Friday, 2015-09-18

*** achanda_ has joined #kolla00:02
*** jerryz has quit IRC00:04
*** achanda has quit IRC00:06
*** achanda_ has quit IRC00:08
*** pbourke_ has quit IRC00:18
openstackgerritsean mooney proposed openstack/kolla: Fix missing openvswitch rpm in neutron-base with centos source build.  https://review.openstack.org/22493600:18
*** pbourke_ has joined #kolla00:19
*** daneyon_ has quit IRC00:19
*** jerryz has joined #kolla00:23
*** jruano has quit IRC00:23
jerryzsdake: ping00:31
jerryzsdake: i just experienced a server shutdown. some of the docker containers including keystone and horizon keep restarting because there was an stale httpd.pid left.00:32
jerryzSamYaple: ^00:35
*** dims_ has joined #kolla00:38
*** banix has joined #kolla00:40
*** jruano has joined #kolla00:42
*** juggler_ has joined #kolla00:43
*** dims_ has quit IRC00:43
*** dims_ has joined #kolla00:43
*** juggler_ has quit IRC00:45
*** juggler_ has joined #kolla00:45
*** juggler_ has quit IRC00:48
*** juggler has joined #kolla00:49
*** juggler has quit IRC00:50
openstackgerritQiu Yu proposed openstack/kolla: Fix ring bulder command examples in swift readme doc  https://review.openstack.org/22494601:00
openstackgerritMerged openstack/kolla: Refactor set_configs.py  https://review.openstack.org/22455701:01
*** juggler has joined #kolla01:02
*** banix has quit IRC01:02
*** erkules_ has joined #kolla01:05
*** blakec has quit IRC01:05
*** erkules has quit IRC01:08
*** weiyu has joined #kolla01:10
*** huikang has quit IRC01:18
*** banix has joined #kolla01:18
*** banix has quit IRC01:24
*** jerryz has quit IRC01:26
*** dims_ has quit IRC01:27
*** dims_ has joined #kolla01:29
*** alisonh has left #kolla01:32
*** dims_ has quit IRC01:37
*** unicell has quit IRC01:46
sdakejerryz yes we have an open bug for that02:00
sdakejerryz the individual fixing it doesn't seem to be following up on the review comments, so someone else may have to take it over02:00
*** alisonh has joined #kolla02:04
*** sdake_ has joined #kolla02:09
*** banix has joined #kolla02:13
*** sdake has quit IRC02:13
*** jruano has quit IRC02:27
*** achanda has joined #kolla02:28
*** jruano has joined #kolla02:32
*** dayou has joined #kolla02:46
*** blakec has joined #kolla03:05
*** tfukushima has joined #kolla03:17
*** dims_ has joined #kolla03:22
*** achanda has quit IRC03:27
*** dims_ has quit IRC03:28
*** sdake has joined #kolla03:29
*** sdake_ has quit IRC03:32
SamYaplesdake: that is not at all what was implemented and i removed his settings from kolla, but didnt disable the ability to do it03:34
SamYaplethats a completely differnet issue03:34
sdakehuh?03:34
sdakecontext?03:34
SamYaple22:02 < sdake> samyaple I believe pbourke had implemented non-root based sudo but you reverted his change this morning03:34
sdakeoh03:34
sdakei have no idea what was done there or why it was reverted03:35
sdakeit seemed like it introduced a defect of some sort03:35
SamYaplewell it shouldn't have been there in the first place but it also broke things, yes03:35
SamYapleits a setting only typically used in the inventory file or the ansible.cfg03:35
SamYapleboth files that kolla should not really be involved with03:35
SamYaplewe just supply a sample inventory file03:36
sdakewfm03:36
SamYapleone of these days someone will comethrough and review ceph for me03:36
SamYapleone of these days....03:36
sdakeoh right i didn't get to it today03:36
sdakeapoloogies for that03:36
SamYapleno problem03:36
SamYapleits a real pita to review03:37
SamYapledid you see my suggestion about using the image_prefix so we dont have to update all teh files again like that?03:37
*** sdake_ has joined #kolla03:38
*** sdake_ has quit IRC03:39
*** sdake_ has joined #kolla03:40
*** sdake_ has quit IRC03:40
*** sdake_ has joined #kolla03:40
sdake_SamYaple re your review of the install_type and install_metatype03:41
sdake_I did try that prefix approach initially03:41
sdake_there was something about it that didn't work03:41
sdake_but i dont mind trying to sort it out03:41
sdake_is the advantage there just to get rid of one of the variables?03:41
*** sdake has quit IRC03:41
SamYapleyea we dont have to ever touch all the files like this again03:42
SamYaplealso it would allow greater control of the image name03:42
sdake_i think the reason may have been i wasn't sure what prefix actually meant ;)03:42
sdake_whereas install_name is clear03:42
sdake_but the images were also not building03:43
SamYapleit is not clear to me at all03:43
SamYapleinstall_name makes no sense to be honest03:43
SamYaplebut thats a different thing03:43
sdake_what does prefix mean then?03:43
sdake_whatever the case, it can'tstay install_type03:44
sdake_so all the docker files need modification03:44
sdake_it takes 30 seconds to type the sed command in so i don't mind changing it03:44
SamYaple"22:02 < sdake> samyaple I believe pbourke had implemented non-root based sudo but you reverted his change this morning03:44
SamYapleoops sorry03:44
SamYapleprefix? it means to put in front of?03:44
SamYaplethats way more clear03:44
sdake_so for -t binary-rdo, prefix = what?03:45
SamYaplecentos-binary-rdo-03:45
SamYaplei mean thats what we already for in the builkd.py03:45
sdake_so you want me to get rid of the base distro in the string?03:45
SamYaplethats not new functionality03:45
SamYapleyea03:45
sdake_in the dockerfiles03:45
SamYaplei put an aexample in celiometer03:45
sdake_oh let me check it out03:46
sdake_ok i guess that makes sense03:47
sdake_the reason i didn't use prefix is because it included centos, and i wanted to keep the base_distro in the from line03:47
sdake_but i dont feel strongly about it03:47
sdake_i see what you mean by never having to modify the docker files again03:48
sdake_so that seems worth it03:48
sdake_if i cn just get my linux box into a bootable state again maybe I can actually do the work :)03:48
sdake_(upgraded from win7 to windows 10, replaced bootloader)03:48
* sdake_ groans03:48
*** jruano has quit IRC03:48
*** sdake_ is now known as sdake03:48
SamYaplei really just dont want to touch all the files again03:50
SamYaplethis wilhelp l future proof the namping03:50
SamYaplenaing*03:50
sdakeright03:52
sdakeoh btw, i am going to call it OSP instead of RHOS03:52
sdakeI guess RHOS is the old naming03:52
sdakeOSP = "OpenStack Platform"03:52
sdakeRHOS = "Red Hat OpenStack"03:52
sdakeanother followup SamYaple  arey ou suggessting having -t binary-rdo convert to the name centos-rdo-base?03:55
sdakeand -t binary conver to the name centos-binary-bae?03:55
sdakei'm a little confused by your comments on this point03:56
sdakein build.py03:56
*** unicell has joined #kolla03:57
sdakei thought the idea was to have type be either binary or source, and the meta type to be the exact specific type in action04:03
openstackgerritSwapnil Kulkarni (coolsvap) proposed openstack/kolla: Spelling corrections in cinder guide  https://review.openstack.org/22499404:23
openstackgerritSwapnil Kulkarni (coolsvap) proposed openstack/kolla: Spell corrections in AIO doc  https://review.openstack.org/22499504:23
openstackgerritSwapnil Kulkarni (coolsvap) proposed openstack/kolla: Spell corrections in dev quickstart guide  https://review.openstack.org/22499604:23
openstackgerritSwapnil Kulkarni (coolsvap) proposed openstack/kolla: Spell corrections in Image building guide  https://review.openstack.org/22499704:23
*** banix has quit IRC04:35
*** chandankumar has joined #kolla04:40
*** achanda has joined #kolla04:40
*** weiyu has quit IRC04:43
*** tfukushima has quit IRC05:03
*** weiyu has joined #kolla05:19
*** coolsvap|holiday is now known as coolsvap05:28
*** openstackgerrit has quit IRC05:31
*** tfukushima has joined #kolla05:31
*** openstackgerrit has joined #kolla05:31
*** weiyu has quit IRC05:38
*** sdake has quit IRC05:49
*** weiyu_ has joined #kolla06:14
*** dayou has quit IRC06:17
*** dayou has joined #kolla06:21
*** achanda has quit IRC06:56
*** gfidente has joined #kolla07:02
*** gfidente has joined #kolla07:02
*** jmccarthy has quit IRC07:19
*** jmccarthy has joined #kolla07:20
*** akwasnie has joined #kolla07:28
*** tfukushima has quit IRC07:41
*** tfukushima has joined #kolla07:42
*** erkules_ is now known as erkules07:51
*** erkules has joined #kolla07:51
*** athomas has joined #kolla07:56
*** chandankumar has quit IRC07:57
*** shardy_ has joined #kolla08:14
*** shardy has quit IRC08:15
*** shardy_ has quit IRC08:19
*** shardy has joined #kolla08:20
*** dims_ has joined #kolla08:30
*** dims_ has quit IRC08:34
*** diga has joined #kolla08:42
*** blakec has quit IRC08:43
*** blakec has joined #kolla08:54
*** tfukushima has quit IRC09:04
*** tfukushima has joined #kolla09:05
*** dayou has quit IRC09:19
*** diga has quit IRC09:21
*** blakec has quit IRC09:24
*** dims_ has joined #kolla09:32
*** weiyu_ has quit IRC09:32
*** weiyu_ has joined #kolla09:34
*** tfukushima has quit IRC09:35
*** dims_ has quit IRC09:36
openstackgerritSwapnil Kulkarni (coolsvap) proposed openstack/kolla: Spell corrections in Image building guide  https://review.openstack.org/22499709:40
*** dims_ has joined #kolla09:50
*** [Leeloo is now known as [Leeloo]10:02
*** athomas has quit IRC10:03
*** athomas has joined #kolla10:08
*** weiyu_ has quit IRC10:40
*** dayou has joined #kolla10:51
*** dayou has quit IRC10:58
*** mfalatic has quit IRC11:00
*** mfalatic has joined #kolla11:06
*** rhallisey has joined #kolla11:11
*** shardy_ has joined #kolla11:55
*** shardy has quit IRC11:56
jmccarthyAnyone else run into any permission errors like this:11:56
jmccarthyTASK: [haproxy | Allowing non-local IP binding]11:56
jmccarthy...11:56
jmccarthyOSError: [Errno 13] Permission denied: '/etc/.ansible_m_sysctl_QTaom0.conf'11:56
jmccarthyI have ansible_ssh_user: "kolla" with sudo rights on the nodes in question11:56
*** dayou has joined #kolla11:58
*** shardy_ has quit IRC12:00
*** shardy has joined #kolla12:01
*** weiyu has joined #kolla12:04
*** weiyu has quit IRC12:21
[Leeloo]Hello, I try to use Kolla under Debian 8  (Ubuntu / source), but got this error --> https://paste.fedoraproject.org/268861/25791271/ :(12:26
*** dwalsh has joined #kolla12:27
*** banix has joined #kolla12:28
akwasnieLeeloo: got the same problem, anyone knows what can be the issue?12:52
SamYaple[Leeloo] akwasnie: what do the logs from the bootstrap container say?12:58
akwasnieHi SamYaple ERROR: openstack The plugin token_endpoint could not be found12:59
openstackgerritMerged openstack/kolla: Implement configuration file for kolla-build  https://review.openstack.org/22444213:02
openstackgerritMerged openstack/kolla: Set namespace to point to local registry  https://review.openstack.org/22341913:03
SamYapleakwasnie: wierd. my guess is upstream keystone changed auth things. i heard murmuring about this. can you file a bug?13:03
akwasniesure, will do13:03
SamYaplejmccarthy: ansible_ssh_user shouldn't be set in the group_vars or the globals for one, its an inventory or ansible.cfg setting only (it breaks some things in ansible)13:08
SamYaplejmccarthy: additionally the playbooks are not gauranteed to work without root so I wouldn't expect them too13:08
SamYaplethe push to fix all the root and privelege dropping will be done in the Mitaka 1 cycle13:09
SamYaplefor liberty the official stance is root is required13:09
*** banix has quit IRC13:21
*** banix has joined #kolla13:22
openstackgerritVladislav Belogrudov proposed openstack/kolla: Fix broken centos-source- build: add rdo repository  https://review.openstack.org/22396313:23
openstackgerritRyan Hallisey proposed openstack/kolla: Replace config-external with a JSON file for Glance  https://review.openstack.org/21850113:33
openstackgerritRyan Hallisey proposed openstack/kolla: Replace config-external with a JSON file for Keystone  https://review.openstack.org/21849713:33
openstackgerritRyan Hallisey proposed openstack/kolla: Replace config-external with a JSON file for Heat  https://review.openstack.org/22489113:33
openstackgerritRyan Hallisey proposed openstack/kolla: Replace config-external with a JSON file for Rabbitmq  https://review.openstack.org/21849613:33
openstackgerritRyan Hallisey proposed openstack/kolla: Replace config-external with a JSON file for Cinder  https://review.openstack.org/22487213:33
openstackgerritRyan Hallisey proposed openstack/kolla: Replace config-external with a JSON file for Neutron (thin)  https://review.openstack.org/22482913:33
openstackgerritRyan Hallisey proposed openstack/kolla: Replace config-external with a JSON file for Nova  https://review.openstack.org/21850513:33
rhalliseyahh my ears13:33
*** jtriley has joined #kolla13:36
*** openstackgerrit has quit IRC13:46
*** openstackgerrit has joined #kolla13:46
jmccarthySamYaple: Do you mean literally ansible_ssh_user=root for hosts in the inventory file ? (for example vs granting the kolla user sudo rights on the hosts in question)13:50
*** dims_ has quit IRC14:20
*** dims_ has joined #kolla14:21
*** sdake has joined #kolla14:27
*** banix has quit IRC14:29
*** banix has joined #kolla14:42
openstackgerritSteven Dake proposed openstack/kolla: Add a customize deployment document  https://review.openstack.org/22484714:44
sdakemorning14:44
rhalliseymorning14:44
*** dims_ is now known as dimsum__15:03
*** shardy is now known as shardy_afk15:03
*** tfukushima has joined #kolla15:13
*** tfukushima has quit IRC15:18
*** tfukushima has joined #kolla15:21
*** tfukushima has quit IRC15:23
openstackgerritHui Kang proposed openstack/kolla: Fix tags declaration in site.yml  https://review.openstack.org/22519615:25
*** blakec has joined #kolla15:34
openstackgerritHui Kang proposed openstack/kolla: Fix tags declaration in site.yml  https://review.openstack.org/22520215:35
*** blakec has quit IRC15:46
*** tfukushima has joined #kolla15:47
*** akwasnie has quit IRC15:49
*** blakec has joined #kolla15:49
*** banix has quit IRC15:52
*** tfukushima has quit IRC15:55
*** unicell has quit IRC16:09
*** blakec has quit IRC16:09
*** mduggan has quit IRC16:13
*** mduggan has joined #kolla16:14
*** mfalatic has quit IRC16:17
*** tfukushima has joined #kolla16:19
*** tfukushima has quit IRC16:20
*** mfalatic has joined #kolla16:20
*** shardy_afk is now known as shardy16:24
*** mfalatic_ has joined #kolla16:26
*** mfalatic has quit IRC16:30
*** jmccarthy has quit IRC16:32
*** mfalatic_ is now known as mfalatic16:37
*** athomas has quit IRC16:45
*** unicell has joined #kolla16:45
*** blahRus has joined #kolla16:56
*** achanda has joined #kolla17:01
*** absubram has joined #kolla17:09
openstackgerritMerged openstack/kolla: Fix ring bulder command examples in swift readme doc  https://review.openstack.org/22494617:18
*** jerryz has joined #kolla17:23
*** achanda has quit IRC17:30
sdakepbourke_ around this fine morning?17:38
*** loth has joined #kolla17:40
sdakenote today is deadline for moving all blueprints that are not in good progress to mitaka17:41
sdakeI am giong to do that work at noon PST, so please get your blueprint information updated if you think your blueprints will make the 25th17:42
*** gfidente has quit IRC17:44
*** blahRus has quit IRC17:49
*** dwalsh has quit IRC17:49
*** blahRus has joined #kolla17:50
*** achanda has joined #kolla17:52
openstackgerritMerged openstack/kolla: Spelling corrections in cinder guide  https://review.openstack.org/22499417:54
openstackgerritMerged openstack/kolla: Spell corrections in AIO doc  https://review.openstack.org/22499517:54
openstackgerritMerged openstack/kolla: Spell corrections in dev quickstart guide  https://review.openstack.org/22499617:55
openstackgerritMerged openstack/kolla: Spell corrections in Image building guide  https://review.openstack.org/22499717:56
sdakeSamYaple would you midn reviewing https://review.openstack.org/#/c/225202/ for correctness18:01
sdakeit looks good to me, but I'm not certain if there is some reason the tags are specified as they are, or if what is there is just a bug18:01
openstackgerritJeff Peeler proposed openstack/kolla: Add Ansible support for Ironic  https://review.openstack.org/21974718:02
sdakejpeeler woudl you mind giving a review of https://review.openstack.org/#/c/224847/18:05
sdakedont just rubber stamp but check for correctness and coherency :)18:05
sdakerhallisey ^^ as well?18:05
* sdake wants m0ar docs18:06
*** dwalsh has joined #kolla18:06
sdakerhallisey ping re the json pathset18:06
rhalliseysdake, hey18:19
sdakeso on a scale of 1 to 10 how well did you test the latet patch set18:19
rhalliseyI haven't yet it's still building18:19
sdakei see18:19
rhalliseybuild should be done soon18:19
rhalliseywhy is it totally bust?18:20
sdakeno idea18:20
sdakethat is why I asked :)18:20
rhalliseyya I haven't tested18:20
jerryzsdake: i know this is a corner case but i yesterday experienced a power down of the physical server and some of the httpd related services couldn't start because they thought httpd (pid 1) already running18:22
sdakejerryz i responded in irc to that18:22
sdakewe know about that problem18:22
jerryzsdake: ok18:22
sdakethe person working on it seems to have not followed up on teh patch18:22
sdakethere is a bugfiled for it iirc18:22
jerryzsdake: also in such circumstances, the orchestration can't be guaranteed.18:23
sdakewell I think it can be guranteed, just not wit hthe curren timplementation :)18:24
jerryzsdake: some services that relies on mysql connection will hang there after attempts to connect to mysql even if mysql was up later18:25
*** shardy has quit IRC18:26
sdakejerryz how did you have this mysql problem yo udescribed?18:28
jerryzsdake: mariadb was up later than nova conductor, consoleauth18:29
*** dwalsh has quit IRC18:34
jerryzsdake: nova conductor would give up aftre 10 attempts18:37
*** unicell has quit IRC18:37
jerryzsdake: roughly 2 mins18:38
*** unicell has joined #kolla18:40
jerryzsdake: and nova conductor daemon wouldn't exit so docker daemon won't restart the container18:45
*** banix has joined #kolla18:47
rhalliseysdake, ok test worked.  Just need to tweak libvirt templat18:55
rhalliseye18:55
*** achanda has quit IRC18:57
sdakejerryz I am curious why the nova condcutor daemon shoud have exited?19:04
sdakejerryz what was your expectation of how the system should have behaved compared to how it acctually behaved?19:04
sdakerhallisey sounds good19:05
jerryzsdake: if conductor just gave up after 10 attempts to connect mysql, then there is no point for it to stay there19:05
sdakeis this using sam's refactor of set_ocnfigs rhallisey ?19:05
rhalliseyya19:05
rhalliseyI rebased on top19:05
sdakeso the ocnductor gives up on mysql but then it doesn't exit?19:06
jerryzsdake: it appears to me so19:06
sdakejerryz would you do me a super solid and file a bug for that please?19:06
sdakethat sounds like a serious problem19:06
jerryzsdake: otherwise, docker daemon would try to restart it right?19:06
sdakei would expect it to keep retrying to connect to the db19:06
sdakeor alternatvely exit if its hit a end of the road19:07
sdakeis there a retry limit or something?19:07
jerryzsdake: seems like an oslo db issue19:08
sdakejerryz http://docs.openstack.org/havana/config-reference/content/section_configuring-database-connections.html19:10
sdakemax_retries = 1019:10
sdakeperhaps it shoul dbe set to -1 as a default in kolla19:10
sdakejerryz how precisely did you get into a situation where the database limit was hit as you described?19:11
jerryzsdake: a server reboot19:11
*** jtriley_ has joined #kolla19:11
jerryzsdake: and mariadb is behind haproxy19:12
sdakehow many nodes did you have in your enviornment?19:12
jerryzsdake: it is unclear to me when haproxy deemed mariadb back online19:12
jerryzsdake: just one controller and one compute19:13
sdakewhich did you reboot, the compute node?19:13
jerryzsdake: controller19:14
sdakei na proper deployment that wouldn't have happened becuase you would have 3 control nodes i think19:14
*** jtriley has quit IRC19:14
jerryzsdake: it is another scenario19:15
sdakebut we should hve max_retries at -1 anyway19:16
jerryzsdake: how about retry interval?19:17
sdakewell that could be cranked up a bit19:17
sdake10 seconds is pretty short19:17
sdakeif max_retries = -1, it will just retry forever19:18
sdakewhat should happen is if the max_retries limit is hit, the service shouldexit19:18
sdakebut clearly that isn't happening19:18
jpeelertests are failing because horizon requires a later RPM version than what epel provides. i assume that would normally be put in delorean19:18
sdakejpeeler huh?19:19
jerryzsdake: in oslo messaging, if iirc, the retry interval would increase after several short attempts19:19
jpeelersdake: the non-voting gates, that's why they are failing right now19:19
jerryzsdake: oslo db seems to use a fixed interval19:19
sdakerequires alater rpm version of which package jpeeler?19:19
jpeelerRequires: python-django-pyscss >= 2.0.2 Available: python-django-pyscss-1.0.5-2.el7.noarch (epel)19:20
sdakejpeeler thanks for hunting that down, i wa going to look into that19:20
openstackgerritRyan Hallisey proposed openstack/kolla: Replace config-external with a JSON file for Glance  https://review.openstack.org/21850119:20
openstackgerritRyan Hallisey proposed openstack/kolla: Replace config-external with a JSON file for Keystone  https://review.openstack.org/21849719:20
openstackgerritRyan Hallisey proposed openstack/kolla: Replace config-external with a JSON file for Heat  https://review.openstack.org/22489119:20
openstackgerritRyan Hallisey proposed openstack/kolla: Replace config-external with a JSON file for Rabbitmq  https://review.openstack.org/21849619:20
openstackgerritRyan Hallisey proposed openstack/kolla: Replace config-external with a JSON file for Cinder  https://review.openstack.org/22487219:20
openstackgerritRyan Hallisey proposed openstack/kolla: Replace config-external with a JSON file for Neutron (thin)  https://review.openstack.org/22482919:20
openstackgerritRyan Hallisey proposed openstack/kolla: Replace config-external with a JSON file for Nova  https://review.openstack.org/21850519:20
jpeelerRequires: python-scss >= 1.3.4 Available: python-scss-1.2.1-1.el7.x86_64 (epel)19:20
rhallisey^ those are all set19:20
sdakerhallisey grats on learning how to use git rebase -i ;)19:20
sdakeit will make you much more productive now :)19:20
rhalliseyyes19:21
rhallisey:)19:21
jpeelersdake: might be best to disable checking of horizon until that RPM gets added (i think that's the proper fix)19:21
sdakejpeeler cool can you fix that up and we can get it merged?19:21
jpeeleri guess so heh19:22
sdakejpeeler did you inform the delorean cats about the repo problem?19:28
openstackgerritJeff Peeler proposed openstack/kolla: Disable horizon build checking for now  https://review.openstack.org/22528619:29
jpeelersdake: no, am i right about it needing to be put in delorean?19:30
sdakeyes clearly something needs to provide that requirement :)19:31
sdakei'm not sure if its delorean or what19:31
sdakebut starting a discussion on the mailing list would probably be helpful to get it fixed19:31
sdakei know why that change is needed, - the current containers install pycss from pip19:31
sdakehorizon can't compress correctly19:31
sdakerhallisey can you ack jpeeler's change please19:31
jpeelerhrm, well i didn't look too closely at horizon. if we're doing something weird, maybe it's not delorean's fault?19:32
sdakejpeeler the pip install occurs after the yum install19:32
sdakeit is failing on the yum install line19:32
jpeelerah ok19:33
sdakethe pip install just installs the latesst via pip packaging19:33
jpeeleris it safe to pip upgrade a RPM installed package?19:34
sdakeymmv19:34
sdakeit works for hoirzon19:34
jpeeleryeah i bet it practically works, just not sure how pip upgrades work19:35
sdakei knew it would be temporary and we would remove it19:35
sdakeso i went with it since its temporary and without it horizon was bustola19:35
sdakeideally we wont be using pip installs with rpm packaged distros19:35
*** absubram has quit IRC19:40
*** smeisner|afk is now known as smeisner19:45
*** sdake has quit IRC19:46
jpeelerwhere did sdake goooooooooooooooo19:53
jpeelerdoes anybody know why we're using kilo packages in kolla master?19:53
*** sdake has joined #kolla19:53
sdakejpeeler can you ack https://review.openstack.org/#/c/223963/19:54
jpeelersdake: why are we using kilo packages in kolla master?19:55
sdakeopenvswitch comes from that repo19:55
sdakedelorean requires installing kilo repo for dependencies19:55
sdakeits right in the documentation on trunk website19:55
sdakei dont know why delorean doesn't just keep a copy of those files directly in its repo19:56
sdakemaybe a maintenance thing19:56
sdakeas in trying to reduce duplication of work19:56
sdakeonce liberty is released in rdo we will have to get rid of the delorean reposentirely in the branch i make on the 25th19:57
sdakecoolsvap around?19:57
openstackgerritDavid Moreau Simard proposed openstack/kolla: Add missing dependency repository for Liberty for EL  https://review.openstack.org/22529419:57
sdakejpeeler do you know this david fellow?19:58
sdakei wonder if that fixes the horizon problem19:58
jpeelerit does19:58
jpeelerthat's what we're talking about19:58
sdakedamn19:58
jpeelerin #rdo19:58
sdakewell i guess we need to revert that horizon patch19:59
jpeeleror at least, that's what he claims19:59
jpeeleronce we see that it passes, we can19:59
jpeelerif not, best to let it go ahead in19:59
sdakerhallisey can you remove your workflow +1 on https://review.openstack.org/#/c/225286/20:00
jpeeleror you could just -2 and it hope the other patch works20:00
openstackgerritDavid Moreau Simard proposed openstack/kolla: Add missing dependency repository for Liberty for EL  https://review.openstack.org/22529420:04
jpeelerwow he actually followed up on it20:05
*** jtriley_ has quit IRC20:07
*** jtriley has joined #kolla20:08
openstackgerritMerged openstack/kolla: Fix broken centos-source- build: add rdo repository  https://review.openstack.org/22396320:10
openstackgerritSteven Dake proposed openstack/kolla: Remove pip install of pyscss in Horizon  https://review.openstack.org/22530120:11
jpeelershould have made that a dependency technically20:12
sdakeya let me edit it20:12
openstackgerritSteven Dake proposed openstack/kolla: Remove pip install of pyscss in Horizon  https://review.openstack.org/22530120:13
sdakejpeeler how does that look20:13
jpeeleri've never seen that before "depends-on"?20:14
jpeeleri meant a gerrit dependency20:14
jpeelerapparently depends-on is used for cross-repo depends because gerrit doesn't do that20:15
jpeelerit'll probably do "the right thing", but the easy way would have been to pull down the other change in your repo and submit.20:22
sdakeok moment illl do that20:24
jpeelerit's no big deal20:26
openstackgerritSteven Dake proposed openstack/kolla: Remove pip install of pyscss in Horizon  https://review.openstack.org/22530120:26
openstackgerritSteven Dake proposed openstack/kolla: Add missing dependency repository for Liberty for EL  https://review.openstack.org/22529420:26
openstackgerritSteven Dake proposed openstack/kolla: Remove pip install of pyscss in Horizon  https://review.openstack.org/22530120:26
openstackgerritSteven Dake proposed openstack/kolla: Add missing dependency repository for Liberty for EL  https://review.openstack.org/22529420:26
sdakethere ya go jpeeler :)20:26
jpeeleryeah, it reset the tests though :/ the downside20:27
jpeelerand apparently it changed the committer, which i did not think would happen20:27
jpeelerso maybe i should have kept my mouth shut20:27
jpeelerand i need to focus on ironic. running out of time20:28
sdakejpeeler reading thread moment20:29
sdakejpeeler i'll accept ironic after 25th20:29
sdakeit looks close enough to being solid20:30
jpeeleroh thanks :) but either way, still need to get it done today20:30
sdakei just dont want people starting new work today and expecting it to hit the repo for the rcs :)20:30
sdakewell we have a full week left before the deadline20:30
jpeelerwhat did you think about the templating of configuration comments?20:31
sdakejpeeler which templating are you talking about20:33
sdakeI was just reading the thread about trivial changes20:33
sdakeadrian's suggestion seems solid20:34
sdakei'll bring it up next wekenesday and we can see if folks are open to relaxing the rules20:34
sdakeI dont mind given adrian's suggested approach20:34
jpeeleryeah i'm glad that thread came up...20:34
jpeeleri was talking about https://review.openstack.org/#/c/224847/20:34
sdakeoh reading your review moment20:35
sdakeya so the "requires much more work" is the thing I am avoidant about :)20:37
sdakethe docs have to be referenced anyway in most cases20:37
sdakeso that isn't really a good enough tradeoff in my opino20:37
sdakeidont htink it conflicts though20:37
sdakeas in "somehow conflicts with additoina lcustom options" I dont think I made that argument?20:38
sdakejpeeler ^^20:38
jpeelerok well, i guess i'm in favor of documenting certain options that have limited choices20:38
jpeelerbut i did finish with a "maybe I'm wrong"20:39
sdakesam takes a pretyt hardline approach on this20:39
sdakehis statement was you can't have it both ways20:40
sdakeI feel you can20:40
sdakebut just because we can doesn't mean we should :)20:40
sdakeI htink we can make things simple but in very limited cases template options20:40
sdakefor example, I think at  some point we will want to template MTU20:40
sdakebecause it will need to be applied and calculated and we don't necessarily want to have the operator figure out the 5-10 config options required to change mtu globally for a deployment20:41
sdakebut for something simple like virt type, the operator can go look at the upstream docs, rather then us documenting the same thing over agian20:41
jpeelermaybe i still don't understand "templating"20:42
sdaketake that virt tyep as an example20:43
jpeelerbasically defining the available choices in textual form, right?20:43
sdaketempatling tha towuld be adding to globals.yml20:43
jpeelerright20:43
sdakea nova_virt_type variable20:43
sdakeand then that would get jammed into the config20:43
jpeeleri guess it is hard to know where to draw the line - wouldn't want kolla doing it for every variable ever20:44
sdakeyup and that is what it leads to20:44
jpeelerbut drawing attention to some i think will be helpful20:44
sdakeI think in very limited circumstances we can do so20:44
sdakelike mtu20:44
sdakethere are some things we can't ovveride in kolla's system20:44
sdakelike how haproxy has timings for ha detection20:45
sdakethe problem is once we template it, its permanently in thesystem as a config option20:45
jpeeleryeah. when you correct the spelling errors my -1 will go away and i won't contest it20:45
sdakemore config options = more complexity for the operator to sort out20:45
sdakeatm kolal is dead simple20:45
sdakeI suspect over time we will come to a general agreement about what we are going to template20:46
jpeelerbut then you're relying on a bunch of default config options20:46
sdakebut I'm going to be pushing back pretty hard in almost allcases :)20:46
sdakethe 2 I am in support of now are the haproxy timeouts and MTU20:46
jpeelerhave you looked at openstack-ansible?20:47
sdakelooked about 1 year ago20:47
sdakeor maybe 8 months or so20:47
jpeeleri'm curious what they do. i know their configuration is located at the service level20:47
sdakehave also looked at bluebox20:47
sdakethey template everything20:47
sdakethey have like a million options20:47
sdakethey both operate the same way wit hteh templating20:48
jpeelerok20:48
sdakethe problem is the config options maybe deprecated20:48
sdakeor be added20:48
sdakeyou end up with hundreds of config options20:49
jpeeleri mean, there's definite maintenance burden20:49
sdakeand it requires a dev/test cycle to change things20:49
sdakeif I was an operator I wouldn't want ot rely on a dev/test cycle and a variable and subjective approval process for customizing my deployment20:49
sdakei'd just wan to jam thevariable in20:50
jpeeleri think i'd be satisified if the original config file contents remained from the RPM commented out20:50
sdakejpeeler what do yo umean - don't understand20:51
jpeelerthe config file that ends up on disk is only what you put in it. i don't honestly understand how the merging works20:52
jpeelerbut i suspect the solution would be to populate the 0 length files in the tree with the contents of the RPM20:52
jpeeler(in etc/kolla/config)20:54
sdakethe merging is necessary because otherwise the operator has to figure out how to get the out of the box configright which wouldbe very challenging20:54
jpeeleri know why you have to set the options20:55
jpeeleri'm just saying it's a bare bones file instead of containing the original RPM contents20:55
jpeeleri think the merging should be done with that20:55
sdakeyou mean we should merge with the rpm dist config?20:56
sdakeI would like to completely remove the dist config loading20:56
sdakei just htink we are short on time to do it this cycle ;)20:56
jpeeleri know it can't be done now. why do you mean completely remove the dist config loading?20:57
jpeeler*what20:57
*** blakec has joined #kolla21:01
openstackgerritSteven Dake proposed openstack/kolla: Add a customize deployment document  https://review.openstack.org/22484721:12
sdakejpeeler have a look at that21:12
sdakejpeeler what I mean is at the moment for example if you load glance, glance has a config file call glance-dist.ini21:13
sdakeor glance-dist.conf I think21:13
sdakeand it gets loaded into glance as a configuration file21:13
jpeeleri thought that the dist file got merged into the main glance configuration file21:13
sdakeit does, but I wan t to remove it completely21:14
sdakethat way there is no variance between distros21:14
jpeelerthat's fine21:14
sdakei want to specify --config-file xyz21:14
jpeelerjust keep the vanilla upstream config file21:14
jpeelerthen it's consistent and complete too21:14
jpeelerso i'm mistaken about the original files being in the RPM. they are merged at build time, so the RPM doesn't help21:17
jpeeleri guess what i want is only possible with source builds then21:17
*** dayou has quit IRC21:19
sdakewhat is it you want exaclty?21:22
jpeelerall the potential available options to be shown21:23
jpeelermaybe since they'd have to be configured trough ansible anyway it doesn't matter to have them in the container21:23
*** diogogmt has joined #kolla21:29
*** dims_ has joined #kolla21:29
*** dimsum__ has quit IRC21:32
*** jtriley has quit IRC21:47
*** blakec has quit IRC21:49
sdakejpeelerit hink that line of thinking is correct21:50
*** blakec has joined #kolla21:51
*** jpeeler has quit IRC22:05
openstackgerritHui Kang proposed openstack/kolla: Fix a small bug of wrong path to validate-docker-execuate  https://review.openstack.org/22535422:21
*** sdake_ has joined #kolla22:40
*** sdake has quit IRC22:43
*** openstackgerrit has quit IRC22:46
*** openstackgerrit has joined #kolla22:46
*** sdake has joined #kolla22:48
*** sdake_ has quit IRC22:51
*** achanda has joined #kolla22:52
*** dims_ has quit IRC22:56
*** banix has quit IRC23:10
*** banix has joined #kolla23:11
*** blahRus has quit IRC23:22
*** dimsum__ has joined #kolla23:23
*** weiyu has joined #kolla23:34

Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!