Monday, 2016-02-22

*** jtriley has quit IRC00:09
*** sdake has joined #kolla00:23
*** sdake has quit IRC00:23
*** subscope has quit IRC00:27
*** achanda has quit IRC00:36
*** Jeffrey4l has quit IRC00:39
*** salv-orl_ has joined #kolla00:42
*** salv-orlando has quit IRC00:45
*** iceyao has joined #kolla01:00
*** salv-orl_ has quit IRC01:00
*** iceyao_ has joined #kolla01:10
*** iceyao__ has joined #kolla01:11
*** iceyao_ has quit IRC01:11
*** iceyao has quit IRC01:14
*** tfukushima has joined #kolla01:17
*** alisonh has quit IRC01:30
*** alisonh has joined #kolla01:35
*** iceyao__ has quit IRC01:36
*** achanda has joined #kolla01:37
*** iceyao has joined #kolla01:37
*** iceyao_ has joined #kolla01:42
*** achanda has quit IRC01:42
*** iceyao has quit IRC01:45
*** achanda has joined #kolla01:52
*** iceyao has joined #kolla02:03
*** iceyao_ has quit IRC02:06
*** Liuqing has joined #kolla02:06
*** salv-orlando has joined #kolla02:38
*** salv-orlando has quit IRC02:41
*** Marga_ has quit IRC02:51
*** salv-orlando has joined #kolla02:55
*** salv-orlando has quit IRC03:01
*** vhosakot has joined #kolla03:02
*** vhosakot has quit IRC03:05
*** Jeffrey4l has joined #kolla03:12
*** yuanying has quit IRC03:20
*** Marga_ has joined #kolla03:28
*** Marga_ has quit IRC03:29
*** Marga_ has joined #kolla03:29
*** klint has joined #kolla03:33
openstackgerritLingfeng Xiong proposed openstack/kolla: Multiple hypervisor driver support for nova  https://review.openstack.org/28294303:38
*** Marga_ has quit IRC03:44
*** Marga_ has joined #kolla03:45
*** Marga_ has quit IRC03:49
*** salv-orlando has joined #kolla03:57
*** chandankumar has joined #kolla04:01
*** Liuqing has quit IRC04:03
*** salv-orlando has quit IRC04:04
*** Marga_ has joined #kolla04:10
*** yuanying has joined #kolla04:10
*** iceyao_ has joined #kolla04:18
*** Marga_ has quit IRC04:20
*** iceyao has quit IRC04:22
*** dave-mcc_ has joined #kolla04:24
*** dave-mccowan has quit IRC04:25
*** dave-mcc_ has quit IRC04:28
*** tfukushima has quit IRC04:34
*** dave-mccowan has joined #kolla04:40
openstackgerritLingfeng Xiong proposed openstack/kolla: Multiple hypervisor driver support for nova  https://review.openstack.org/28294304:50
*** Marga_ has joined #kolla05:05
*** Allen_Gao has quit IRC05:17
*** chandankumar has quit IRC05:19
*** Allen_Gao has joined #kolla05:21
openstackgerritLingfeng Xiong proposed openstack/kolla: Allow cinder-volume to be configured to use nfs  https://review.openstack.org/27991305:28
*** Marga_ has quit IRC05:30
*** tfukushima has joined #kolla05:32
*** chandankumar has joined #kolla05:32
*** dave-mccowan has quit IRC05:37
*** iceyao_ has quit IRC05:40
*** iceyao has joined #kolla05:40
*** Liuqing has joined #kolla05:44
*** daneyon_ has quit IRC05:45
*** mbound has joined #kolla05:46
*** tpot has quit IRC05:49
*** Liuqing has quit IRC05:59
openstackgerritLingfeng Xiong proposed openstack/kolla: Allow cinder-volume to be configured to use nfs  https://review.openstack.org/27991306:02
*** mbound has quit IRC06:08
*** salv-orlando has joined #kolla06:11
*** salv-orlando has quit IRC06:14
*** Marga_ has joined #kolla06:22
openstackgerritLingfeng Xiong proposed openstack/kolla: Multiple hypervisor driver support for nova  https://review.openstack.org/28294306:28
*** alger has joined #kolla06:29
*** Liuqing has joined #kolla06:38
*** alger has quit IRC06:43
*** alger has joined #kolla06:43
*** Liuqing has quit IRC06:45
*** jasonsb has joined #kolla06:52
openstackgerritLingfeng Xiong proposed openstack/kolla: fix neutron-dhcp-agent start failure  https://review.openstack.org/28297106:52
*** salv-orlando has joined #kolla06:54
*** salv-orlando has quit IRC07:02
*** alger has quit IRC07:28
*** jasonsb has quit IRC07:33
*** jasonsb has joined #kolla07:36
*** salv-orlando has joined #kolla07:40
*** mbound has joined #kolla07:40
*** alger has joined #kolla07:45
*** mbound has quit IRC07:46
*** subscope has joined #kolla07:51
*** subscope has quit IRC07:51
*** subscope has joined #kolla08:07
*** cui has joined #kolla08:11
*** akwasnie has joined #kolla08:14
*** macsz has joined #kolla08:15
*** mikelk has joined #kolla08:17
*** subscope has quit IRC08:17
cuiHi, can i deploying nova-compute with kolla now?08:21
*** jmccarthy has quit IRC08:24
*** alger has quit IRC08:24
*** salv-orlando has quit IRC08:25
*** subscope has joined #kolla08:26
*** iceyao_ has joined #kolla08:28
*** iceyao has quit IRC08:28
elemoine_cui: yes, kolla does support deploying nova-complute08:30
ajafoo/08:31
*** athomas has joined #kolla08:36
*** subscope has quit IRC08:40
cuielemoine_:thanks! but as i know,iscsid can't work in the docker: https://bugzilla.redhat.com/show_bug.cgi?id=110291108:40
openstackbugzilla.redhat.com bug 1102911 in iscsi-initiator-utils "can't run iscsid in a docker container" [Unspecified,Closed: duplicate] - Assigned to cleech08:40
cuihttps://bugzilla.redhat.com/show_bug.cgi?id=110000008:41
openstackbugzilla.redhat.com bug 1100000 in iscsi "can't run iscsid in a docker container" [Unspecified,Assigned] - Assigned to cleech08:41
*** alger has joined #kolla08:41
*** achanda has quit IRC08:42
*** jmccarthy has joined #kolla08:44
*** achanda has joined #kolla08:46
cuielemoine_:How did kolla resoved this issue?08:49
*** achanda has quit IRC08:58
openstackgerritMichal Rostecki proposed openstack/kolla-mesos: [WIP] Add openvswitch support  https://review.openstack.org/28238409:00
*** subscope has joined #kolla09:00
akwasniemorning guys09:04
nihiliferhi akwasnie09:04
*** iceyao_ has quit IRC09:04
*** iceyao has joined #kolla09:05
*** iceyao has quit IRC09:08
*** iceyao has joined #kolla09:08
mdnadeem.09:08
elemoine_morning09:11
*** bmace has quit IRC09:15
*** alger has quit IRC09:17
*** tzn has joined #kolla09:23
*** subscope has quit IRC09:25
*** athomas has quit IRC09:27
*** subscope has joined #kolla09:29
*** pbourke has joined #kolla09:30
*** iceyao_ has joined #kolla09:38
*** iceyao has quit IRC09:38
openstackgerritMD NADEEM proposed openstack/kolla: Swift ubuntu binary container  https://review.openstack.org/27729409:39
*** athomas has joined #kolla09:42
*** athomas has quit IRC09:48
*** athomas has joined #kolla09:51
*** cui has quit IRC10:02
*** iceyao_ has quit IRC10:11
*** iceyao has joined #kolla10:11
elemoine_hi akwasnie I'll open a Gerrit CR setting up Heka/ES integration this week10:17
akwasniehi elemoine_:great, please let me know if you need any help10:22
*** salv-orlando has joined #kolla10:26
elemoine_I hope we can collaborate on this10:28
openstackgerritMichal Rostecki proposed openstack/kolla-mesos: Add openvswitch support  https://review.openstack.org/28238410:29
*** salv-orlando has quit IRC10:30
*** bmace has joined #kolla10:39
akwasnieelemoine_:yes sure, looking forward to10:40
*** subscope has quit IRC10:57
openstackgerritEric Lemoine proposed openstack/kolla: Run yum clean all in Heka Dockerfile  https://review.openstack.org/28303410:59
*** iceyao has quit IRC10:59
*** akwasnie has quit IRC11:01
*** tzn has quit IRC11:10
*** tfukushima has quit IRC11:16
*** akwasnie has joined #kolla11:24
*** subscope has joined #kolla11:30
*** salv-orlando has joined #kolla11:33
*** gfidente has joined #kolla11:37
openstackgerritMerged openstack/kolla-mesos: Updated from global requirements  https://review.openstack.org/28274611:39
*** elemoine_ has quit IRC11:40
*** elemoine_ has joined #kolla11:40
*** zigo has quit IRC11:40
*** zigo has joined #kolla11:41
*** salv-orlando has quit IRC11:43
*** subscope has quit IRC11:45
*** subscope has joined #kolla11:45
*** chandankumar has quit IRC12:00
*** limamauricio has joined #kolla12:03
*** liyi has joined #kolla12:05
liyiafter deploy, I try to login with admin:password, keep getting "  Unauthorized. Please try logging in again."12:06
liyianyone experiencing similar problem12:06
limamauricioi have this error when i try to build centos/binary http://pastebin.com/eRwUnu2V12:10
*** tzn has joined #kolla12:10
*** tzn has quit IRC12:11
pbourkelimamauricio: there is a review open for that issue12:13
limamauricio:)12:13
*** gfidente has quit IRC12:14
limamauriciopbourke, can you send me the review?12:14
pbourkehttps://review.openstack.org/#/c/282906/12:15
limamauricioThanks12:15
*** JoseMello has joined #kolla12:15
liyipbourke, this the change solve the problem in ubuntu as well?12:16
pbourkeliyi: good question12:16
pbourkeliyi: does ubuntu have the same problem?12:16
akwasniepbourke: I checked and there is no problem with ubuntu12:16
liyisorry guy, I mistake the topic12:18
liyiI thought you are talking about "  Unauthorized. Please try logging in again." as i am experiencing12:18
*** iceyao has joined #kolla12:19
liyi@akwasnie, are you able to deploy and use openstack on ubuntu at the moment?12:20
*** rhallisey has joined #kolla12:21
*** tzn has joined #kolla12:23
akwasnieliyi: testing right now12:24
limamauricioakwasnie you use ubuntu?12:24
liyi@akwasnie Thanks, looking forward to your good news12:25
*** stvnoyes has quit IRC12:35
*** chandankumar has joined #kolla12:36
*** stvnoyes has joined #kolla12:37
*** salv-orlando has joined #kolla12:40
*** salv-orlando has quit IRC12:44
*** chandankumar has quit IRC12:52
akwasnielimamauricio: yes12:54
limamauricioubuntu/binary or ubuntu/source?12:55
akwasnieubuntu/source12:55
limamauriciovocê pode iniciar uma instância com sucesso?12:56
limamauriciocan you launch a instance successfully?***12:56
openstackgerritEric Lemoine proposed openstack/kolla: Make cleanup-containers remove bootstrap_heka  https://review.openstack.org/28306512:57
openstackgerritEric Lemoine proposed openstack/kolla: Make cleanup-containers remove Heka volumes  https://review.openstack.org/28306613:00
akwasnielimamauricio: I am rebuilding everything right now13:01
limamauriciook13:01
openstackgerritDaniel Gonzalez Nothnagel proposed openstack/kolla: Change ubuntu base_image for vagrant to wily64  https://review.openstack.org/28196813:03
openstackgerritDaniel Gonzalez Nothnagel proposed openstack/kolla: Harmonise vagrant provider defaults  https://review.openstack.org/28306813:03
openstackgerritDaniel Gonzalez Nothnagel proposed openstack/kolla: Fix vagrant bootstrapping for ubuntu  https://review.openstack.org/28306913:03
openstackgerritEric Lemoine proposed openstack/kolla: Create "heka" group and add "heka" user to it  https://review.openstack.org/28307113:06
*** kproskurin has joined #kolla13:11
*** chandankumar has joined #kolla13:12
*** dwalsh has joined #kolla13:14
openstackgerritDaniel Gonzalez Nothnagel proposed openstack/kolla: Fix vagrant bootstrapping for ubuntu  https://review.openstack.org/28306913:26
openstackgerritDaniel Gonzalez Nothnagel proposed openstack/kolla: Harmonise vagrant provider defaults  https://review.openstack.org/28306813:26
openstackgerritDaniel Gonzalez Nothnagel proposed openstack/kolla: Change ubuntu base_image for vagrant to wily64  https://review.openstack.org/28196813:26
*** cui has joined #kolla13:29
*** chandankumar has quit IRC13:32
openstackgerritEric Lemoine proposed openstack/kolla: Remove duplicate code in nova extend_start.sh  https://review.openstack.org/28308213:43
*** salv-orlando has joined #kolla13:45
*** salv-orlando has quit IRC13:52
*** v1k0d3n has joined #kolla13:53
*** subscope has quit IRC13:56
openstackgerritEric Lemoine proposed openstack/kolla: Change Heka sudoers file mode bits  https://review.openstack.org/28308613:57
*** klint has quit IRC13:58
*** fthiagogv has joined #kolla14:02
*** macsz has quit IRC14:03
*** chandankumar has joined #kolla14:04
openstackgerritRyan Hallisey proposed openstack/kolla: Add a precheck for MountFlags=shared  https://review.openstack.org/28127914:06
*** subscope has joined #kolla14:09
*** dwalsh has quit IRC14:10
openstackgerritMichal Rostecki proposed openstack/kolla: Pin Mesos to 0.27.0 version  https://review.openstack.org/28309114:10
*** wadeholler has joined #kolla14:11
*** dwalsh has joined #kolla14:12
wadehollerhi all.  brand new user from a koala perspective.  trying to deploy and hit a early ansible error "cannot access /var/lib/mysql/mariadb.pid: No such file or directory"14:12
wadehollerill put the message on pastbin now14:12
*** dwalsh has quit IRC14:13
wadeholleransible error is @ http://pastebin.com/83jL0ege14:13
wadeholleranyone on that can help ?14:13
*** dave-mccowan has joined #kolla14:14
*** dwalsh has joined #kolla14:14
*** openstackgerrit has quit IRC14:17
*** openstackgerrit has joined #kolla14:17
*** blahRus has joined #kolla14:19
elemoine_wadeholler: what does "docker logs mariadb" tell you?14:21
*** sdake has joined #kolla14:21
sdakeyo14:21
sdakemasters of orion out next week14:21
* sdake taking the month off14:21
kproskurinOh it really depends it it will be the same MoO and not something like MoO314:26
wadehollerelemoine: - just looks like it is starting to me.  maybe start is hung ? http://pastebin.com/nxuVkJC914:26
elemoine_kproskurin: wrong channel?14:26
kproskurinIn a way :-)14:27
sdakeJeffrey4l ping14:30
elemoine_wadeholler: and when you do the operation manually: "docker exec mariadb ls /var/lib/mysql/mariadb.pid" ?14:30
wadehollereleomine: ls: cannot access /var/lib/mysql/mariadb.pid: No such file or directory14:31
rhalliseysdake, what the heck is masters of orion14:32
sdakedude14:32
sdakejust the best video game in the world14:32
sdakediablo like from the old days :)14:32
sdakebuild sapceships conquer the universe14:33
sdakeapparentl ythe new one is online14:33
sdakeever play "tradwars"14:33
sdakeits like a grphical version of that game14:33
wadehollereleomine: other files are there - mysql.sock,  gallera.cache, .....14:33
rhalliseysdake, I skipped a lot of 90's video games14:34
pbourkewadeholler: ansible is waiting for mariadb to start by checking for the pid file 6 times14:34
rhallisey:)14:34
elemoine_wadeholler: you use kolla master right?14:34
sdakeprobablybecause youwere 5 ;)14:34
pbourkewadeholler: I would check the mariadb logs to see why its failing to start14:34
rhalliseysdake, yes :).  The first game I played like crazy on pc was warcraft 314:35
sdakemariadb failing to start because of the pdi issue?14:35
wadehollerelemoine: yes. cloned down per the quick start on saturday14:35
sdakepid issue?14:35
pbourkewadeholler: if the mariadb container is running you can exec into it and look under /var/log14:35
elemoine_wadeholler: pbourke or look at /var/lib/docker/volumes/kolla_logs/_data/mariadb/mariadb.log14:35
sdakedocker exec first_3_numbers_of_container_id ls /var/lib/docker/volumes/kolla_logs14:35
pbourkeah yes I forgot elemoine_ is our logging expert now :)14:35
sdakei'd just run docker logs on it though14:36
sdaketo make sure its not the pid issue14:36
sdakebeen seeing alot of reports of that lately14:36
elemoine_sdake: wadeholler already provided a pastebin with the output of docker logs mariadb14:36
sdakeelemoine_ i was not here14:36
sdakewas it the PID issue?14:36
sdakethis happens over and over to people14:37
sdakei'm not sure what the resolution is14:37
pbourkedocker logs is not reliable in this case14:37
sdakebut we need a bug  filed!14:37
pbourkeactually no scratch that14:37
pbourkeignore me14:37
wadehollerend of mariadb.log http://pastebin.com/LhUztxb614:38
openstackgerritMerged openstack/kolla: Create "heka" group and add "heka" user to it  https://review.openstack.org/28307114:39
wadehollerI have re-ran the ansible deploy, docker stop, docker rm all containers, docker rmi mariadb, tools/build.py mariadb, re-ran deploy - same result...... what next ?14:39
pbourkeI wonder is it a case of a bad data vol left over from previous deploy14:41
*** salv-orlando has joined #kolla14:41
elemoine_wadeholler: are you running aio or multinode? (just for me to know)14:42
wadehollermultinode14:42
pbourkewadeholler: try copying tools/cleanup-containers to your target node(s)14:42
*** chandankumar has quit IRC14:42
pbourkerun that, then re-run ansible14:42
wadehollerwill do.14:42
sdakepbourke he didn't remove the named volumes14:43
sdakebut we should not have ot remove named volumes in order for a redeploy to work14:44
pbourkesdake: we shouldn't but I think in this case it might be the problem14:44
openstackgerritEric Lemoine proposed openstack/kolla: Run yum clean all in Heka Dockerfile  https://review.openstack.org/28303414:44
pbourkesdake: maybe not, it's all I've got right now :)14:44
*** jtriley has joined #kolla14:45
wadehollerp.s. - I think it doesn't matter but  for koala-toolbox I had to remove "mariadb-libs" from the list to get it to build.  otherwise it faulted with a yum conflict on my.cnf14:45
*** akwasnie has quit IRC14:45
pbourkewadeholler: that's correct, there is a patch for that14:45
wadehollerand I will eventually fix the spellchecker for koala -> kolla :)14:46
elemoine_wadeholler: yes that's  a known issue that is being addressed by https://review.openstack.org/#/c/282906/14:46
sdakepbourke link to patch please?14:46
elemoine_sdake: https://review.openstack.org/#/c/282906/14:46
wadehollerpbourke: it appears to be working, mariadb.pid exists and ansible is into the nova section now - THANKS!14:48
sdakewadeholler what did you do14:49
pbourkewadeholler: good to hear14:49
sdakeok pbourke lets not treat the symptom, lets fix the problem ;)14:49
elemoine_wadeholler: was it your first deploy?14:49
sdakethis happens over and over - like 7+ times on irc in the last week from different individuals14:50
sdakeso lets understand what just worked, and why it worked14:50
sdakei'm out of the loop on what hte solution was14:50
sdakewould someone mind offering that up?14:50
sdakepbourke ^^14:50
sdakeI always hear "clean up your environment" - that is a really bad answer to redeploy14:51
pbourkesdake: when you say it's happened 7+ times over the past week, it may not be the same issue14:51
sdakemariadb.pid problem right?14:51
dave-mccowanhello ansible gurus... is there an ansible module that will simply concatenate two files?  I'm trying to avoid using "shell: cat file1 file2 > bigfile", but that is the exact functionality i want.14:52
SamYaplemorning14:53
pbourkesdake: mariadb failed to start. the only way ansible can tell this is by the lack of pid, so it shows as a series of 'ls /var/lib/mysql/mariadb.pid file not found' errors14:53
wadehollersdake: yes I think: stderr: "ls: cannot access /var/lib/mysql/mariadb.pid: No such file or directory" http://pastebin.com/83jL0ege14:53
SamYaplesdake: "cleanup you environemtn" is a great answer for _redeploy_14:53
pbourkesdake: why did it fail to start? that's the hard part, but from experience its due to a previous failure in the deploy process, which leaves the data vol in a state where it can't be reclustered14:54
pbourkesdake: the failure that left it in that state could be anything14:54
pbourkesdake: I understand trashing the env is not a good solution to the problem, I just also want to help new users get running quickly14:55
cuiHi,all. Anybody know if kolla resolved this issue "iscsid can't work in docker container"? https://bugzilla.redhat.com/show_bug.cgi?id=110000014:55
openstackbugzilla.redhat.com bug 1100000 in iscsi "can't run iscsid in a docker container" [Unspecified,Assigned] - Assigned to cleech14:55
cuiwhen deploying nova-compute14:55
SamYaplei mean for a full redploy you _do_ want a clean environment pbourke14:55
sdakein the field, do we expect people to onl yrun "deploy" once?14:56
pbourkealso SamYaple was not around and I don't know much about mysql :(14:56
SamYapleautomating a "cleaning" is a really bad idea (thats going to trash production environments"14:56
pbourkeSamYaple: no one is advocating that14:56
pbourkeunless I missed something14:56
SamYaplesdake has suggested we automatically fix those issues14:57
sdakepbourke i was not advocating deleting anything14:57
sdakeok lets come to a common definition of how "deploy" operation is supposed to work14:57
sdakein the past it has meant upgrade + reconfigure14:57
sdakesince we are breaking those steps apart14:57
sdakedo we expect then deploy to just happen once?14:57
ccesariois there planned any kind of UI (eg Kolla-UI) for manage/monitor the docker images/services!?14:58
sdakei am fine with deleting the environment on a failed deploy if it happens once14:58
SamYapledave-mccowan: hmmm i know of no simple task to do that. you hve a few options, 3 tasks : read file1, read file2, write bigfile14:58
SamYapledave-mccowan: option 2 is to create a simple module to do this14:58
sdakebut if kolla-ansible deploy happens over and over (as is the case now) on the same box to upgrade/reconfgiure, then we have a problem14:59
SamYaplesdake: unless something has changes, we said back in liberty if a bootstrap action fails it is expected you either rekick or reset just that one service14:59
SamYaplebootstrap action being key14:59
SamYaplethats a one time special action that happens once per environment14:59
sdakei understand how bootstrap works14:59
SamYapleno one said you didnt14:59
SamYaplemy point was _if_ that fails its quickest to have the user reset teh evnrionemtne14:59
sdakelet me explain how I have been using deploy14:59
sdakekolla-ansible deploy15:00
sdakehey look my environment is up, let me reconfigure15:00
SamYapletechnically they could fix the issue, but the path to quickest working is reset teh envrionemtn15:00
sdakekolla-ansible deploy15:00
sdakehey look I need to change something again15:00
sdakekolla-ansible deploy15:00
SamYaplethe issue here is your premises is a working kolla-ansible deploy the first time15:00
sdakeso we have this fixed with reconfigure15:00
SamYaplethats where you have you flaw in flow15:00
SamYaplethe initial bootstrap is failing15:00
sdakewell lets figure out what the flow should be and get it documented15:00
sdakealso initial bootstrap should not be failing imo ;-)15:01
SamYapleif bootstrap fails, rekick15:01
SamYapleit fails for alot of reasons aoutside our control15:01
SamYapleincluding user error _alot_15:01
dave-mccowanSamYaple as a code reviewer, which option would you prefer for https://review.openstack.org/#/c/282733/ ?  For this use case, maybe "shell: cat" is OK?15:01
SamYapledave-mccowan: any use of shell module will be met with -115:02
sdakeI don't like shell module15:02
SamYapleits a vary dangeours one we avoid at all costs15:02
sdakesecurity problem15:02
SamYaplethere is always a way around it15:02
sdakeif deploy is only meant to happen one time, we should not allow a second deploy to occur if the first one has proceeded in any way15:03
sdakebut i'm not hearing any answers on if deploy is meant to only happen once or not15:03
SamYapledeploy is idempotent15:03
sdakeit clearly is not idempotent15:03
sdakeyou mean the intent ?15:03
SamYapleit is if you look up what idempotent means15:04
sdakeI know what idempotent means15:04
SamYaplebut the issue is if it progresses to a certain point, it will think bootstrap is finished and not do it again15:04
SamYapleclearly not15:04
SamYapledeploy _is_ idempotent15:04
openstackgerritEric Lemoine proposed openstack/kolla: Fix Swift logging  https://review.openstack.org/28311815:04
sdakedoesn't deploy pull in new configs?15:05
SamYapleit will push new configs15:05
SamYapleit doesnt restart services15:05
sdakeneed a mullligan optoin ;)15:06
sdakeok I guess I misunderstood in thinking it restarted the services on config changes15:06
SamYapleif a bootstrap fails, theserivce needs to be cleaned up15:06
SamYapleeasiest way to do that is a full rekick15:06
SamYapleits not a requiremetns15:06
SamYaplebut its the fastest way to a working deploy15:06
pbourkethats kind of what I was getting at15:06
SamYaple^^15:06
SamYaplewe talked about this at lenght before liberty15:06
pbourkeideally it should never fail but there are many reasons it can, many not our fault15:06
sdakeok, well now I understand the problem, it can be documented15:07
sdakethe mariadb pid thing seems to havebeen failign alot recently15:07
sdakeas i said 7+ in the last week15:07
pbourkethere may be some common reasons its failing that we *can* address15:07
SamYaplesdake: thats because of partial deploys without full cleans15:07
sdakeits not a dirty environment, its a failed deploy15:07
SamYapleit was because of a dirty environemnt15:07
SamYapleit always is15:07
openstackgerritEric Lemoine proposed openstack/kolla: Fix Swift logging  https://review.openstack.org/28311815:08
sdakethe dirty environment is a failed deploy ?15:08
SamYapleno it failed because of dirty environment15:08
sdakehere is the order of operations15:08
sdakedeploy15:08
sdakedeploy fails15:08
SamYaplesomeone deployed, something broke or they cancelled it15:08
sdakedeploy again15:08
SamYaplethey redeploy and it fails15:08
sdakemariadb implodes15:08
SamYaplei have yet to see a non-user error problem cause this fyi15:08
sdakei'm not proposing to fix that15:08
SamYapleif the prechecks are run, then the deploy is run if it breaks it has always been user error15:09
sdakewhat user error15:09
sdakeVIP=local IP?15:09
*** inc0 has joined #kolla15:09
SamYapleuser had a bad configuration. user canceled the process15:09
*** inc0_ has joined #kolla15:09
SamYapleuser did this user did that15:09
SamYaplenot once have i found an issue with the process as it currently is (we had issue in the past, sure015:10
sdakehere is how i would handle these problems in the future on irc15:10
sdake"You deployed once, and it failed in some way"15:10
sdake"Kolla does not delete data on a failed deploy"15:10
sdake"the fastest way to get running again is clean up your environment and deploy again"15:10
sdake"we find deploy errors are nearly almost always operator error"15:10
sdakeexplain WHY15:11
SamYaplei have indeed done this15:11
SamYaplei agree with your approach15:11
pbourkesounds good to me15:11
sdakebecause without  the WHY they are left wondering if the code is solid enough to use in an eval15:11
sdakethanks pbourke  ;)15:11
sdakeSamYaple i msut  have missed the explination, but its good to repeat each time15:11
sdakeor even put in the documenation ;)15:11
*** chandankumar has joined #kolla15:12
sdakei'll write docs for that now15:13
SamYaplesdake: post a docs patch and ill approve!15:13
SamYapleyay15:13
pbourkesdake: I suspect the why is rooted in how the cluster is formed and how it doesn't recover well from being halted during that process15:13
*** inc0 has quit IRC15:13
*** inc0_ has quit IRC15:13
pbourkesdake: if you know anything about that process would definitely be great to get into doc form15:14
*** inc0 has joined #kolla15:14
sdakepbourke I don't know how that particular cluster forms15:14
*** inc0_ has joined #kolla15:14
pbourkedidnt you write it? :)15:14
sdakepbourke I think we need to figure out if there is still a problem with mariadb15:14
sdakepbourke mariadb + galera? no15:14
pbourkeah yeah you were corosync15:14
sdakei had help with corosync15:15
sdakebut most of totem I wrote myself15:15
*** gfidente has joined #kolla15:15
*** spisarski has joined #kolla15:15
sdakeand tested on thousands of customers ;-)15:15
*** jtriley has quit IRC15:15
sdakebtw our meetings are moved aroudn now15:16
sdakeI'm not sure which is an odd week and which is an even week15:16
sdakebut 23:00 is the new voted time15:16
*** inc0___ has joined #kolla15:16
SamYaplepbourke: I will tell you how!15:17
pbourkeSamYaple: i should read myself and then come back with questions15:17
SamYaplepbourke: so in /var/lib/mysql is a file called "galera.dat" it contains information about the cluster15:17
wadehollerok next prob. bootstrap_neutron failure  http://pastebin.com/wdeLqp1R  sorry to be so needy15:18
pbourkeSamYaple: but I can do the lazy way if you have time :p15:18
SamYaplepbourke: you need to find the galera.dat file with the highst seq number, thats your new master15:18
*** inc0 has quit IRC15:18
SamYaplepbourke: once you have that you need to start _taht_ node first15:18
*** inc0 has joined #kolla15:18
pbourkewadeholler: looks possible like a bug in recent logging changes. elemoine_ ?15:19
SamYapleto start that node you need to pass '--wsrep-new-cluster' or set the 'gcomm://' list to empty in the mysql.cnf15:19
*** inc0_ has quit IRC15:19
*** ayoung has joined #kolla15:19
SamYapleelemoine_: wadeholler pbourke that is a problem, ill submit a patch right now15:19
*** inc0 has quit IRC15:20
*** inc0___ has quit IRC15:20
elemoine_pbourke: looking15:20
SamYapleelemoine_: its a problem with the sudoers file15:21
elemoine_pbourke: you're right, that's certainly related to the recent logging changes15:21
SamYaplei suppose this wasnt tested split node15:22
SamYapleelemoine_: ive got a patch brewing15:22
elemoine_SamYaple: cool15:22
*** inc0 has joined #kolla15:24
inc0hey good morning15:25
openstackgerritSam Yaple proposed openstack/kolla: Fix sudoers files issue for neutron from Heka  https://review.openstack.org/28312715:25
SamYaplemornign inc015:25
openstackgerritJeff Peeler proposed openstack/kolla: Playbook for glance upgrade  https://review.openstack.org/27804015:26
elemoine_SamYaple: is tools/deploy_aio.sh used on the gates?  anywhere else?15:26
SamYapleelemoine_: only in gate15:27
*** iceyao has quit IRC15:27
sdakewhat removes the named volumes?15:27
sdakedo we have a script for that already?15:27
elemoine_sdake: cleanup_containers15:27
sdakecoool thanks15:27
*** inc0 has quit IRC15:28
*** inc0 has joined #kolla15:28
*** tzn has quit IRC15:29
wadehollerSamYaple: After adding your patch changes - do I just  remove my failed bootstrap container and re-run deploy or should I just kick the whole thing with cleanup-containers ?15:29
sdakeif your in eval, just do cleanup-containers15:29
wadeholleram in eval.  will do15:30
sdaketo kick the boottrap container you also need to remove the named volumes15:30
sdake(if there are any associated)15:30
inc0so anyone here wants an ATC? I think we still need migration script from data_containers to thin volumes15:30
openstackgerritMerged openstack/kolla: fix neutron-dhcp-agent start failure  https://review.openstack.org/28297115:32
openstackgerritSteven Dake proposed openstack/kolla: Add docs about failures during evaluation  https://review.openstack.org/28312915:33
*** limamauricio has quit IRC15:33
*** achanda has joined #kolla15:34
openstackgerritMerged openstack/kolla: Swift ubuntu binary container  https://review.openstack.org/27729415:35
*** dwalsh has quit IRC15:35
openstackgerritSteven Dake proposed openstack/kolla: Add docs about failures during evaluation  https://review.openstack.org/28312915:35
elemoine_SamYaple: regarding https://review.openstack.org/#/c/283127, I don't get it.  What code does "sudo chown neutron: /var/log/kolla/neutron"?15:35
SamYapleelemoine_: what do you mean? like what does chown do?15:37
sdakeSamYaple thanks for your comments15:37
sdakei'll add them in in a bit - school rush atm15:37
sdakewould you mind telling me, is there a kolla-ansible prechecks command?15:37
elemoine_SamYaple: I mean, I don't see who is the user of that sudo you added15:38
*** dwalsh has joined #kolla15:38
SamYapleelemoine_: i dont understand what yo uare asking. it was failing due to not having permissions right?15:38
SamYaplefyi im rebuilding my images right now to test this patch15:38
SamYapleyou may understand the issue better than me at this point, i just dont understand your question15:39
sdakeelemoine_ remember the 400 chmod15:39
sdake!!!15:39
openstacksdake: Error: "!!" is not a valid command.15:39
*** chandankumar has quit IRC15:39
sdakeif your sudo is failing it is because of the chmod 400 that is missing15:40
elemoine_sdake: I created a patch for this already15:40
elemoine_https://review.openstack.org/#/c/283086/15:40
sdakeare you running with it?15:40
sdakei missed your failure15:40
elemoine_not my failure, someone else's15:41
sdakewhat was the failure15:41
elemoine_SamYaple: you added "chown neutron: /var/log/kolla/neutron" to the neutron_sudoers file while I don't see any bash code in kolla that does "sudo chown neutron: /var/log/kolla/neutron"15:42
elemoine_sdake: http://pastebin.com/wdeLqp1R15:43
wadehollerI just tried to test and got same failure.  Was looking for the same thing - maybe that should be in the Dockerfile on the chown -R line ?15:43
sdake /var/log/kolla is a named volume15:44
sdakeits chown should happen in the start file15:44
*** achanda has quit IRC15:45
SamYapleelemoine_: i think youre right, but i wont be able to look into it for a few minutes15:45
SamYapleplease feel free to fix the issue the way you see fit15:46
elemoine_SamYaple: yes15:46
*** jtriley has joined #kolla15:47
*** inc0 has quit IRC15:49
elemoine_wadeholler: do you have the heka container running on that node?15:53
*** inc0 has joined #kolla15:53
wadehollerelemoine: yes15:54
*** mikelk has quit IRC15:56
elemoine_are you able to run "docker inspect bootstrap_neutron"?15:56
wadeholleryep - will pastebin it- standby15:57
*** inc0_ has joined #kolla15:57
elemoine_wadeholler: thanks15:58
openstackgerritSteven Dake proposed openstack/kolla: Add docs about failures during evaluation  https://review.openstack.org/28312915:58
*** inc0 has quit IRC15:59
wadehollerelemoine_: http://pastebin.com/37uJyMf416:03
elemoine_wadeholler: the problem is related to kolla_logs:/var/log/kolla not being mounted into the bootstrap_neutron container16:04
elemoine_but it should be: https://github.com/openstack/kolla/blob/master/ansible/roles/neutron/tasks/bootstrap.yml#L5316:04
openstackgerritMerged openstack/kolla: Playbook for glance upgrade  https://review.openstack.org/27804016:05
openstackgerritSteven Dake proposed openstack/kolla: Add docs about failures during evaluation  https://review.openstack.org/28312916:06
sdakeupgrades almost done !!16:07
inc0_sdake, not yet, on my todo list we still have all of infra and docs16:08
inc0_but I'll leave docs for post-M3 I guess16:08
elemoine_wadeholler: can you check the content of your ansible/roles/neutron/tasks/bootstrap.yml file, I really don't get it…16:08
inc0_or at least when we have rest done16:08
*** chandankumar has joined #kolla16:10
sdakeinc0_ we can sort out docs in the rcs16:10
wadehollerelemoine: checking...16:10
sdakeinc0_ I could live with just openstack upgrades for this cycle, but yup, lets try to get all the ugprades going ;)16:11
dgonzalezhi SamYaple, are you here?16:11
SamYapledgonzalez: i am here16:11
*** Allen_Gao has quit IRC16:12
inc0_sdake, let's try, if we will have problems we can focus on required infra upgrades, if there are any16:13
wadehollerelemoine_: http://pastebin.com/46vtsV9z16:13
dgonzalezhey, i am the author of https://review.openstack.org/#/c/283069. Just a short question: do you think i should also update the centos-bootstrap.sh script with [[]] instead of []?16:13
elemoine_wadeholler: so it's correct, as expected16:14
inc0_SamYaple, do you have your haproxy config changes that will change ports for air?16:14
inc0_aio16:14
elemoine_wadeholler: I guess that "docker volume ls" reports that the volume kolla_logs exists16:14
SamYapleinc0_: what do you mean? no changes are needed?16:14
SamYaplei run AIO with haproxy all the time...16:14
inc0_well if you run it on vm VIPs won't work16:15
inc0_on public cloud16:15
SamYapledgonzalez: yes, always use [[ ]]16:15
SamYapleinc0_: if you run AIO vips still work...16:15
openstackgerritEric Lemoine proposed openstack/kolla: Remove Rsyslog entirely  https://review.openstack.org/28315016:15
SamYaplethe gate is a VM running on AIO in a public cloud16:15
SamYapleit uses haproxy16:15
inc0_but I won't be able to connect to it, that's my point16:15
inc0_I don't have second IP for it16:16
SamYaplebut even running with haproxy and a dumb non-routable ip other address is still listening16:16
SamYapleim not in favor of having different oports like you asked for just now16:17
wadeholler[root@cpn00011 kolla]# docker volume ls DRIVER              VOLUME NAME local               kolla_logs16:17
SamYapleyou are welcome to set that up yourself with static mappings (or just not use haporxy)16:17
openstackgerritDaniel Gonzalez Nothnagel proposed openstack/kolla: Fix vagrant bootstrapping for ubuntu  https://review.openstack.org/28306916:17
inc0_SamYaple, it will be hart not to use haproxy in haproxy upgrades;)16:17
dgonzalezSamYaple: ok thanks, i just updated my commit16:17
inc0_I don't use haproxy otherwise16:17
elemoine_wadeholler: did you try "docker rm -f bootstrap_neutron" and redeploying?  not saying that's a solution, just wondering…16:18
wadeholleri used clean_containers16:18
SamYapleinc0_: well if you dont have an evironment to test the upgrades properly, you probaly shouldnt be doing them inc0_16:20
elemoine_wadeholler: do you see errors in the Docker daemon logs?16:20
wadehollertotally new to this -probably obvious- looking now -16:21
elemoine_wadeholler: or in /var/lib/docker/volumes/kolla_logs/_data/ansible.log?16:22
inc0_I will test them properly soon, my env is coming in soon I hope16:22
inc0_and I can test them properly, just let me make little patchiest16:22
*** Allen_Gao has joined #kolla16:23
SamYapleinc0_: hoenstly, i really dont think you need to make any changes to kolla to make this work....16:24
SamYaplejust use a 169.x.x.x address for the vip16:25
SamYapleyou can still address it by your public address just like oyu owuld without hte vip16:25
wadehollerelemoine_: I don't understand why the time is all wrong in that (ansible.log) log (time on host is fine).  I will clean-containers make sure the volumes and log of course is gone, and re-run.16:26
SamYaplewadeholler: time in container is UTC i believe16:26
inc0_SamYaple, I can make new var "mariadb_external_port"16:26
SamYapleUTC time is the recommended time to run openstack on16:26
inc0_and override these16:26
SamYapleinc0_: just but were not going to merge that...16:27
SamYaplewhy do all that extra work?16:27
wadehollerSamYaple: ah, thank you ... again,16:27
inc0_SamYaple, why? oO16:27
elemoine_wadeholler: after cleanup-containers also run docker volume rm heka_socket kolla_logs heka16:27
inc0_what's your issue with that?16:28
SamYapleinc0_: what is the use case? its just extra cruft16:28
inc0_configuring ports is cool16:28
SamYaplewe already do that16:28
elemoine_as cleanup-containers does not remove the Heka-related volumes at this point (https://review.openstack.org/#/c/283066/)16:28
wadehollerjust noticed they were still there.  will do.16:28
inc0_you might want to have different ports at internal for whatever reason16:29
inc0_and externally expose normal endpoints that people are used to16:29
SamYapleill need to see a use case or request first then16:29
inc0_I see no reason why not to have this16:29
SamYaplebecause as it stands i know no reason _to_ have this16:30
inc0_I can't test aio properly without it16:30
inc0_there you go, use case16:30
SamYapleyou can too16:30
SamYaplei just told oyu how16:30
inc0_ehh...whatever. you always fight things.16:30
openstackgerritEric Lemoine proposed openstack/kolla: Make cleanup-containers remove Heka volumes  https://review.openstack.org/28306616:31
SamYaplesame to you buddy16:31
SamYaplebut seriuosly, you can without changing the kolla code base, currently run AIO with haproxy with your setup16:31
SamYapleand do testing16:31
inc0_not always16:31
SamYaplegive me a sceanrio16:31
inc0_ok, well, you want to run haproxy on controller nodes16:31
SamYaplethis is about AIO right?16:32
SamYapleor are you changing the arguement?16:32
inc0_real world example right16:32
SamYapleso changing the arguement, got it.16:32
SamYaplego16:32
inc0_for aio I did show you example and you didn't accept it16:32
SamYaplei told you why you were wrong...16:33
SamYapleand how to set it up16:33
SamYapleso yes, i didnt accept it16:33
SamYaplethe gate is AIO and does haproxy. you can too16:33
inc0_so example - people have running openstack on different ports16:33
inc0_(which people do sometimes)16:33
inc0_and they want to keep same ports for endpoints in koala deploy16:33
inc0_kolla16:33
inc0_but they want to keep things openstackish inside16:34
SamYaplethats not a use case...16:34
inc0_why?16:34
SamYaplebecasue no one does that?16:34
SamYaplethere are about a thousand scenarios where what you describe is a theoretical "use-case"16:34
SamYapleare we going to blow kolla up to cover all of those?16:34
SamYaplei see no real world use to this and no one has asked or requested this16:35
inc0_I've seen deploys of people who do16:35
inc0_so yes, people do use that16:35
SamYaplewhat it will do is change _alot_ of logic and alot of things16:35
SamYapleso if we are going to do it, plan it for newton please16:35
*** dwalsh has quit IRC16:36
inc0_ah..I just noticed it won't be simple new variable16:36
inc0_it this case I agree, I don't want to change logic16:36
SamYapleok. want to add a bp for this so we can discuss for newton?16:37
inc0_no16:38
inc0_too big of a change16:38
inc0_I thought it will be simpler16:38
openstackgerritSteven Dake proposed openstack/kolla: Add docs about failures during evaluation  https://review.openstack.org/28312916:39
inc0_with 2 vips tho16:39
inc0_this will be simpler16:39
inc0_but 2 vips is not merged yet16:39
SamYaplei agree with those statements16:39
SamYapleso discuss for newton?16:39
inc0_with 2 vips I'll just probably submit a patch and review16:40
inc0_nothing to discuss really16:40
inc0_we'll have kolla_external_ip, we might have mariadb_external_port16:40
inc0_or even no, because mariadb won't have external facing port16:40
SamYapleit stil lwont be simple16:40
SamYapleand then there are precheck changes and additions16:40
sdakemitaka is locked up fornew features16:41
sdakeif you want something new, submit blueprint, we will get it in newton or later ;)16:41
sdakewe have about 12 days left until mitaka-316:41
wadehollereleomine_: that helped.  now I made it to an error where the dhcp_agent containers are Created but the deploy failed http://pastebin.com/M72mjNqA16:43
openstackgerritEric Lemoine proposed openstack/kolla: Run yum clean all in Heka Dockerfile  https://review.openstack.org/28303416:44
wadehollerdocker logs neutron_dhcp_agent doesn't show anything - i.e. blank16:44
SamYaplewadeholler: how long ago did you rebuild your contaienrs?16:45
SamYaplethat sounds like its hanging on waiting for a socket16:45
elemoine_wadeholler: sounds like a different issue indeed16:46
wadehollerif by rebuild you mean tools/build.py - then last night.  If you mean deploy then directly before this deploy I did a clean-containers && docker rm ... heka volumes and koala_logs, ...16:46
openstackgerritMerged openstack/kolla: Add a precheck for MountFlags=shared  https://review.openstack.org/28127916:48
sdakeif you eat 3 meals a day for 2 weeks straight, how many meals is that?16:48
sdaketrust me, you will feel better if you do :)16:48
SamYaplei want to build a quote bot that just randomly quotes what sdake says16:50
SamYaplei mean sometimes _even_ with context....16:51
sdakesounds like a waste of time, but whatever works :)16:51
openstackgerritEric Lemoine proposed openstack/kolla: Fix Swift logging  https://review.openstack.org/28311816:53
wadehollerelemoine_: running away for lunch - sdake made me wicked hungry.  will hop back on after lunch.  Thanks for all the help thus far!16:53
*** haplo37 has joined #kolla16:53
elemoine_wadeholler: thank you for reporting these issues16:53
*** tzn has joined #kolla16:54
sdakepbourke I'm not sure what happend hered17:03
sdakehttps://review.openstack.org/#/c/276855/17:03
sdakeone way to fix it is  to resubmit the patch with the same changeid I think17:03
sdakelet me try a recheck first tho17:03
*** jtriley has quit IRC17:04
SamYaplesdake: already done17:04
SamYaplepbourke: no reason to resubmit17:05
SamYaplegates been failing alot, thats on infra though17:05
*** vhosakot has joined #kolla17:05
pbourkesdake: SamYaple: ok thanks17:06
elemoine_this gate thing does not sound easy to manage :)17:07
openstackgerritSam Yaple proposed openstack/kolla: DO NOiT MERGE  https://review.openstack.org/28317017:07
sdakenihilifer can you rebase this please https://review.openstack.org/#/c/275798/17:08
sdakenihilifer can you rebase htis please https://review.openstack.org/#/c/278045/17:09
sdakecore reviewers, please have a look at https://review.openstack.org/#/c/281380/117:18
openstackgerritEric Lemoine proposed openstack/kolla: Make Heka collect Mistral logs  https://review.openstack.org/28317717:19
*** daneyon has joined #kolla17:21
*** xionglingfeng has joined #kolla17:21
*** daneyon_ has joined #kolla17:21
*** jtriley has joined #kolla17:24
*** daneyon has quit IRC17:25
*** tzn has quit IRC17:29
openstackgerritMerged openstack/kolla: Make cleanup-containers remove Heka volumes  https://review.openstack.org/28306617:29
openstackgerritMerged openstack/kolla: Run yum clean all in Heka Dockerfile  https://review.openstack.org/28303417:30
*** jasonsb has quit IRC17:30
*** unicell1 has quit IRC17:44
*** subscope has quit IRC17:47
*** thumpba has joined #kolla17:51
sdakepbourke please review https://review.openstack.org/#/c/283068/217:52
pbourkesdake: change looks good but I no longer use vagrant...17:53
pbourkeso am not the approver you're looking for!17:53
sdakeif you use vagrant17:53
sdakein the past17:53
sdakethat i sgood enough for me17:53
openstackgerritMerged openstack/kolla: Change ubuntu base_image for vagrant to wily64  https://review.openstack.org/28196817:58
xionglingfengPlease suggest how shall I do to improve https://review.openstack.org/#/c/282943/17:59
xionglingfengthank you everyone :-)17:59
openstackgerritMerged openstack/kolla: Harmonise vagrant provider defaults  https://review.openstack.org/28306818:00
openstackgerritMerged openstack/kolla: Make cleanup-containers remove bootstrap_heka  https://review.openstack.org/28306518:00
openstackgerritMerged openstack/kolla: Change Heka sudoers file mode bits  https://review.openstack.org/28308618:00
openstackgerritMerged openstack/kolla-mesos: Fix cleanup with empty volumes  https://review.openstack.org/28269318:02
*** gtt116_ has joined #kolla18:04
*** gtt116 has quit IRC18:07
sdakeelemoine_ before you make a bajillion patches for all of the other services please read my comments here: https://review.openstack.org/#/c/283082/118:07
*** dwalsh has joined #kolla18:12
openstackgerritLingfeng Xiong proposed openstack/kolla: Use external address in novncproxy and spice  https://review.openstack.org/28319918:14
*** unicell has joined #kolla18:17
openstackgerritMerged openstack/kolla: Upgrade task for Swift  https://review.openstack.org/27685518:18
openstackgerritLingfeng Xiong proposed openstack/kolla: Glance image path customization  https://review.openstack.org/28284018:20
*** tzn has joined #kolla18:26
*** dwalsh has quit IRC18:27
*** dwalsh has joined #kolla18:31
*** dtturner has joined #kolla18:34
*** dwalsh has quit IRC18:36
*** salv-orl_ has joined #kolla18:42
*** salv-orlando has quit IRC18:44
*** Marga_ has quit IRC18:55
*** macsz has joined #kolla18:56
*** sdake has quit IRC18:58
*** dwalsh has joined #kolla18:58
elemoine_sdake, I will18:58
*** tzn has quit IRC18:59
*** sdake has joined #kolla19:03
*** macsz has quit IRC19:03
*** subscope has joined #kolla19:04
openstackgerritMerged openstack/kolla: Use external address in novncproxy and spice  https://review.openstack.org/28319919:14
openstackgerritLingfeng Xiong proposed openstack/kolla: Multiple hypervisor driver support for nova  https://review.openstack.org/28294319:23
*** chandankumar has quit IRC19:28
*** neophy has joined #kolla19:30
*** Jeffrey4l_ has joined #kolla19:32
openstackgerritwilliam snow proposed openstack/kolla: Minor update add prechecks into workflow path Add kolla-ansible prechecks prior to deploy to make sure targets are ready for deploy TrivialFix  https://review.openstack.org/28321919:33
*** Jeffrey4l has quit IRC19:34
openstackgerritwilliam snow proposed openstack/kolla: Minor update add prechecks into workflow path TrivialFix  https://review.openstack.org/28321919:34
openstackgerritwilliam snow proposed openstack/kolla: Minor update add prechecks into workflow path  https://review.openstack.org/28321919:36
*** xionglingfeng has quit IRC19:36
openstackgerritMichal Jastrzebski (inc0) proposed openstack/kolla: Upgrade stub for elasticsearch  https://review.openstack.org/28322219:43
inc0_can I get quick merge on this one plz?19:44
SamYaplew00t. i broke top 10 on reviews this cycle according to stackalytics!19:47
SamYaplei know its a pointless stat... but still it makes me happy :)19:47
rhalliseySamYaple, nice!19:49
*** sdake_ has joined #kolla19:49
*** sdake has quit IRC19:50
*** tzn has joined #kolla19:55
sdake_SamYaple is this valid jinja2 syntax https://review.openstack.org/#/c/283118/3/ansible/roles/common/templates/heka.json.j219:56
*** Marga_ has joined #kolla19:56
sdake_and if so, what does -%} do?19:56
sdake_whitespace stripping?19:57
kproskurinsdake_, yes19:59
SamYaplesdake_: it is, but we dont need to use it there I dont think19:59
SamYapleeither way its valid19:59
*** fthiagogv has quit IRC20:00
*** inc0___ has joined #kolla20:15
*** inc0_ has quit IRC20:17
*** Guest40848 has quit IRC20:20
*** akscram has quit IRC20:20
*** mkoderer has quit IRC20:20
*** inc0_ has joined #kolla20:20
*** inc0_ has quit IRC20:21
*** inc0_ has joined #kolla20:21
*** akscram has joined #kolla20:21
*** mkoderer has joined #kolla20:21
*** Guest40848 has joined #kolla20:21
*** athomas has quit IRC20:22
openstackgerritMerged openstack/kolla: Minor update add prechecks into workflow path  https://review.openstack.org/28321920:22
*** inc0___ has quit IRC20:23
*** tzn has quit IRC20:29
openstackgerritSteven Dake proposed openstack/kolla: Add docs about failures during evaluation  https://review.openstack.org/28312920:33
vhosakotsdake_: SamYaple: I got my kolla up! :)  after adding --registry "--registry 192.168.122.41:4000"  to kolla-build20:34
vhosakotI can access horizon now!20:34
*** athomas has joined #kolla20:34
sdake_huh20:34
*** sdake_ is now known as sdake20:35
sdakethat sounds like a bug20:35
vhosakotwell, I have a list of commands to update the quickstart20:35
sdakemore like update build.py ;)20:35
vhosakotsdake: well, I checked with SamYaple and he said it is not a bug20:35
SamYaplevhosakot: awesome!!20:36
sdakeSamYaple --registry "--registry xyz:port" is intended?20:36
vhosakotsdake: do you think kolla-build must force the user to use --registry ?  SamYaple said --registry is not a mandatory flag and can be anything20:36
SamYaplesdake: i think vhosakot mistyped20:36
SamYaple"--registry 192.168.122.41:4000"20:36
sdakeyes i mentioned that several times over the last few days20:37
sdakevhosakot did you mistype?20:37
vhosakotwhat mistype.. I added  "-registry 192.168.122.41:4000"  to kolla-build and it worked!20:37
SamYaplew00t!20:38
sdake[13:34:29]  <vhosakot>sdake_: SamYaple: I got my kolla up! :)  after adding --registry "--registry 192.168.122.41:4000"  to kolla-build20:38
vhosakotI see all the images prefixed with "192.168.122.41:4000/kollaglue/"  in the output of "docker images"20:38
sdakevhosakot yes I mentioned that severalt imes over the last few weeks ;)20:38
SamYapleyou got it20:38
sdakelast few days20:38
SamYapleglad its working20:38
vhosakotsdake: right, I know you mentioned it.. agreed 100% :)   I'm wondering why you said  "that sounds like a bug"  ? :)20:39
openstackgerritMerged openstack/kolla: Upgrade stub for elasticsearch  https://review.openstack.org/28322220:39
wadehollerhey all, can someone validate with me that I took the appropriate steps to avoid https://bugs.launchpad.net/kolla/+bug/154679820:40
openstackLaunchpad bug 1546798 in kolla "neutron failed to deploy in ubuntu binary" [Low,Triaged] - Assigned to Sam Yaple (s8m)20:40
sdakeI thought you meant you typed the command build.py --registry "--registry IP:port"20:40
SamYaplewadeholler: looking20:40
sdakeubuntu binary does not deploy afaik20:40
wadehollerI think that I am hitting that error and really thought I followed the right steps20:40
wadeholleros is centos20:40
wadeholleroutput / pastern - http://pastebin.com/9waBv6pV20:40
SamYaplewadeholler: no read the quickstart guide wadeholler20:40
vhosakotsdake: ah ok cool.. BTW, I will update the quickstart soon20:40
SamYaplefor centos you need to set the flag in docker.service20:40
sdakesandbox error20:40
wadehollerI think that I did, and its in the paste bin and I restarted the service20:41
wadehollerI'm sure its something simple.  sorry to be a pain20:41
SamYaplewadeholler: now remove the failed container and retry20:41
SamYaplewadeholler: if you are positive you restarted the service20:41
SamYapledocker rm -v -f neutron_dhcp_agent20:41
wadehollerok20:41
vhosakotSamYaple: I think  "r = requests.get(source['source'])"  in kolla/kolla/cmd/build.py  needs a timeout... that line hung/blocked for me many times... let me know what you think20:42
vhosakotSamYaple: I think kolla-build should not block/hang..20:42
SamYaplevhosakot: i mean timeouts are fine. i would be curious as to why you time out but if you need one, sure20:42
SamYaplei agree20:42
SamYaplei would think fixing the cause of the hang would be prefered though20:42
SamYaplesdake: got my cloud going. complete with ssl certs20:43
sdakeSamYaple nice ;)20:43
vhosakotSamYaple: intermittent network issues, and tarballs not reachable at http://tarballs.openstack.org/  (due to lot of threads/requests ?)20:43
SamYaplevhosakot: timeouts seem reasonable. please submit a patch :)20:44
sdakevhosakot bug first then patch plz20:44
vhosakotSamYaple: ok, what timeout should I specify, 60 secs, 120 secs, should we read it from globals ?20:44
SamYaplevhosakot: i think 120 is reasonable20:44
vhosakotsdake: bug first, then patch... yes sir20:45
sdakesir is for old people20:45
vhosakotSamYaple: cool, I will hardcode 120 then..20:45
sdakecall me sdake steak steve ;)20:45
SamYaplethe tarballs are like 1mb in size so 2 minutes is fine20:45
sdakeany of those work20:45
SamYaplesdake: im from the south, sir is proper for al20:45
SamYapleall*20:45
vhosakotsdake: hahaha :)20:45
sdakewell us people in the west don't hanker to such formalities :)20:46
SamYapleheathens20:46
*** openstackgerrit has quit IRC20:47
*** openstackgerrit has joined #kolla20:47
*** neophy has quit IRC20:48
vhosakotAfter I change code, I need to run "tox" to run unit tests and test sanity... is there any other comman/script I need to run before pushing my patch set ?20:51
SamYaplevhosakot: nope20:52
vhosakotSamYaple: just "tox" ?20:53
SamYapletechnically you dont _need_ to20:53
SamYaplebut its a good thing to do20:53
SamYapleit can catch a gate failure earlier20:53
vhosakotSamYaple: what ? no tox before pushing ?20:53
SamYapleyou dont have to, its a good idea though20:53
vhosakotSamYaple: wow, I dont want to fail at the gate20:53
kproskurin:-) Just do it, it's a good habbit20:53
SamYapleagreed20:53
vhosakotSamYaple: I will run tox.. cool20:54
SamYaplebut, like, I am not going to do it if i change a docs patch, you know?20:54
vhosakotyeah yeah... only if we touch code20:54
vhosakotcore code especially20:54
SamYaplefor that matter, only python and bash scripts20:54
SamYaplethere is no tox check for ansible stuff20:54
vhosakotSamYaple: how can tox make sense for an ansible playbook ?20:55
SamYapleansible-lint, i just havent implemented it yet20:55
vhosakotSamYaple: ah,20:55
vhosakotSamYaple: cool20:55
*** thumpba has quit IRC20:55
SamYaplesdake: you around?20:56
*** thumpba has joined #kolla20:56
sdakeshoot20:56
openstackgerritLingfeng Xiong proposed openstack/kolla: Allow cinder-volume to be configured to use nfs  https://review.openstack.org/27991320:56
*** thumpba has quit IRC21:00
*** pbourke_ has joined #kolla21:03
*** ayoung has quit IRC21:06
*** unicell has quit IRC21:11
*** unicell has joined #kolla21:11
*** ayoung has joined #kolla21:12
sdakerhallisey can you look at the last comment in https://review.openstack.org/#/c/279913/821:13
rhalliseyya21:25
wadeholler-- user error in that I did not "systemctl daemon-reload" before restarting docker.21:25
wadehollerdeploy finished successfully.  thank everyone.  now the real fun begin!21:26
wadehollerexit21:26
*** wadeholler has left #kolla21:26
sdakeuser error ftw21:27
openstackgerritLingfeng Xiong proposed openstack/kolla: Multiple hypervisor driver support for nova  https://review.openstack.org/28294321:27
sdaked_code's patch should fix that up ;)21:27
openstackgerritSam Yaple proposed openstack/kolla: Multiple hypervisor driver support for nova  https://review.openstack.org/28294321:29
*** gfidente has quit IRC21:30
openstackgerritLiz DiTucci proposed openstack/kolla: Minor updates to Tips and Tricks  https://review.openstack.org/28326721:36
SamYaplesdake: your -1 in not valid here https://review.openstack.org/#/c/28294321:36
SamYaplei have replied to your comments21:36
openstackgerritLingfeng Xiong proposed openstack/kolla: Allow cinder-volume to be configured to use nfs  https://review.openstack.org/27991321:40
*** tzn has joined #kolla21:44
sdakethat xen implementation thing needs to be reworked later21:44
sdakepbourke is right, and as I see it, it could turn into a real mess with multiple hypervisor implementations in  the same container21:45
sdaketime for a break, bbl21:46
SamYaplesdake: no the way the xen thing is now is fine21:46
SamYaplethats not what pbourke was talking about21:46
SamYaplei had him remove all the stuff pbourke was talking about as a condition of merging this xen stuff21:46
SamYaple(also had him remove the /dev:/dev binding as that change worried me right now)21:47
SamYaplethe stuff that _is_ in that patch should be safe for now21:47
SamYapleno changes to non-xen nodes21:47
openstackgerritMerged openstack/kolla: Fix Swift logging  https://review.openstack.org/28311821:49
*** tpot has joined #kolla21:50
openstackgerritSam Yaple proposed openstack/kolla: Unpin elasticsearch  https://review.openstack.org/28327521:52
*** sdake has quit IRC21:57
*** thumpba has joined #kolla22:03
openstackgerritMichal Jastrzebski (inc0) proposed openstack/kolla: HAProxy and Keepalived upgrade  https://review.openstack.org/28327822:03
inc0_Jeffrey4l_, when you read it, could you please reassign bp for upgrade keepalived to me? I did both haproxy and keepalived in same patchiest22:05
inc0_patchset22:06
SamYapleinc0_: that PS looks good22:06
SamYaplei like it :)22:07
inc0_PS?22:07
SamYaplepatchset22:07
inc0_ahh22:07
inc0_ok22:07
inc0_well it's not done yet22:07
inc0_but that's how I want to handle it22:07
SamYaplefighting over a compliment even!22:07
SamYapleim kidding :P22:07
inc0_downtime should be minimal with this approach22:07
*** pbourke_ has quit IRC22:07
openstackgerritMichal Jastrzebski (inc0) proposed openstack/kolla: HAProxy and Keepalived upgrade  https://review.openstack.org/28327822:08
openstackgerritMichal Jastrzebski (inc0) proposed openstack/kolla: HAProxy and Keepalived upgrade  https://review.openstack.org/28327822:10
inc0_ok I'll finish it up tomorrow22:12
inc0_SamYaple, so you are ok with this approach? if so, I'll just clean it up and test the hell out of it22:12
*** sdake has joined #kolla22:13
*** thumpba has quit IRC22:13
SamYaplelooks good to me. one thing you will want it the -V option for keepalived22:13
SamYaplealso it could be optomized if you want22:13
inc0_ok, I'll ping you tomorrow about it22:13
*** thumpba has joined #kolla22:13
SamYaplethe _lowest_ failover time would be to upgrade haproxy on the off nodes22:13
sdakeSamYaple i am thinking beyond xen22:13
SamYaplekick the vip22:13
sdakethat is why it requires rework22:14
SamYapleso doesnt time is the time it takes to reesstablish ssh22:14
SamYaplesdake: even beyond xen... what needs to change?22:14
SamYaplenothing I can think of22:14
inc0_well kicking the vip will cause other stuff to happen as well22:14
sdakeeach hypervisor installs hypervisor specific software22:14
inc0_and redeploy of container is fast enough22:14
sdakewe dont wantto install hypervisor specific software in the libvirt container22:14
SamYaplesdake: not vmware22:14
sdakecould get messy22:14
inc0_so it's like a middle ground between lower impact22:14
SamYapleyea. absolutely we do. its just a few packages that sit idle. like a few MB22:15
inc0_but name is misleading22:15
SamYaplebesides at this point im pretty sure we can do them all suprisingly22:15
SamYapleinc0_: sure no worries. like i said get it working then optomize22:15
inc0_ok22:15
inc0_we can just rename it to nova-hypervisor22:15
inc0_or whatever22:16
sdakexen-hypervisor22:16
inc0_nah, we have nova-libvirt22:16
sdakei already chnaged the ittle22:16
SamYaplebut thats not acurate22:16
*** jtriley has quit IRC22:16
SamYapleits nova-libvirt22:16
SamYaplenova-libvirt doesnt me kvm22:16
SamYapleit means libvirt22:16
inc0_nova doesn't use libvirt for xen management tho right?22:16
SamYapleyes22:16
SamYaplethats my point22:17
inc0_it does? then I'm completely fine with it;)22:17
*** tpot has quit IRC22:17
SamYaplefor, say, vmware and hyper-v the nova-libvirt container isn't even needed22:17
sdakea more optimal approach would be nova-libvirt-xen22:17
sdakenova-libvirt-kvm22:17
sdakeetc22:17
SamYaplei disagree22:17
SamYaplethen kolla has to be aware of hypervisor22:17
SamYaplekolla-ansible*22:17
inc0_also sdake keep in mind that these might change between hosts22:17
inc0_unlikely, but possible22:18
sdakezomg22:18
sdakebrain implode22:18
inc0_I'd rather keep us out of it as much as we can22:18
SamYaplei like what we have22:18
inc0_this way we're hypervisor agnostic and that's cool22:18
SamYaplewe already overload the containers with ceph, even if you dont use ceph22:18
sdakewell we aren't changing it before newton22:18
sdakeso it doesn't matter22:18
SamYaplenow nfs22:18
*** tpot has joined #kolla22:18
inc0_anyway, I'm going home22:18
SamYaplei think a few MB of extra packages is fine22:18
inc0_cya all tomorrow22:18
sdakelater inc0_22:18
SamYaplebye inc0_22:18
sdakeyes, we need to have a coordinated discussion in austin about how to handle these plugins22:19
sdakeatm we have punted for two releases22:19
SamYaplei also disagree with that22:19
SamYaplei havent punted22:19
SamYaplei thought we agreed22:19
SamYaplea few MB is fine...22:19
sdakei dont mind mb22:20
SamYaplethats all this is though22:20
sdakeas long as they dont conflict22:20
SamYaplethey dont22:20
sdakei understand22:20
sdakethat is why I approved the patch22:20
sdakeand approved the blueprint22:20
*** inc0_ has quit IRC22:22
*** thumpba has quit IRC22:23
sdakesamyaple elasticsearch appears to not be in the ubuntu repos22:26
sdakecan you check into it?22:26
sdakehttp://logs.openstack.org/88/269688/14/check/gate-kolla-dsvm-build-ubuntu-source/ca21673/console.html.gz#_2016-02-19_18_30_45_64122:26
sdakethe centos gating work is being sorted out by jeffrey4l22:26
sdakeperhaps you need a special thing in the gate - like you added for iirc mariadb22:27
SamYaplesdake: i do, its the gates fault22:27
SamYaplecost of violating DRY22:27
sdakethe failure is persistent, is it fixable?22:28
*** JoseMello has quit IRC22:28
*** thumpba has joined #kolla22:28
openstackgerritSam Yaple proposed openstack/kolla: Add elasticsearch repo to gate  https://review.openstack.org/28327522:29
sdakei'd like the gate green asap ;)22:29
sdakethanks22:29
SamYapleit shouldnt have been merged initially22:29
SamYaplesomeone wasnt watching the gates22:29
sdakeyup that happens22:29
sdakeatleast we are getting a list of repos we need infra to mirror ;)22:31
*** tzn has quit IRC22:37
*** sdake has quit IRC22:44
openstackgerritRyan Hallisey proposed openstack/kolla: Allow cinder-volume to be configured to use nfs  https://review.openstack.org/27991322:45
openstackgerritRyan Hallisey proposed openstack/kolla: Allow cinder-volume to be configured to use nfs on Centos  https://review.openstack.org/28329022:45
*** haplo37 has quit IRC22:49
openstackgerritLingfeng Xiong proposed openstack/kolla: Glance image path customization  https://review.openstack.org/28284022:51
*** ayoung has quit IRC22:54
openstackgerritVikram Hosakote proposed openstack/kolla: Updated doc/quickstart.rst to help new contributors  https://review.openstack.org/28329822:56
vhosakotI've updated  doc/quickstart.rst  to help new contributors and submitted for review.  Could folks please review it and let me know their comments!22:58
*** dims has joined #kolla23:03
*** salv-orl_ has quit IRC23:08
*** sdake has joined #kolla23:10
*** salv-orlando has joined #kolla23:11
openstackgerritMerged openstack/kolla-mesos: Add openvswitch support  https://review.openstack.org/28238423:12
openstackgerritVikram Hosakote proposed openstack/kolla: Updated doc/quickstart.rst to help new contributors  https://review.openstack.org/28329823:16
*** cui has quit IRC23:20
asalkeldsdake: you about? want to chat about https://blueprints.launchpad.net/kolla/+spec/custom-extend-start23:23
*** spisarski has quit IRC23:24
openstackgerritLingfeng Xiong proposed openstack/kolla: Multiple hypervisor driver support for nova  https://review.openstack.org/28294323:27
*** vhosakot has quit IRC23:32
*** dwalsh has quit IRC23:32
sdakeasalkeld  shoot23:32
asalkeldhangon just adding a comment to that bp23:33
asalkeldsdake: look now23:35
asalkeldbasically we have to be the calling all actions23:35
asalkeldto keep track of dependencies23:35
asalkeld(you have ansible for that)23:35
asalkeldbootstrap is just one kind of command23:35
asalkeldand at the moment it is kind of hard baked into the image23:36
asalkeldwe just want it not hard baked in23:36
*** tzn has joined #kolla23:37
sdakei'm not keen on having it outside of the container because then there could potentially be deviations23:37
*** kzaitsev_mb has joined #kolla23:37
sdakebut controlling when seems ok23:38
*** kzaitsev_ws has joined #kolla23:38
asalkeldsdake: also see these service files more as ansible config tasks23:38
asalkeld(this stuff https://github.com/openstack/kolla-mesos/blob/master/services/mariadb/mariadb.yml.j2)23:38
asalkeldwe need an equivelent23:38
sdakegood lets make one abi to do it23:39
sdakerather then two23:39
asalkeldsdake: huh?23:39
sdakethat is passed into the container to configure it by set_configs.py right?23:39
sdakethe format of that file is the ABI23:39
sdakeansible has one mesos has one23:40
asalkeldsdake: that is nuts23:40
sdakethey should be the same23:40
asalkeldnoooo23:40
asalkeldthat file has marathon stuff in it23:40
asalkeldlike the ansible files have ansible stuff23:40
asalkeldthis is just not a problem23:40
sdakewhich ansible stuff specifically?23:40
asalkeldwell look at the ansible tasks23:41
asalkeldstart/config etc..23:41
sdakethose are not parsed by set_configs.py23:41
asalkeldexactly23:41
asalkeldall we want is bootstrap task not to be automatically run23:41
*** tzn has quit IRC23:41
asalkeld(or a way to turn it off)23:41
asalkeldso we can run it, and keep track of the fact that it has run23:42
*** tzn has joined #kolla23:42
sdakeasalkeld let me tell you what i'm after23:43
sdakeand you can do as you please to make that happen ;)23:43
*** tzn has quit IRC23:43
asalkeldnot if i disagree;)23:43
sdakewhere there is commonality, I don't' want deviation23:43
sdake(this was from midcycle ,not my own opinion)23:43
sdakewhere there must be deviation, deviation is fine23:44
asalkeldsdake: we are as common as we can can be23:44
sdakeSamYaple seemed to think that was not the case23:44
kzaitsev_mbI've got a question. I'm working on murano and starting with L we have optional support for glance-glare repository as storage for the binary assets. This is done with a glare plugin (a stevedore plugin, which is in fact just a python-package), that needs to be installed in the env, where glare api operates. The catch is that plugin currently lives in murano codebase. This poses some challenges for pack23:44
asalkeldthere is much different in the *way* things are deployed23:44
kzaitsev_mbagers. I was wondering if this would be a problem for you guys and are there any best practices on where we should keep such plugins (afair kolla is capable of deploying murano now)23:44
sdakethere was duplciation23:44
kzaitsev_mbsame might be true for, say searchlight plugins.23:45
asalkeldsdake: more in the actual templates23:45
asalkeld(which i think is a good thing)23:45
asalkeldas it allows use to make sure that the config of services is largely the same23:45
sdakeasalkeld dont shoot messenger plz23:46
sdakeperhaps i got message wrong23:46
sdakei tried to capture all the details, but only type 110wpm ;)23:46
asalkeldi'll chat to sam and see what he was referering to23:46
sdakepeople tlak faster then i can type23:46
sdakehe was talking about the extend scripts23:47
*** blahRus has quit IRC23:47
sdakewhich are duplicated between projects23:47
sdakebetween repos i mean23:47
asalkeldsdake: that is because of this issue of the extend scripts23:47
sdakeneedlessly according to nihilifer23:47
asalkeldthe automatic running of ti23:47
sdakewell lets make it configurable and maek that part of the abi23:47
asalkeldkill that and things will get better23:47
asalkeldok23:48
sdakei want a standard abi to the containers23:48
kzaitsev_mbshould I come around and ask that question later? =) anyway pls ping me if anyone can shed some light on the situation23:48
sdakeif that means the containers have to change to implement ansible and mesos, so be it23:48
sdakekzaitsev_mb  can't context dump at this time i'll be right with you23:48
sdakeasalkeld atm the abi is the BOOTSTRAP environment variable23:49
sdakeand the yaml fiel that tells which command to run and which files to copy to where23:49
sdakei'd like that to stay intact between repos23:49
*** subscope has quit IRC23:50
asalkeldsdake: sure, but the extend start is still sourced23:50
asalkeldand we want to be the one to run that23:50
sdakeyup and i'd like extend source to work for both repos not one23:50
asalkeldcool, i'll post something23:50
sdakeso if the sourcing of the start script needs to be controlled23:50
sdakei'm totally open to that23:50
sdakethat seems like a reasonable change23:50
sdakei just want one start file, one abi :)23:51
sdakebtw when I say I, this was feelign of the midcycle23:51
sdakeheka makes it more compelx as well because it introduces a second extension23:51
asalkeldwhen you say "start file", you mean extend_start.sh?23:51
sdakeright23:51
asalkeldok, totally fine with that23:51
sdakeyes, i also want one start  file (the stuff that runs set_configs.py)23:52
sdakeso both of those should be common in the container23:52
sdakefor whatever underlays we implement23:52
sdakeI recognize it gets messy in one file23:52
asalkeldour start.py gets run by the base start file23:52
sdakebut the other optoin, two sets of scripts is worse23:52
sdakeit doubles the amount of maintenance that needs to be done23:52
sdakeand knowledge of one repo doesn't transfer easily to the other23:53
asalkeldand it calls all the commands (one being the bootstrap)23:53
sdakereally want to keep one community here, rather then fragment23:53
asalkeldsure23:53
sdakethis is how we keep one community, one set of files ;)23:53
sdakeone abi23:53
asalkeldtho' ansible is just different to microservices23:53
asalkeldshrug, doing what we can23:54
sdakei think if we work at it, we can make things common23:54
sdakeasalkeld this is long term work, if it cant be done immediatey start thinking about it23:54
sdakei'm not acusing you of not doing what you can :)23:54
sdakei really dont know tbh if we have one abi or not23:54
sdakefrom that last file you inked, it appears not23:55
sdakebecause you have a cmd in there, that is matched by the yaml files used in ansible23:55
sdakeservice->daemon->command23:55
asalkeldsdake: so one reason we use set_configs very little is that we render the templates within the container23:55
sdakethere is no good reason to duplicate that23:55
asalkeld*confused*23:55
sdakemem, cpus, i can understand23:56
asalkeldwhat is a duplication?23:56
sdakehttps://github.com/openstack/kolla-mesos/blob/master/services/mariadb/mariadb.yml.j2#L1123:56
sdakemoment23:56
asalkeldi am not going to learn how to parse anisble files23:56
sdakehttps://github.com/openstack/kolla/blob/master/ansible/roles/mariadb/templates/mariadb.json.j2#L323:57
sdakethis is not an ansible file23:57
sdakethis file is parsed by set_configs.py23:57
sdakeansible copies this into the container23:57
sdake(after jinja2ifying it)23:57
asalkeldexcept set_config does not track dependencies23:57
asalkeldso one option is we can somehow replace set_config.py23:58
sdakethe second link could be expanded to merge the abi in the first mariadb.yml23:58
sdakeyou see how there are two mariadb.yml "config" loaders23:58
sdakethat is the abi23:58
sdakei prefer one23:58
asalkeldthere is a lot more going on23:59
sdakeif you tell me it can't be done, i'll take your word for it23:59
sdakebut I believe that was the objection raised at the midcycle23:59
asalkeldnot sure, "can't be done", but not sure if it is worth the effort23:59

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