstack/puppet-glance: Fix documentation for log_dir parameter  https://review.openstack.org/384508
14:24 rpittau kind of new to this thhing ^^
14:24 iurygregory if you need help let us know =)
14:25 rpittau I sure will! thanks!
14:27 hexlibris joined #puppet-openstack
14:29 ansiwen iurygregory: it's a known race that also apprears in the ec2-api once in a while. he couldn't catch it yet. I just issued a recheck.
14:29 iurygregory ansiwen, tks
14:34 openstackgerrit ayoung proposed openstack/puppet-tripleo: Optionally include mod auth mellon for apache  https://review.openstack.org/359942
14:52 dmsimard ansiwen: we tend to be quick on the ban hammer for flap-sensitive tests, let us know if we can help troubleshooting it
14:52 agrebennikov joined #puppet-openstack
14:53 iurygregory meeting in 8 minutes
14:56 ansiwen dmsimard: andrey-mp is the one trying to find the race... I'm just relaying the messages between the channels :-)
14:57 ansiwen dmsimard: but I will tell him
14:57 dmsimard ack
14:58 EmilienM mwhahaha: /me leaving home now. I'll miss the meating. Let me know if anything.
14:58 mwhahaha k
15:00 ansiwen dmsimard: so, would that mean to kick out ec2api from tripleo as well?  since tests are a requirement for tripleo inclusion this can have quite a chain reaction
15:00 bklei joined #puppet-openstack
15:00 iurygregory davideagle, join #openstack-meeting-4
15:00 dmsimard ansiwen: if the test is flapping, it has an impact on tripleo and all openstack puppet modules
15:01 iurygregory meeting in #openstack-meeting-4 people o/
15:06 openstackgerrit Mykyta Karpin proposed openstack/puppet-ceilometer: Fix documentation for log_dir parameter  https://review.openstack.org/384463
15:07 paramite joined #puppet-openstack
15:10 ansiwen dmsimard: if it results into a feature removed from tripleo that's also an impact... :-)
15:10 dmsimard ansiwen: a valid point to make sure it's fixed then :p
15:11 ducttape_ joined #puppet-openstack
15:13 ansiwen dmsimard: I'm afraid I'm not able to fix it within reasonable time, never ever looked at the code of ec2api... I'm just the guy that tries to get the feature integrated into tripleo... >_<
15:13 petems joined #puppet-openstack
15:22 openstackgerrit Iury Gregory Melo Ferreira proposed openstack/puppet-openstack-guide: Meeting #94  https://review.openstack.org/388074
15:25 mwhahaha ansiwen: i can help out if you need another set of eyes on the issue
15:32 openstackgerrit François Charlier proposed openstack/puppet-ceph: Notify radosgw service on configuration change  https://review.openstack.org/388080
15:33 openstackgerrit Merged openstack/puppet-monasca: Provide default service_name for keystone endpoint  https://review.openstack.org/380811
15:36 mwhahaha dmsimard: did marathon get packaged (and promoted)?
15:36 dmsimard mwhahaha: need to fix a nitpick in the spec file review
15:36 dmsimard not yet
15:36 mwhahaha k
15:36 dmsimard maybe tomorrow
15:44 openstackgerrit Merged openstack/puppet-openstack-guide: Meeting #94  https://review.openstack.org/388074
15:47 davideagle joined #puppet-openstack
15:50 ansiwen mwhahaha: as I said, andrey-mp is looking at it before, and he has an idea, but he couldn't reproduce it yet.
15:50 xarses joined #puppet-openstack
15:51 mwhahaha Sure just offering if you need some help :)
15:52 iurygregory dmsimard, the only party i know is the one you put in operator list now haha
15:52 dmsimard bah it's not a party as is
15:53 iurygregory we should have a puppet dinner XD
15:53 iurygregory mwhahaha, what you think about? haha
15:53 mnaser when a patch you submitted in june gets merged today :P
15:53 mnaser or yesterdaya rather
15:54 iurygregory mnaser, put in meeting XD so we don't forget to give attention
15:54 mnaser actaully i didnt bother too much with it, someone rebased it and merged it in! :P
15:54 mwhahaha iurygregory: i'm not going, so drink for me :D
15:55 iurygregory mwhahaha, what?! :O and who will be the chair for our sessions? (Emilien?)
15:55 mwhahaha yea
15:55 iurygregory oh ok o/
15:56 iurygregory we should have a confraternization i think, since there is no OpenStack parties haha
16:03 rmart04 left #puppet-openstack
16:04 openstackgerrit Riccardo Pittau proposed openstack/puppet-trove: fixing storage_namespace; the correct location is under trove.common  https://review.openstack.org/388090
16:12 openstackstatus NOTICE: pycparser 2.16 released to fix assertion error from today.
16:19 markvoelker joined #puppet-openstack
16:21 openstackgerrit Alex Schultz proposed openstack/puppet-tripleo: Add barbican profile rspec testing  https://review.openstack.org/388099
16:45 pester joined #puppet-openstack
17:22 EmilienM fyi, tripleo CI had outage, specially on stable/newton patches. All should be back to normal once https://review.openstack.org/#/c/387750/ will merge (soon hopefully)
17:22 EmilienM so if you had a puppet backport that failed to run tripleo jobs, please recheck when this patch is merged ^
17:23 iurygregory ack
17:24 EmilienM ok now I have 8h of flight to learn spanish
17:33 jgrassle1 joined #puppet-openstack
17:33 Anticimex joined #puppet-openstack
17:33 _matthias_ joined #puppet-openstack
17:33 LarsErikP joined #puppet-openstack
17:33 Hatsjoe joined #puppet-openstack
17:34 dhollinger joined #puppet-openstack
17:34 beddari joined #puppet-openstack
17:34 Hatsjoe joined #puppet-openstack
17:34 basilAB joined #puppet-openstack
17:34 vern joined #puppet-openstack
17:34 lucasagomes joined #puppet-openstack
17:34 jistr joined #puppet-openstack
17:34 jpena|off joined #puppet-openstack
17:34 khappone joined #puppet-openstack
17:34 EmilienM joined #puppet-openstack
17:35 haaj joined #puppet-openstack
17:35 EmilienM joined #puppet-openstack
17:38 _stowa joined #puppet-openstack
17:39 patrickeast joined #puppet-openstack
17:40 kgalanov joined #puppet-openstack
17:41 igorbelikov joined #puppet-openstack
17:42 aglarendil joined #puppet-openstack
17:51 jperry joined #puppet-openstack
18:07 dprince joined #puppet-openstack
18:12 dencaval hey guys
18:13 dencaval How can I resolve puppetlabs-rabbitmq (v5.5.0) dependency: UNMET DEPENDENCY puppet-staging (>=0.3.1 <2.0.0)?
18:17 iurygregory puppet module install <name> --version ?
18:17 iurygregory i never saw this problem for unmet dependency
18:17 openstackgerrit Merged openstack/puppet-heat: Fix documentation for log_dir parameter  https://review.openstack.org/384494
18:17 openstackgerrit Merged openstack/puppet-barbican: Update barbican for Debian based systems  https://review.openstack.org/384691
18:18 dencaval iurygregory I'have try to install it, but: Currently, 'nanliu-staging' (v1.0.4) is installed to that directory
18:18 iurygregory ody, Hunner you guys have any idea about this? ^
18:20 Hunner yeah, this is "module changed namespace and module dependency resolution doesn't understand it"
18:20 Hunner nanliu-staging vs puppet-staging
18:20 iurygregory tks Hunner o/
18:21 Hunner it happens most often with highly-used modules, like when ripienaar-concat changed to puppetlabs-concat
18:21 Hunner puppet-staging is the right way to go, so any modules still on nanliu-staging should have their metadata.json fixed
18:22 iurygregory yeah =) tks
18:22 dencaval Thanks, I'll try to remove the nanliu and install the puppet-staging
19:05 openstackgerrit Merged openstack/puppet-openstack-integration: Add puppet-qdr and puppet-cloudkitty  https://review.openstack.org/387970
19:07 openstackgerrit Merged openstack/puppet-trove: Change worker defaults to ::os_workers  https://review.openstack.org/383957
19:19 pester joined #puppet-openstack
19:31 fultonj joined #puppet-openstack
19:49 petems joined #puppet-openstack
19:53 openstackgerrit Merged openstack/puppet-ceilometer: Fix documentation for log_dir parameter  https://review.openstack.org/384463
19:54 openstackgerrit Merged openstack/puppet-barbican: Manage package created apache config file  https://review.openstack.org/387027
19:55 openstackgerrit Merged openstack/puppet-keystone: Fix documentation for log_dir parameter  https://review.openstack.org/384588
19:55 openstackgerrit Merged openstack/puppet-tripleo: Remove explicit service_name setting from nova manifest  https://review.openstack.org/386522
19:56 dencaval_ joined #puppet-openstack
19:58 dencaval__ joined #puppet-openstack
20:01 michapma_alt joined #puppet-openstack
20:02 openstackgerrit Matthew J Black proposed openstack/puppet-barbican: Manage package created apache config file  https://review.openstack.org/388195
20:03 mjblack EmilienM: I think the barbican CI updates might need to wait till newton is officially released, the current uca packages may have bad code in them
20:03 mjblack for newton that is
20:05 iurygregory mjblack, he is travelling to Barcelona at this moment
20:10 mjblack oh
20:10 mjblack well
20:10 iurygregory he will read XD but not know
20:10 mwhahaha i thought UCA had officially released newton
20:10 mjblack uhhh
20:11 mjblack when?
20:11 iurygregory i saw an email in openstack-dev i think
20:11 mwhahaha yea like last week
20:11 mjblack hmmm
20:11 iurygregory [openstack-dev] [Openstack] OpenStack Newton B3 for Ubuntu 16.04 LTS and Ubuntu 16.10
20:12 mjblack well I guess it just means keep looking
20:12 mjblack barbican devs were saying it seems like the cinder key manager is messed up on the ubuntu build
20:13 iurygregory enr .-.
20:13 * mjblack goes to sleep
20:43 cdelatte joined #puppet-openstack
20:51 ayoung joined #puppet-openstack
20:53 EmilienM dprince: fyi, mwhahaha's comment on https://review.openstack.org/#/c/387974/
20:54 EmilienM mwhahaha: the idea was to remove the logic from puppet-tripleo and add it in puppet modules, so we can use include ::class instead of declaring the class and compute the parameters
20:54 mwhahaha yea i still don't like it :D
20:54 mwhahaha i hate both methods to be honest
20:55 mwhahaha but we shouldn't be duplicating the parameter like that
20:55 EmilienM ok
20:55 EmilienM what method do you like?
20:56 mwhahaha so can't heat do it?
20:56 openstackgerrit Sven Anderson proposed openstack/puppet-openstack-integration: Disable flaky ec2-api tempest test.  https://review.openstack.org/388212
20:56 mwhahaha the address/port collection
20:56 EmilienM mwhahaha: apparently, it could. dprince knows how to do it. But I didn't like it to be honnest
20:57 EmilienM because this thing affects people not using heat
20:57 mwhahaha since we already have the memecached_servers we should be using that unless you want to update every module to do both
20:58 EmilienM the pb is ipv6
20:58 mwhahaha but i really don't want to add the duplicated flow to just one module either
20:58 EmilienM we need to make sure params have brackets
21:01 mwhahaha then we need to pick one method and apply it everywhere
21:01 mwhahaha i don't like the separate ips and port because that assumes it's universal
21:01 mwhahaha it doesn't have to be
21:01 openstackgerrit Trygve Vea proposed openstack/puppet-cinder: Make EMC VNX backend class compatible with FC driver  https://review.openstack.org/387263
21:02 mwhahaha if you push the logic up the stack then it's the consumer of the modules responsibility to cleanup their server list
21:02 dprince mwhahaha: long term I want it in Heat.
21:03 mwhahaha i prefer the single server list array parameter than the ip param and  port param
21:03 dprince mwhahaha: I poked at the heat approach a bit yesterday and it needed more time than I had. So I pushed https://review.openstack.org/#/c/387624/ for backwards compat with what we did in mitaka
21:04 mwhahaha dprince: did you read my comment on perf around that?
21:04 dprince mwhahaha: the gist of that ticket is to fix a regression from Newton where we removed the nova.conf authtoken memcached setting
21:04 mwhahaha yea i got that part, the problem i was pointing out is that that configuration may not be ideal
21:04 dprince mwhahaha: fair enough, but we are getting called out on it
21:05 * mwhahaha <inserts face palm here>
21:05 mwhahaha software for numbers sake, cool
21:06 dprince mwhahaha: I don't think it is that way at all. FWIW, I just saw your comment so we *can* address this
21:06 dprince mwhahaha: I don't think anybody realized our previous config wasn't optimal
21:06 dprince mwhahaha: but a regression is a regression
21:06 mwhahaha not when a regression is a feature :D
21:07 mwhahaha it's fine i've just battled this memcached server stuff for like 3 months now and it's a mess
21:08 dprince mwhahaha: the list conversion stuff in heat (to append the port) isn't ideal. It would be nice if puppet-* did that in one place
21:09 openstackgerrit Tim Rozet proposed openstack/puppet-tripleo: Fixes missing ODL ML2 Authentication info  https://review.openstack.org/386820
21:09 dprince mwhahaha: then we can set simple hiera in t-h-t... for all the services. That was my thought for ocata.
21:09 mwhahaha dprince: so i had previously spent a bunch of time trying to align all the memcache settings for the puppet modules and we just got them universally there
21:10 mwhahaha dprince: i'm not keen on doing it again which is why i'd like to see a plan to get it done
21:10 dprince mwhahaha: but, that can't be easily backported into Newton.
21:10 dprince mwhahaha: this was the bug that spawned all this FWIW https://bugzilla.redhat.co​m/show_bug.cgi?id=1384031
21:10 openstack bugzilla.redhat.com bug 1384031 in rhel-osp-director "Openstack components uses in-process cache instead of memcache" [High,On_dev] - Assigned to dprince
21:11 dprince mwhahaha: it really is a simple fix for the regression I think. And it uses the new puppet modules
21:11 openstackgerrit Tim Rozet proposed openstack/puppet-tripleo: Fixes missing ODL ML2 Authentication info  https://review.openstack.org/386820
21:12 mwhahaha dprince: i'm talking about the first patch not your second one
21:12 mwhahaha https://review.openstack.org/#/c/387974/ <- not keen on that one
21:12 dprince mwhahaha: my patch was first I think?
21:12 dprince mwhahaha: okay, well lets debate that for ocata then
21:12 mwhahaha https://review.openstack.org/#/c/387624/ <- regression ok, but we should fix the caching method
21:13 dprince mwhahaha: in the short term I would like something we can (quickly) backport to newton to resolve the nova regression
21:13 dprince mwhahaha: either that, or we punt and say forget it
21:13 dprince mwhahaha: I don't think we have time to introduce new feature/logic into newton to handle memcache stuff better there. Which is why I bailed on the heat approach for newton
21:13 mwhahaha dprince: yea thats why i said ok and backport the fix for that one
21:14 dprince mwhahaha: so for Emilien's fix you want it to be always a single server?
21:14 dprince mwhahaha: liek that is fine, but if this is really the preferred case why is the config a list at all?
21:15 mwhahaha it's not a single server
21:15 mwhahaha it's an array
21:15 dprince mwhahaha: but your suggestion was to always configure just one server in that array, right?
21:16 mwhahaha my understanding is that for the best performance it should just be a single server
21:16 EmilienM i'm afk from now, might come back before takeoff, not sure.
21:16 AndyWojo joined #puppet-openstack
21:16 mwhahaha because the python memcache library pools like a db and doesn't use memcache as it was designed
21:16 mwhahaha so when a server is down, you are hit with the connection penalty
21:16 mwhahaha so caching makes it
21:17 mwhahaha so caching makes it terrible
21:17 ducttape_ joined #puppet-openstack
21:18 dprince mwhahaha: so what logic should we use (in t-h-t) to select a server then?
21:18 dprince mwhahaha: we can do this with yaql for example
21:18 mwhahaha we should probably be deploying a single small memcache instance on servers with APIs
21:18 dprince mwhahaha: all of our controllers currently have this
21:19 dprince mwhahaha: so you are suggesting we just use 127.0.0.1 then?
21:19 mwhahaha for the keystone authtoken caching, yes
21:19 saneax-_-|AFK joined #puppet-openstack
21:20 ducttape_ joined #puppet-openstack
21:21 dprince mwhahaha: well I think this actually makes things a lot simpler
21:22 dprince mwhahaha: we could actually just enable memcache only if the memcache service is configured on a given role, for the rest of the API's
21:22 dprince mwhahaha: would you be up for filing a bug so we capture this for ocata then?
21:22 mwhahaha dprince: sure
21:23 dprince mwhahaha: cool. I like that it eliminates the complexity around appending the port to the array of IP addresses in heat (and puppet-*)
21:24 dprince mwhahaha: and, we don't even need to track that array anymore
21:24 mwhahaha now there might be more cases where it would be ideal to have multiple servers but it should be fewer
21:24 mwhahaha because there's a bunch of different caches but specifically for keystone authtoken it should just be one
21:24 dprince mwhahaha: yeah, perhaps we should reach out to bandini, and gfidente and see if they have thoughts on this too (HA guys)
21:24 mwhahaha otherwise you ends up with an api request that exceeds slas
21:25 dprince mwhahaha: what we just spoke of leaves the other memcache lists alone I think. It would be authtoken specific I think
21:25 dprince mwhahaha: and FWIW, the other cache lists I've seen are handled more nicely in heat in that they just take a list of IP addresses. Not a list of IP:Port strings
21:26 dprince mwhahaha: the second case we can solve in heat anyways for ocata
21:26 mwhahaha dprince: well i think it uses 11211 by default so maybe no one has noticed yet
21:26 dprince mwhahaha: again, my reason for not doing it that way for Newton was largely to keep the patch size tight
21:27 dprince mwhahaha: could be all for naught :)
21:27 mwhahaha :D
21:27 hjensas joined #puppet-openstack
21:27 hjensas joined #puppet-openstack
21:31 openstackgerrit Sven Anderson proposed openstack/puppet-tripleo: Implement Nova ec2api profile  https://review.openstack.org/336562
21:58 dgurtner joined #puppet-openstack
21:58 dgurtner joined #puppet-openstack
22:13 EmilienM mwhahaha: I'm investigating in all modules / master if we need to backport some things we missed into stable/newton and if yes, propose new releases (like we did for some of them)
22:13 EmilienM so don't be surprised to see some backports / releease requests this week
22:14 EmilienM I'm aware some downstreams projects (RDO, etc) use tags as reference for deploying some projects (ex: TripleO). So I want to make sure our releases are in good shape
22:14 EmilienM I'll only propose bugfixes backports.
22:15 openstackgerrit Merged openstack/puppet-oslo: Allow setting state for backend packages  https://review.openstack.org/387503
22:15 openstackgerrit Merged openstack/puppet-ceilometer: Release 9.4.1 (newton)  https://review.openstack.org/386663
22:16 mwhahaha EmilienM: ok
22:16 gfidente mwhahaha so the reason for memcached there is the vnc auth
22:17 gfidente are you suggesting to deploy a single instance and just point to that from all nova-api(s) ?
22:17 mwhahaha No
22:17 mwhahaha Gimme a minute, I'll explain
22:19 xarses unless someone got around to fixing the python memcache clients, each host should use it's own, non clustered memcache host
22:19 mwhahaha -^
22:19 gfidente xarses yeah well they arent' clustered
22:20 mwhahaha clustered being a collection of like memcache
22:20 mwhahaha not actual cluster
22:20 gfidente yeah they don't replicate data
22:20 mwhahaha gfidente: so we should deploy a small memcache instance locally for any keystone caching needs
22:20 gfidente though if it's local it's not shared amongst the nova-api(s) right?
22:20 xarses you should not leave your host for memcache service
22:21 mwhahaha it's a cache
22:21 mwhahaha it's not the sharing part that you're getting the benefit from
22:21 mwhahaha it's the not having to go back to the source
22:21 mwhahaha it's better to configure many small (~32M) instances for the temporary cache than one giant one
22:21 mwhahaha also this reduces your risk when it ultimately fails
22:22 mwhahaha if you restrict it to a single host, you only impact the one service runing on that host
22:22 mwhahaha there's a larger performance related issue with python memcache and then the caching in the workers and everything else. unlike something like java where the state information can be better shared amungst an app
22:23 mwhahaha so the best solution for the deployment is to deploy a small memcache instance on any host providing an api endpoint and configure all *::keystone::authtoken::memcached_servers to use that
22:23 mwhahaha gfidente: make sense?
22:24 gfidente mwhahaha it does but I am pretty sure we had to use memcached to *share* the tokens amongst the consoleauth service
22:24 mwhahaha now the cache could be shared between nova/glance/cinder/whatever
22:24 gfidente so it could be shared and be local if it was serving the sole purpose of a cache
22:24 mwhahaha gfidente: that's a different cache then what he's configuring
22:24 gfidente (shared amongst the services)
22:24 mwhahaha gfidente: keystone_token cache is not the nova cache
22:25 mwhahaha so what you describe is possible but unrelated to this specific caching configuration
22:25 gfidente mwhahaha ack then there the issue is then
22:25 gfidente or seems to be
22:25 mwhahaha https://github.com/openstack/puppet-​nova/blob/master/manifests/cache.pp vs https://github.com/openstack/puppet-nova/b​lob/master/manifests/keystone/authtoken.pp
22:25 mwhahaha what i'm refering to is nova::keystone::authtoken
22:26 mwhahaha the vnc issue uses a different cache setting
22:29 gfidente agreed what consoleauth uses is DEFAULT/memcached_serves
22:30 openstackgerrit Merged openstack/puppet-aodh: Release 9.4.1 (newton)  https://review.openstack.org/386662
22:32 mwhahaha So that also seems like a terrible configuration but less important than the performance hit on api services
22:32 gfidente mwhahaha which is already there https://github.com/openstack/puppet-tripleo/b​lob/master/manifests/profile/base/nova.pp#L73
22:32 mwhahaha so yea for vncconsole auth you'd probably want to configure many servers
22:32 gfidente so I wonder if we want to enable use of memcached at all when not needed
22:32 gfidente for APIs
22:32 mwhahaha yea it speeds up access
22:33 gfidente yes but I mean given issue with having a pool of servers
22:33 gfidente we might use it only for consoleauth
22:33 gfidente for which config is working
22:33 gfidente s/might/might want/
22:33 mwhahaha the whole point of memcache is to have lots of little ones
22:33 mwhahaha there's no reason we can't just universally run memache on localhost with like 32M
22:34 gfidente well that doesn't work for consoleauth
22:34 mwhahaha not for console auth
22:34 mwhahaha you need a specific configuration for consoleauth
22:34 mwhahaha which is just stupid
22:34 openstackgerrit Merged openstack/puppet-tacker: Populating init.pp and parameters  https://review.openstack.org/387419
22:34 mwhahaha that's a worked in devstack right t here
22:36 gfidente mwhahaha but we'll have to deal with the string manipulation for consoleauth
22:36 gfidente maybe we can remove the 11211
22:36 mwhahaha why?
22:36 gfidente as it's the default
22:36 gfidente could make it easier to mangle in heat
22:36 mwhahaha no i meant the string stuff
22:37 gfidente to build the list of servers for consoleauth
22:37 mwhahaha i guess i need to go digging into the memcache list in heat
22:37 mwhahaha i'm missing something i guess
22:38 gfidente well as you said, we'd really need to distribute two different configs
22:38 gfidente one for consoleauth, localhost for the rest
22:38 mwhahaha so you use the controllers for consoleauth
22:38 mwhahaha and localhost for the rest
22:39 gfidente yeah seems like it could do?
22:39 mwhahaha that would be my choice
22:39 mwhahaha or you could even just use all the nova api servers via yaql or something
22:39 gfidente so we do the localhost with 32M which you suggested
22:39 gfidente and point there to speed up
22:40 mwhahaha ideally it'd be great to do some scale testing to better understand what would be a good setting but that would be a start
22:41 gfidente and it's only this stuff https://github.com/openstack/puppet-tripleo/b​lob/master/manifests/profile/base/nova.pp#L59 we need to move into templates
22:41 mwhahaha we don't even need that
22:41 mwhahaha if we just write a parser function to deal with it
22:41 mwhahaha i don't like those ipv6 flags anwyay
22:55 openstackgerrit Merged openstack/puppet-horizon: Remove unnecessary 'processorcount' facts  https://review.openstack.org/383741
23:08 dencaval_ joined #puppet-openstack
23:14 openstackgerrit Merged openstack/puppet-mistral: Switch mistral api to use os_workers  https://review.openstack.org/386691
23:14 openstackgerrit Merged openstack/puppet-trove: fixing storage_namespace; the correct location is under trove.common  https://review.openstack.org/388090
23:15 dgurtner joined #puppet-openstack
23:15 dgurtner joined #puppet-openstack
23:17 ducttape_ joined #puppet-openstack
23:17 xarses joined #puppet-openstack
23:27 openstackgerrit Merged openstack/puppet-tripleo: Set memcached_servers for nova API  https://review.openstack.org/387624
23:32 dencaval_ joined #puppet-openstack
23:36 dencaval__ joined #puppet-openstack
23:41 markvoelker joined #puppet-openstack
23:42 dencaval_ joined #puppet-openstack
23:43 dencaval__ joined #puppet-openstack
23:44 dencaval___ joined #puppet-openstack
23:46 dencaval___ joined #puppet-openstack
23:49 EduNeto joined #puppet-openstack
23:51 iurygregory_ joined #puppet-openstack
23:54 jtriley joined #puppet-openstack

| Channels | #puppet-openstack index | Today | | Search | Google Search | Plain-Text | summary