| *** markvoelker has joined #openstack-ansible | 00:02 | |
| *** BjoernT has quit IRC | 00:04 | |
| openstackgerrit | Kevin Carter proposed stackforge/os-ansible-deployment: Fix the package pinning problem within the galera roles https://review.openstack.org/176134 | 00:06 |
|---|---|---|
| *** markvoelker has quit IRC | 00:08 | |
| openstackgerrit | Kevin Carter proposed stackforge/os-ansible-deployment: Fix the package pinning problem within the galera roles https://review.openstack.org/176137 | 00:10 |
| openstackgerrit | Kevin Carter proposed stackforge/os-ansible-deployment: Fix the package pinning problem within the galera roles https://review.openstack.org/176134 | 00:11 |
| openstackgerrit | Kevin Carter proposed stackforge/os-ansible-deployment: Fix the package pinning problem within the galera roles https://review.openstack.org/176137 | 00:12 |
| cloudnull | javeriak still around ? | 00:12 |
| javeriak | yep | 00:12 |
| cloudnull | i've seen that issue when the lb is not able to contact the galera cluster . | 00:12 |
| cloudnull | you should be able to ssh to the container that is exhibiting the issue and then simply run `mysql`. | 00:13 |
| cloudnull | if that fails I'd check the lb vip | 00:13 |
| javeriak | ssh from the deploy node? | 00:13 |
| javeriak | cloudnull: okay, so i was getting this on the keystone container and running 'mysql' it says " Can't connect to MySQL server on '172.29.236.7" | 00:15 |
| javeriak | and the only place i see that defined is in "etc/rpc_deploy/rpc_user_config.yml" which has the correct value | 00:17 |
| cloudnull | if you run `mysql -h 172.29.236.7` can you connect to it ? | 00:17 |
| cloudnull | also running haproxy? | 00:17 |
| javeriak | cloudnull: yes "mysql -h 172.29.236.7" gives the same, and haproxy is running | 00:18 |
| cloudnull | did you run the haproxy play to set that up and if so, can you run `hatop -s /var/run/haproxy.stat` and see that its serving content to the backend vip? | 00:19 |
| javeriak | okay i managed to get the command to run, turns out the container wasnt able to connect to the haproxy node, removed some extra routes | 00:20 |
| javeriak | thanks cloudnull | 00:20 |
| cloudnull | ah, that'll do it. | 00:20 |
| cloudnull | :) | 00:20 |
| cloudnull | glad it wasnt something else terrible going on ;) | 00:21 |
| javeriak | yea phew :) .. btw that stat command, so the ones I see down, i should go check correct? | 00:21 |
| cloudnull | yes. | 00:22 |
| openstackgerrit | Kevin Carter proposed stackforge/os-ansible-deployment: Fix the package pinning problem within the galera roles https://review.openstack.org/176137 | 00:29 |
| openstackgerrit | Kevin Carter proposed stackforge/os-ansible-deployment: Fix the package pinning problem within the galera roles https://review.openstack.org/176134 | 00:32 |
| *** jaypipes has quit IRC | 00:37 | |
| *** stevemar has joined #openstack-ansible | 01:56 | |
| *** markvoelker has joined #openstack-ansible | 02:01 | |
| *** darrenc is now known as darrenc_afk | 02:02 | |
| openstackgerrit | Kevin Carter proposed stackforge/os-ansible-deployment: Fix the package pinning problem within the galera roles https://review.openstack.org/176134 | 02:08 |
| openstackgerrit | Kevin Carter proposed stackforge/os-ansible-deployment: Fix the package pinning problem within the galera roles https://review.openstack.org/176137 | 02:10 |
| cloudnull | palendae ^ updated based on your comments. | 02:10 |
| palendae | cloudnull: Cool | 02:10 |
| palendae | Conversation in #openstack-infra sounds like rechecks are broken | 02:11 |
| *** javeriak has quit IRC | 02:12 | |
| *** darrenc_afk is now known as darrenc | 02:16 | |
| miguelgrinberg | cloudnull palendae need an opinion. I would like to create a "gate" test suite to replace the "scenario" that we are currently using at the gate. This new group will contain the same scenario tests we are running, plus one basic heat smoke test. Yay or nay? | 02:20 |
| palendae | miguelgrinberg: So you'd propose making that the default job in the script? | 02:22 |
| miguelgrinberg | right, that's the proposal | 02:22 |
| miguelgrinberg | there are no heat scenario tests in tempest | 02:23 |
| palendae | I'm not opposed. We don't have any heat coverage at all right now | 02:23 |
| miguelgrinberg | right, we missed the auth problem last week | 02:23 |
| *** galstrom_zzz is now known as galstrom | 02:47 | |
| *** sacharya has joined #openstack-ansible | 02:49 | |
| cloudnull | miguelgrinberg i say yay. | 02:59 |
| miguelgrinberg | cloudnull palendae: okay, patch is coming soon | 03:02 |
| Sam-I-Am | stop working :) | 03:02 |
| miguelgrinberg | Sam-I-Am: says who? | 03:03 |
| Sam-I-Am | says me! heh | 03:04 |
| Sam-I-Am | because i am too | 03:04 |
| miguelgrinberg | much worse for you at 10pm than me at 8pm :) | 03:05 |
| Sam-I-Am | this is true | 03:05 |
| cloudnull | Miguelgrinberg : with the proposed change , is that something that we could gate on now but could go upstream into tempest as an actual scenario test? | 03:10 |
| miguelgrinberg | cloudnull: wasn't clear. I did not add a scenario test to tempest. I'm enabling an api test in addition to the scenario tests that we currently have. | 03:10 |
| miguelgrinberg | that's why I created a new test list called "gate", to not mix non-scenario tests in the "scenario" list | 03:11 |
| cloudnull | Ah OK. | 03:11 |
| miguelgrinberg | I haven't investigated why, but all the scenario tests for heat were moved into the heat repo | 03:11 |
| cloudnull | Sorry. Carry on. Nothing to see here. ;) | 03:11 |
| cloudnull | Miguelgrinberg that's a thing happening to all the projects. | 03:12 |
| miguelgrinberg | what's the reason? | 03:12 |
| cloudnull | So the tempest project doesn't have to maintain them. | 03:12 |
| cloudnull | I don't know all the reasons behind it. But its something that's come up in the defcore meetings. | 03:13 |
| miguelgrinberg | so in the future we'll have to import tests from each project | 03:13 |
| cloudnull | I believe so. | 03:13 |
| *** javeriak has joined #openstack-ansible | 03:13 | |
| openstackgerrit | Miguel Grinberg proposed stackforge/os-ansible-deployment: Add a basic orchestration test to the gate check https://review.openstack.org/176166 | 03:14 |
| miguelgrinberg | cloudnull: ^ let me know what you think | 03:14 |
| cloudnull | Looks good to me. | 03:15 |
| *** galstrom is now known as galstrom_zzz | 03:15 | |
| cloudnull | Question, could we combine the scenario + api tests as they are? | 03:16 |
| cloudnull | The gate job will take longer but that's OK for more test coverage. | 03:17 |
| *** galstrom_zzz is now known as galstrom | 03:17 | |
| *** JRobinson__ is now known as JRobinson__afk | 03:30 | |
| *** galstrom is now known as galstrom_zzz | 03:43 | |
| *** galstrom_zzz is now known as galstrom | 03:44 | |
| *** sacharya1 has joined #openstack-ansible | 04:01 | |
| *** fawadkhaliq has joined #openstack-ansible | 04:02 | |
| *** sacharya has quit IRC | 04:02 | |
| *** sacharya has joined #openstack-ansible | 04:04 | |
| *** sacharya1 has quit IRC | 04:05 | |
| *** mahito has joined #openstack-ansible | 04:15 | |
| openstackgerrit | Kevin Carter proposed stackforge/os-ansible-deployment: Fix the package pinning problem within the galera roles https://review.openstack.org/176137 | 04:16 |
| openstackgerrit | Kevin Carter proposed stackforge/os-ansible-deployment: Fix the package pinning problem within the galera roles https://review.openstack.org/176134 | 04:17 |
| *** galstrom is now known as galstrom_zzz | 04:20 | |
| *** sacharya has quit IRC | 04:24 | |
| *** sdake has joined #openstack-ansible | 04:29 | |
| *** sdake_ has joined #openstack-ansible | 04:32 | |
| *** sdake has quit IRC | 04:33 | |
| *** JRobinson__afk is now known as JRobinson__ | 04:38 | |
| *** fawadkhaliq has quit IRC | 04:43 | |
| openstackgerrit | Miguel Grinberg proposed stackforge/os-ansible-deployment: Add a basic orchestration test to the gate check https://review.openstack.org/176166 | 05:05 |
| *** galstrom_zzz is now known as galstrom | 05:09 | |
| miguelgrinberg | cloudnull: if this patch I proposed passed we have a "gate" test list that we can fill out with other api tests. As far as Heat goes not all tests pass, so we can't select all of them at this time, that's something I will probably look into tomorrow. We also need a more convenient way to select tests, the egrep based solution gets tiring for complex selections. | 05:13 |
| *** galstrom is now known as galstrom_zzz | 05:15 | |
| *** fawadkhaliq has joined #openstack-ansible | 05:16 | |
| *** javeriak has quit IRC | 05:39 | |
| *** sdake_ has quit IRC | 05:42 | |
| *** markvoelker has quit IRC | 06:10 | |
| *** stevemar has quit IRC | 06:31 | |
| *** sdake has joined #openstack-ansible | 06:34 | |
| *** mahito has quit IRC | 06:38 | |
| *** markvoelker has joined #openstack-ansible | 06:40 | |
| *** markvoelker has quit IRC | 06:45 | |
| mancdaz | sup | 06:56 |
| mattt | morning mancdaz | 07:02 |
| *** mrodden has quit IRC | 07:06 | |
| *** mrodden has joined #openstack-ansible | 07:07 | |
| *** javeriak has joined #openstack-ansible | 07:13 | |
| *** mrodden has quit IRC | 07:20 | |
| *** mrodden has joined #openstack-ansible | 07:21 | |
| *** JRobinson__ has quit IRC | 07:25 | |
| *** sdake has quit IRC | 07:35 | |
| *** markvoelker has joined #openstack-ansible | 07:41 | |
| *** markvoelker has quit IRC | 07:46 | |
| *** sdake has joined #openstack-ansible | 07:47 | |
| *** sdake has quit IRC | 08:00 | |
| odyssey4me | werd | 08:01 |
| mattt | morning odyssey4me | 08:03 |
| odyssey4me | I see that galera is once again a pain in the behind. | 08:04 |
| *** sandywalsh has quit IRC | 08:24 | |
| *** sandywalsh has joined #openstack-ansible | 08:24 | |
| *** ccrouch has quit IRC | 08:31 | |
| *** fawadkhaliq has quit IRC | 08:36 | |
| *** markvoelker has joined #openstack-ansible | 08:42 | |
| *** markvoelker has quit IRC | 08:47 | |
| javeriak | hey guys, i wrote a new play that is supposed to run on the 'neutron_all' group, i.e. execute only on all neutron-containers, but for some reason it also runs on the compute hosts. | 08:52 |
| mattt | javeriak: the neutron-linuxbridge-agent runs on the compute hosts | 08:53 |
| mattt | javeriak: so neutron_all will contain neutron_linuxbridge_agent which will contain your compute hosts | 08:55 |
| javeriak | matt: ah okay, is there an easy way to determine what hosts are included in which groups, rpc_enviroment.yml seems to give a top-level distribution | 08:55 |
| mattt | javeriak: you can see this in your rpc_inventory.json file | 08:56 |
| javeriak | matt: oh okay, i see that only the infra_neurtron_server containers are listed under neutron_server group but neutron conf files are also present on the agents container, will I have to specify both groups for mutual plays | 09:02 |
| *** fawadkhaliq has joined #openstack-ansible | 09:07 | |
| openstackgerrit | Jesse Pretorius proposed stackforge/os-ansible-deployment: Fix the package pinning problem within the galera roles https://review.openstack.org/176134 | 09:22 |
| mattt | javeriak: yeah, the neutron_server group is for neutron-server service itself, but you'll find any of the neutron services will have neutron-configurations on them | 09:23 |
| mattt | javeriak: there was a recent change in master (it's probably in kilo branch) to narrow down neutron configurations based on what service you're running -- for example, there is no need for the neutron.conf to have db configurations if you're just using the linuxbridge-agent for example | 09:24 |
| javeriak | matt: yes that makes more sense, as currently its a bit harder to track down all places configurations may need to be propagated for one service that has many containers | 09:25 |
| mattt | javeriak: in that case you'd be best off using neutron_all | 09:28 |
| openstackgerrit | Jesse Pretorius proposed stackforge/os-ansible-deployment: Fix the package pinning problem within the galera roles https://review.openstack.org/176134 | 09:28 |
| javeriak | matt: but then things like neuton.conf settings error out for compute nodes that are also grouped in neutron_all, | 09:31 |
| mattt | javeriak: why would it error out ? | 09:34 |
| javeriak | matt: oh wait, if i were to introduce a new compute into an already running system, i would need to add it to the inventory json you mentioned, so that it gets the appropriate configs, hence thats why it cant find neurton.conf | 09:36 |
| javeriak | matt: the actual neutron server process should only be running on the neutron-server containers? i dont seem to have a problem when i try to restart it for neutron-all | 09:39 |
| mattt | javeriak: if you are adding a new compute node you'd just have to update rpc_user_config.yml, you wouldn't have to modify the inventory json | 09:39 |
| javeriak | matt: the inventory would have to be regenerated though right? | 09:40 |
| mattt | javeriak: ansible should take care of that as far as i'm aware, i'll need to test first hand but i don't believe anything further is needed | 09:41 |
| *** markvoelker has joined #openstack-ansible | 09:43 | |
| javeriak | matt: i was working off of juno and i added another compute to the rpc_user_config.yml and ran all playbooks, i did not come into any direct errors, however the new compute seems to be missing some configs as when i run my custom play on it, it cannot find some configs | 09:43 |
| javeriak | matt: would this have to do with clearing the previous inventory cache? it does not re-generate if it finds the files already there? | 09:44 |
| mattt | javeriak: ah i see what you are saying now, sorry ... yeah there you'd need to run the neutron and nova plays | 09:44 |
| *** markvoelker has quit IRC | 09:47 | |
| mattt | javeriak: you can probably limit the plays that you run to like nova-compute.yml, nova-compute-keys.yml, and neutron-linuxbridge-agent.yml, but you'd have to test this to double-check | 09:50 |
| javeriak | matt: yes got it, thanks alot | 09:50 |
| openstackgerrit | Jesse Pretorius proposed stackforge/os-ansible-deployment: Fix the package pinning problem within the galera roles https://review.openstack.org/176134 | 09:51 |
| openstackgerrit | Jesse Pretorius proposed stackforge/os-ansible-deployment: Fix the package pinning problem within the galera roles https://review.openstack.org/176134 | 09:57 |
| openstackgerrit | Jesse Pretorius proposed stackforge/os-ansible-deployment: Fix the package pinning problem within the galera roles https://review.openstack.org/176134 | 09:57 |
| openstackgerrit | Matt Thompson proposed stackforge/os-ansible-deployment: Fix the package pinning problem within the galera roles https://review.openstack.org/176137 | 10:19 |
| openstackgerrit | Jesse Pretorius proposed stackforge/os-ansible-deployment: Set gerrit default branch to kilo https://review.openstack.org/176246 | 10:25 |
| mattt | javeriak: cool, any further questions just ping, seems like we're all learning how this stuff works :) | 10:30 |
| *** daneyon_ has joined #openstack-ansible | 10:34 | |
| *** daneyon has quit IRC | 10:35 | |
| *** markvoelker has joined #openstack-ansible | 10:43 | |
| *** markvoelker has quit IRC | 10:48 | |
| *** markvoelker has joined #openstack-ansible | 11:44 | |
| *** markvoelker has quit IRC | 11:45 | |
| *** markvoelker has joined #openstack-ansible | 11:45 | |
| openstackgerrit | Matt Thompson proposed stackforge/os-ansible-deployment: Ensure horizon's static content has correct ownership https://review.openstack.org/176263 | 11:53 |
| openstackgerrit | Merged stackforge/os-ansible-deployment: Reorder common apt repo tasks https://review.openstack.org/175554 | 12:05 |
| *** fawadkhaliq has quit IRC | 12:13 | |
| *** jaypipes has joined #openstack-ansible | 12:15 | |
| *** KLevenstein has joined #openstack-ansible | 12:22 | |
| *** mnestheu1 has joined #openstack-ansible | 12:31 | |
| *** galstrom_zzz is now known as galstrom | 12:50 | |
| *** javeriak has quit IRC | 12:53 | |
| *** galstrom is now known as galstrom_zzz | 12:54 | |
| b3rnard0 | i did include the bug that kevin mentioned but i know we prob need multiple eyes on it | 13:00 |
| *** galstrom_zzz is now known as galstrom | 13:12 | |
| *** yaya has joined #openstack-ansible | 13:17 | |
| *** galstrom is now known as galstrom_zzz | 13:19 | |
| *** KLevenstein has quit IRC | 13:25 | |
| *** stevemar has joined #openstack-ansible | 13:29 | |
| *** stevemar has quit IRC | 13:37 | |
| *** Mudpuppy has joined #openstack-ansible | 13:50 | |
| *** Mudpuppy has quit IRC | 13:52 | |
| *** Mudpuppy has joined #openstack-ansible | 13:52 | |
| *** Mudpuppy has quit IRC | 13:57 | |
| *** sigmavirus24_awa is now known as sigmavirus24 | 14:05 | |
| *** Mudpuppy has joined #openstack-ansible | 14:05 | |
| *** prometheanfire has quit IRC | 14:06 | |
| *** prometheanfire has joined #openstack-ansible | 14:07 | |
| *** stevemar has joined #openstack-ansible | 14:07 | |
| *** KLevenstein has joined #openstack-ansible | 14:08 | |
| *** Mudpuppy has quit IRC | 14:10 | |
| *** Mudpuppy has joined #openstack-ansible | 14:11 | |
| *** erikmwilson is now known as Guest95658 | 14:11 | |
| *** Guest95658 has quit IRC | 14:11 | |
| *** erikmwilson has joined #openstack-ansible | 14:11 | |
| *** erikmwilson_ has joined #openstack-ansible | 14:11 | |
| openstackgerrit | Kevin Carter proposed stackforge/os-ansible-deployment: Set gerrit default branch to kilo https://review.openstack.org/176246 | 14:16 |
| cloudnull | odyssey4me ^ updated dependency to the pin update. | 14:18 |
| odyssey4me | cloudnull yeah, saw that - thanks | 14:18 |
| odyssey4me | pretty much any patches going in now will need that other patch as a dependence | 14:18 |
| odyssey4me | *cy | 14:18 |
| cloudnull | yup. we need to tackle icehouse juno too :'( | 14:19 |
| odyssey4me | sorry about rewriting the galera patches without consulting you, by the way - I figured it was more important to get it done asap | 14:19 |
| odyssey4me | yeah, busy looking into juno, but can't replicate the issue so far | 14:19 |
| odyssey4me | it's there in openstack-infra, but not there in any cloud server AIO or multi-node builds | 14:20 |
| cloudnull | no rewriting them was the right thing to do . | 14:20 |
| cloudnull | so no worries brother . | 14:20 |
| cloudnull | as for the issue you need to make sure your on the latest Ubuntu LTS image release. | 14:20 |
| cloudnull | in the iad3 lab i didn't see the problem which was running 14.04.01 | 14:21 |
| odyssey4me | cloudnull before doing a build I made sure to 'apt-get dist-upgrade', does that not achieve the same thing? | 14:21 |
| cloudnull | but in the qe lab we ran into is running 14.04.02 | 14:21 |
| cloudnull | dist-upgrade should do it. | 14:21 |
| odyssey4me | yep: DISTRIB_DESCRIPTION="Ubuntu 14.04.2 LTS" | 14:22 |
| cloudnull | maybe a "full-upgrade" is what needs to be done? | 14:22 |
| mattt | i'm just trying with an image that is already 14.04.2 LTS | 14:22 |
| palendae | Checking nova image list against RAX, none of the 14.04 images have the 3rd number =\ | 14:23 |
| mattt | the previous i built was 14.04.1 i believe and something upgraded it along the way | 14:23 |
| odyssey4me | I suspect that the issue relates to a dependant library somewhere along the way, so when mariadb tries to install the old version we're pinning to apt simply refuses and tries something else. | 14:25 |
| *** stevemar has quit IRC | 14:26 | |
| mattt | cloudnull: did you try juno or just kilo/master ? | 14:26 |
| odyssey4me | We may just be able to update the pinned version to a newer release, but to do that we need to update the right mirror and wait for sync. I just want to make sure that this is the problem before we waste time on that rabbit hole. | 14:27 |
| sigmavirus24 | cloudnull: are we going to remove the rpc_deployment directory for Liberty? | 14:28 |
| cloudnull | odyssey4me it is a lib issue for maria this is the output when you run into it. | 14:28 |
| cloudnull | The following packages have unmet dependencies: | 14:28 |
| cloudnull | libmariadbclient-dev : Depends: libmariadbclient18 (>= 5.5.42+maria-1~trusty) but it is not going to be installed | 14:28 |
| cloudnull | mariadb-client : Depends: mariadb-client-5.5 (= 5.5.42+maria-1~trusty) but it is not going to be installed | 14:28 |
| cloudnull | sigmavirus24 that was the plan | 14:28 |
| sigmavirus24 | :thumbsup: | 14:29 |
| cloudnull | so in truth we could remove it in master whenever | 14:29 |
| * sigmavirus24 was just checking his santiy | 14:29 | |
| sigmavirus24 | *sanity | 14:29 |
| odyssey4me | cloudnull in your testing environment, try updating the mariadb repo to this one: http://archive.mariadb.org/mariadb-5.5.42/repo/ubuntu/ | 14:29 |
| * cloudnull doing | 14:30 | |
| odyssey4me | mattt there was another archive mirror in europe that was faster - what was the URL again? | 14:30 |
| odyssey4me | archive.mariadb.org had severe rate limiting on it | 14:30 |
| openstackgerrit | Ian Cordasco proposed stackforge/os-ansible-deployment: Remove the rpc_deployment directory https://review.openstack.org/176332 | 14:31 |
| mattt | openstackgerrit: looking | 14:31 |
| mattt | odyssey4me: http://ftp.hosteurope.de/mirror/archive.mariadb.org | 14:31 |
| sigmavirus24 | mattt: how nice of you to review my patchset so quickly ;) | 14:31 |
| cloudnull | odyssey4me same error libmariadbclient18 : Depends: libmysqlclient18 (= 5.5.42+maria-1~trusty) but 5.5.43-0ubuntu0.14.04.1 is to be installed | 14:31 |
| odyssey4me | cloudnull did you clear the packages installed first, do an upt-get update, then run the play? | 14:34 |
| cloudnull | i cleared the package install, and then run update. | 14:34 |
| cloudnull | i was installing by hand | 14:34 |
| odyssey4me | bummer, it's only one patch rev from the version we're pinned to anyway :/ | 14:35 |
| cloudnull | but if i drop the pin to the domain this go | 14:39 |
| mattt | cloudnull: so the pin gets dropped on juno already | 14:41 |
| mattt | but we're hitting this same issue | 14:41 |
| *** Bjoern___ has joined #openstack-ansible | 14:41 | |
| odyssey4me | cloudnull to the archive.mariadb domain? | 14:41 |
| cloudnull | mattt the pin that we have in juno / icehouse we pin with an option. | 14:41 |
| cloudnull | which is o=MariaDB | 14:41 |
| cloudnull | and now ubuntu's packages satisfy that option. | 14:42 |
| cloudnull | so if we do `Pin: origin <domain>` | 14:42 |
| odyssey4me | cloudnull that only works if the mariadb and ubuntu mirrors used are different host names | 14:43 |
| *** sdake has joined #openstack-ansible | 14:43 | |
| odyssey4me | I tested that this morning, which is why I switch to the org pinning for master/kilo. | 14:43 |
| cloudnull | which they are for us | 14:43 |
| odyssey4me | cloudnull they would be for us and for others who're using their own local mirrors... so we need another way to differentiate | 14:44 |
| *** sdake_ has joined #openstack-ansible | 14:44 | |
| odyssey4me | do the ubuntu packages honestly match that now? | 14:44 |
| cloudnull | it seems that way . | 14:45 |
| mattt | odyssey4me: doesn't explain why kilo/master are now passing | 14:45 |
| odyssey4me | how do we see this metadata? | 14:46 |
| mattt | odyssey4me: it's the Release file | 14:46 |
| mattt | i'm wondering if mirror.rs has some sync issue | 14:47 |
| cloudnull | odyssey4me: apt-cache show | 14:47 |
| *** sdake has quit IRC | 14:48 | |
| cloudnull | mattt just for the sake of argument the boxes im using aren't using the rs mirrors they're pointed at "http://gb.archive.ubuntu.com/ubuntu" | 14:48 |
| mattt | origin in http://gb.archive.ubuntu.com/ubuntu/dists/trusty-updates/universe/binary-amd64/Release is not MariaDB | 14:52 |
| mattt | that'd break a lot of other stuff | 14:52 |
| odyssey4me | cloudnull have you found a way to tell which match the 'release o=MariaDB' | 14:52 |
| mattt | odyssey4me: ^^^ | 14:53 |
| odyssey4me | versus: http://mirror.rackspace.com/mariadb/repo/5.5/ubuntu/dists/trusty/Release | 14:53 |
| odyssey4me | and http://archive.mariadb.org/mariadb-5.5.42/repo/ubuntu/dists/trusty/Release | 14:54 |
| odyssey4me | what I am seeing, though, is that apt-cache show mariadb-common on a server doesn't show Origin: MariaDB for the MariaDB packages | 14:55 |
| cloudnull | so as a compromise to the issue, which will in fact resolve it, we can pin to both the domain and the release. | 14:57 |
| odyssey4me | cloudnull it does fix it? seems odd as the domain is the same for both except for the situation you're in right now | 14:58 |
| *** galstrom_zzz is now known as galstrom | 14:58 | |
| odyssey4me | I'm game for another patching attribute, but ideally something unique | 14:59 |
| odyssey4me | origin & label perhaps? | 14:59 |
| cloudnull | that may work. | 15:00 |
| odyssey4me | alternatively the actual file version? | 15:00 |
| cloudnull | the situation my boxes are in is likely going to be the most common as deployers adopt the system. | 15:00 |
| odyssey4me | yeah, but it's not a guaranteed win | 15:01 |
| cloudnull | its not. | 15:01 |
| odyssey4me | the file version may also work, but I thought we also has a pin for this | 15:01 |
| cloudnull | the other option here would be to update mariadb to the latest stable. | 15:01 |
| cloudnull | instead of staying with this old release. | 15:02 |
| odyssey4me | cloudnull the latest for mariadb is only a minor patch release up - that's the archive repo I sent you earlier | 15:02 |
| odyssey4me | unless you're talking v10? | 15:02 |
| cloudnull | no i mean mariadb10 | 15:02 |
| odyssey4me | eek, that's a bit rough for icehouse/juno | 15:02 |
| cloudnull | agreeded | 15:03 |
| cloudnull | but master/kilo could adopt that a more updated system. | 15:03 |
| odyssey4me | agreed - that's a possibility for kilo/master, although we'd need to get upgrades tested | 15:03 |
| cloudnull | as for juno / icehouse pinning to the maria repos as provided by the maria people would make this problem go away as well. | 15:04 |
| odyssey4me | the mariadb version numbers are all 5.5.41+maria-1~trusty | 15:05 |
| odyssey4me | different to ubuntu's packages | 15:05 |
| odyssey4me | a version pin may be the simplest way to get this right | 15:06 |
| odyssey4me | I have a pretty simple suggestion for icehouse/juno - we add a pin for the conflicting files to ensure that they use the mariadb versions | 15:07 |
| odyssey4me | we also pin at a level that replaces any previously installed package by the same name | 15:08 |
| odyssey4me | the conflicts are mysql-common and libmysqlclient18 | 15:08 |
| sigmavirus24 | andymccr: do you agree with me on https://bugs.launchpad.net/openstack-ansible/+bug/1439892 ? | 15:08 |
| openstack | Launchpad bug 1439892 in openstack-ansible trunk "Bindmount for Glance" [High,Confirmed] - Assigned to Ian Cordasco (icordasc) | 15:08 |
| *** galstrom is now known as galstrom_zzz | 15:08 | |
| mattt | sigmavirus24: i wouldn't make that change | 15:09 |
| sigmavirus24 | mattt: the bindmount or the change to the frequency with which we manage the cache? | 15:10 |
| mattt | sigmavirus24: not the change you suggested but using bind mount for images :) | 15:10 |
| odyssey4me | the version to pin against is 5.5.41+maria-1~trusty | 15:11 |
| sigmavirus24 | mattt: yeah, I think andymccr, you, and I are all in agreement to *not* bindmount the container just because the cache fills up quickly | 15:11 |
| odyssey4me | packages: mysql-common, libmariadbclient18, libmariadbclient-dev | 15:11 |
| sigmavirus24 | mattt: as a compromise I'm thinking of having the cache management scripts run every 12 hours (11 hrs 59 minutes) instead of every 24 | 15:12 |
| andymccr | yeh i think thats a bad solution sigmavirus24. | 15:12 |
| sigmavirus24 | I would hope that would be frequent enough to not cause too many cache misses while preventing the cache from becoming too large | 15:12 |
| sigmavirus24 | andymccr: changing the cron job is a bad solution? | 15:12 |
| *** yaya has quit IRC | 15:12 | |
| cloudnull | odyssey4me this seems to work http://cdn.pasteraw.com/tmyov5r4qjul9eoid6vbgy2g4ls2t5h | 15:13 |
| odyssey4me | cloudnull triple whammy! | 15:14 |
| odyssey4me | to replace a file already deployed I think the pin priority needs to be above 1000+ | 15:14 |
| andymccr | oh no i mean the bind mount sigmavirus24, sorry half dug in on this mysql package and my context switching is clearly terrible | 15:14 |
| mattt | odyssey4me: yep | 15:14 |
| cloudnull | better one http://cdn.pasteraw.com/chr6z64pkgcfuy1qfmgztglgepuzm0t | 15:14 |
| sigmavirus24 | andymccr: no worries | 15:14 |
| andymccr | i like your solution much better sigmavirus24. itmakes more sense to me | 15:14 |
| sigmavirus24 | andymccr: cool. sending patches now | 15:15 |
| cloudnull | so the version is the highest priority, then the released, then the domain | 15:15 |
| *** fawadkhaliq has joined #openstack-ansible | 15:16 | |
| cloudnull | we'd need to set the version as a var in the default if we did something like that. | 15:17 |
| cloudnull | it could become part of the hash. | 15:17 |
| cloudnull | and rendered in the template if defined. | 15:17 |
| odyssey4me | cloudnull yeah, that's how it's done in icehouse/juno | 15:18 |
| cloudnull | isnt icehouse / juno doing https://github.com/stackforge/os-ansible-deployment/blob/juno/rpc_deployment/roles/common/templates/mariadb-priority | 15:19 |
| cloudnull | which has no version data within it | 15:19 |
| odyssey4me | cloudnull we can use the apt-pinning which is being done for lxc, etc - it's a different task | 15:21 |
| cloudnull | using this https://github.com/stackforge/os-ansible-deployment/blob/juno/rpc_deployment/roles/common/templates/apt_pinned_packages.j2 | 15:22 |
| odyssey4me | but we can just keep it simple in one file | 15:22 |
| odyssey4me | ok, I'll put together revised master/kilo patches in a monent - just doing icehouse/juno patches first | 15:24 |
| *** jwagner_away is now known as jwagner | 15:25 | |
| odyssey4me | cloudnull fyi - it appears that for a pin-priority of 1000 and below, apt will allow itself to select an alternative origin... whereas for 1001+ it'll force the pin origin for the install | 15:26 |
| odyssey4me | http://blog.opperschaap.net/2009/12/12/using-the-preferences-file-on-debian-and-debian-based-distributions/ | 15:26 |
| cloudnull | i used https://wiki.debian.org/AptPreferences for the source of truth | 15:26 |
| cloudnull | but that sounds about right | 15:27 |
| *** rromans has quit IRC | 15:31 | |
| openstackgerrit | Miguel Alejandro Cantu proposed stackforge/os-ansible-deployment: Implement Ceilometer[WIP] https://review.openstack.org/173067 | 15:31 |
| *** rromans has joined #openstack-ansible | 15:33 | |
| d34dh0r53 | make it over 9000 | 15:36 |
| *** sacharya has joined #openstack-ansible | 15:39 | |
| cloudnull | andymccr odyssey4me for master i think we should be doing something like this https://gist.github.com/cloudnull/4d7e2be5e391c1061b9b | 15:39 |
| cloudnull | which pins on the version, domain, and release | 15:40 |
| andymccr | cloudnull: so one of the problems is that the package exists already and needs to get downgraded - which it doesnt like. | 15:40 |
| openstackgerrit | Ian Cordasco proposed stackforge/os-ansible-deployment: Run the glance cache cron jobs more frequently https://review.openstack.org/176365 | 15:43 |
| odyssey4me | cloudnull digging into a possibility, feedback in a bit | 15:45 |
| openstackgerrit | Ian Cordasco proposed stackforge/os-ansible-deployment: Run the glance cache cron jobs more frequently https://review.openstack.org/176368 | 15:46 |
| openstackgerrit | Ian Cordasco proposed stackforge/os-ansible-deployment: Run the glance cache cron jobs more frequently https://review.openstack.org/176370 | 15:47 |
| sigmavirus24 | andymccr: mattt ^ reviews appreciated | 15:47 |
| *** sdake_ has quit IRC | 15:48 | |
| *** erikmwilson has quit IRC | 16:11 | |
| *** erikmwilson has joined #openstack-ansible | 16:11 | |
| openstackgerrit | Jesse Pretorius proposed stackforge/os-ansible-deployment: Implement higher priority for mariadb packages https://review.openstack.org/176382 | 16:12 |
| *** erikmwilson has quit IRC | 16:13 | |
| *** erikmwilson has joined #openstack-ansible | 16:13 | |
| openstackgerrit | Jesse Pretorius proposed stackforge/os-ansible-deployment: Fix the package pinning problem within the galera roles https://review.openstack.org/176134 | 16:17 |
| odyssey4me | cloudnull, we did a bit more testing and found that as long as the pin priority is above 1000 (ie 1001) then the mariadb stuff is properly preferred | 16:18 |
| cloudnull | we know that will work, until it inevitably doesn't this is the forever game of wack-a-mole | 16:18 |
| odyssey4me | the source of the issue appears to be a mysql library that pre-exists in Ubuntu 14.04.2 which wasn't there in 14.04.1 - which is why upgrading to the new image doesn't display the issue | 16:18 |
| odyssey4me | setting the pin priority to 1001 forces apt to replace any conflicting packages with the MariaDB versions, whereas anything below it tries to leave it in place | 16:19 |
| cloudnull | which is why i think that we need to pin to more things or removing pinning within OSAD | 16:20 |
| odyssey4me | well, MariaDB is the only thing in OSAD being pinned for kilo/master | 16:20 |
| cloudnull | it is. | 16:20 |
| Apsu | Well technically the pinning isn't the issue exactly. It's part of it, but | 16:21 |
| Apsu | The issue is that apt-get is somehow not treating -y as "really do it yes" for certain downgrade scenarios | 16:21 |
| Apsu | It requires --force-yes, which is not one of the underlying dpkg --force-* options | 16:22 |
| Apsu | It's apt-get itself (config item APT::Get::force-yes) | 16:22 |
| Apsu | But it also doesn't explain Why it needs further confirmation, aside from the fact we're downgrading | 16:22 |
| odyssey4me | Apsu that's resolve by using force in the apt task for ansible | 16:22 |
| Apsu | odyssey4me: Can actually solve it by dropping a file with the config item in /etc/apt/apt.conf.d | 16:22 |
| cloudnull | and its broken. but the greater issue is that this happened exactly like this for libvirt a few months ago. and for rabbit a few months before that, its happened for maria consistently and all throughout the havana / other releases it happened for lots of other packages. im not sure why we keep doing this . | 16:23 |
| openstackgerrit | Jesse Pretorius proposed stackforge/os-ansible-deployment: Fix the package pinning problem within the galera roles https://review.openstack.org/176134 | 16:23 |
| odyssey4me | Apsu that sounds unmanagable | 16:23 |
| cloudnull | pinning should be a deployer problem not an OSAD one. | 16:23 |
| d34dh0r53 | ^ | 16:24 |
| odyssey4me | cloudnull agreed, which is why pinning in itself needs to move out of OSAD and be the deployer's solution using something like aptly which can aggregate packages, etc | 16:24 |
| odyssey4me | but if we're using alternative repositories with common files in OSAD, we're forced to deal with pinning - this is the issue for MariaDB | 16:24 |
| Apsu | odyssey4me: cloudnull: d34dh0r53: I don't disagree with where pinning/forcing should live. But wherever it lives and how, right now priority == 1001 doesn't solve it alone in downgrade situations. | 16:25 |
| odyssey4me | why are we using a seperate repo for mariadb instead of just the Ubuntu packages? | 16:25 |
| Apsu | The actual apt-get run needs force-yes enabled. | 16:25 |
| Apsu | So that needs to happen somehow if we plan to continue managing package versioning and choose a pinning priority designed for downgrading. | 16:25 |
| cloudnull | odyssey4me the ubuntu packages dont have some of the galera packages, or rather didn't. | 16:25 |
| cloudnull | idk if thats still true. | 16:25 |
| odyssey4me | Apsu please see https://review.openstack.org/176134 (master) and https://review.openstack.org/176382 (juno) - forced package installs has been included | 16:26 |
| odyssey4me | if there is review consensus for those two patches, the methods can be backported to kilo/icehouse - if not, they can be revised until there is consensus | 16:28 |
| cloudnull | this doesn't seem to want to work when on the upstream ubuntu repos. this may be a Red herring but i'm looking more into it. | 16:31 |
| odyssey4me | cloudnull the approach I'm taking for kilo/master is to pin as little as possible and add as little extra config as possible | 16:31 |
| odyssey4me | if we really have to also pin on version, this will be messy | 16:32 |
| andymccr | odyssey4me: with force-yes i believe that works | 16:33 |
| andymccr | so i dont think you have to pin on version | 16:33 |
| andymccr | i think that change should/would work. | 16:34 |
| odyssey4me | cloudnull not working for you? | 16:34 |
| andymccr | dont be that guy cloudnull | 16:34 |
| cloudnull | it does not. the version table seems to want to only use "5.5.43-0ubuntu0.14.04.1 0" im cleaning up using dpkg to test more. | 16:38 |
| andymccr | ugh wtf | 16:38 |
| andymccr | so i had the 5.5.43 but with force-yes it did downgrade it successfully when the priority is 1001 | 16:38 |
| cloudnull | let me give that a go. | 16:39 |
| sigmavirus24 | So Kevin's (not cloudnull) feedback https://bugs.launchpad.net/openstack-ansible/+bug/1439892/comments/8 on https://bugs.launchpad.net/openstack-ansible/+bug/1439892 is pretty valid | 16:41 |
| openstack | Launchpad bug 1439892 in openstack-ansible trunk "Bindmount for Glance" [High,In progress] - Assigned to Ian Cordasco (icordasc) | 16:41 |
| sigmavirus24 | I can imagine a user creating a snapshot of a VM and trying to boot from it where the VM could be upwards of 20 GB. | 16:41 |
| sigmavirus24 | I'm going to see if the glance PTL sees value in having a way to say "Don't cache images over X size" | 16:42 |
| sigmavirus24 | Because I can see why caching basic images would be useful while caching snapshots may be less useful | 16:43 |
| Bjoern___ | sigmavirus24: thats usually the case for windows guests. They need to have at least 40G instance storage to even get it insalled | 16:43 |
| Bjoern___ | also we have customers with 1.2TB glance images | 16:43 |
| *** Bjoern___ is now known as BjoernT | 16:43 | |
| sigmavirus24 | Yeah. I think disabling the cache by default isn't the best solution | 16:43 |
| sigmavirus24 | That said, for os-a-d as a project, this is an issue | 16:43 |
| odyssey4me | sigmavirus24 I would think that snapshots whould not be cached at all | 16:44 |
| odyssey4me | *should | 16:44 |
| sigmavirus24 | Lucky for us we have a glance core who can maybe help add some flexibility to the image service so we don't have to give up the benefits of caching and we can not fill up the container by trying to cache something larger than what we have available | 16:44 |
| odyssey4me | that said, for OSAD we could possibly just make the use of the cache configurable and essentially disable it entirely, but I suppose that makes boot time suffer too much | 16:45 |
| sigmavirus24 | odyssey4me: it depends on many factors | 16:45 |
| sigmavirus24 | and really it should be a per-installation thing based on the knowledge of the people using that installation | 16:46 |
| odyssey4me | sigmavirus24 unfortunately any changes upstream will only be useful to us when liberty releases, unless it gets sneaked in as a minor patch | 16:46 |
| sigmavirus24 | "Do they boot tons of VMs from reasonably sized images? Then caching will probably help" | 16:46 |
| sigmavirus24 | odyssey4me: yeah I realize that | 16:46 |
| sigmavirus24 | I'm thinking of a more forward looking solution while we figure out what to do for juno/kilo | 16:46 |
| *** subscope_ has joined #openstack-ansible | 16:48 | |
| *** yaya has joined #openstack-ansible | 16:49 | |
| *** Mudpuppy has quit IRC | 16:51 | |
| *** Mudpuppy has joined #openstack-ansible | 16:52 | |
| *** Mudpuppy has quit IRC | 16:56 | |
| *** subscope_ has quit IRC | 17:07 | |
| *** subscope_ has joined #openstack-ansible | 17:07 | |
| sigmavirus24 | So row 15 in https://docs.google.com/spreadsheets/d/1vCTZBJKCMZ2xBhglnuK3ciKo3E8UMFo5S5lmIAYMCSE/edit?pli=1#gid=827503418 already has a liberty summit session dedicated to caching | 17:11 |
| sigmavirus24 | In the meantime, I'm going to send a separate patch to make the cache configurable (if it isn't already) | 17:11 |
| *** jwagner is now known as jwagner_lunch | 17:14 | |
| *** sdake has joined #openstack-ansible | 17:24 | |
| *** sdake has quit IRC | 17:29 | |
| *** sdake has joined #openstack-ansible | 17:30 | |
| *** woodard has joined #openstack-ansible | 17:35 | |
| *** Mudpuppy has joined #openstack-ansible | 17:42 | |
| *** Mudpuppy has quit IRC | 17:43 | |
| *** Mudpuppy has joined #openstack-ansible | 17:44 | |
| *** woodard_ has joined #openstack-ansible | 17:50 | |
| *** woodard has quit IRC | 17:50 | |
| *** woodard has joined #openstack-ansible | 17:58 | |
| *** woodard_ has quit IRC | 17:59 | |
| *** jwagner_lunch is now known as jwagner | 18:02 | |
| *** javeriak has joined #openstack-ansible | 18:25 | |
| *** stevemar has joined #openstack-ansible | 18:50 | |
| *** Shrews has joined #openstack-ansible | 18:54 | |
| *** Shrews has left #openstack-ansible | 18:55 | |
| openstackgerrit | Kevin Carter proposed stackforge/os-ansible-deployment: Fix the package pinning problem within the galera roles https://review.openstack.org/176134 | 18:59 |
| cloudnull | so with origin and release pinning set to > 1000 my issues on the physical gear were resolved. | 19:01 |
| cloudnull | and in the data structure version pinning is also a possibility, should the deployer decide to do that. | 19:03 |
| palendae | I like that approach | 19:03 |
| palendae | At the very least, exposing it should deployers wish to use it | 19:04 |
| *** lkoranda has quit IRC | 19:04 | |
| palendae | The omit filter might be of use in combination with those (not for mariadb, but offering pinning priorities) http://docs.ansible.com/playbooks_filters.html#omitting-undefined-variables-and-parameters | 19:04 |
| cloudnull | yes thats interesting default filter. i've not used it in the past, but i can see the potential . | 19:05 |
| palendae | Yeah, someone showed it to me the other day. New in 1.8, so we've not been using versions that include it for too long | 19:06 |
| palendae | So what can I do on an AIO to help test the galera patch? I ran an earlier patch set, before the > 1000 values were put in, and had a success | 19:08 |
| palendae | So I'm guessing I didn't have the right set up - I was on 14.04.2 | 19:08 |
| Apsu | There's two separate issues or concerns. | 19:09 |
| Apsu | One is the ability to install an older package from a 3rd-party repo when a newer version is available in a base repo | 19:09 |
| Apsu | P > 1000 solves that | 19:09 |
| openstackgerrit | Kevin Carter proposed stackforge/os-ansible-deployment: Fix the package pinning problem within the galera roles https://review.openstack.org/176134 | 19:10 |
| Apsu | The other issue is that if you already have a newer version of a package *installed*, apt-get -y isn't sufficient to downgrade it. | 19:10 |
| cloudnull | palendae ^ updated for a doc blurb | 19:10 |
| Apsu | That requires --force-yes, which Ansible's apt module "force: yes" parameter enables. | 19:10 |
| Apsu | Having a newer version installed already has two concerns around it | 19:11 |
| Apsu | One, dealing with regressions where we want to downgrade as part of a new release | 19:11 |
| *** lkoranda has joined #openstack-ansible | 19:11 | |
| palendae | Ok, so I'm testing with an AIO that's only been upgraded to 14.04.2,s o it's not going to have an existing install | 19:11 |
| palendae | So I might want to try with 2 AIOs | 19:11 |
| Apsu | Two, deployment on base installs that for some reason or another (such as hardware monitoring packages) already have a newer package on | 19:11 |
| palendae | Ahh, ok | 19:12 |
| Apsu | If you just test on our public cloud images, you probably won't run into it | 19:12 |
| palendae | Alright, so AIO/the gate are not too useful for this | 19:12 |
| Apsu | I think the pinning priority and the force: yes should be separately dealt with, because they have very different concerns and potential impacts | 19:12 |
| palendae | At least without installing some conflicting patches | 19:12 |
| palendae | s/patches/packages/ | 19:12 |
| Apsu | right | 19:12 |
| Apsu | Should also note that --force-yes is not a dpkg option. | 19:12 |
| Apsu | It's unrelated to --force-things, --force-downgrade or --force-all | 19:13 |
| palendae | cloudnull: Thanks for the doc change | 19:13 |
| Apsu | It's an apt-get option specifically (APT::Get::force-yes), and Only designed to make sure that by default you have to explicitly say "yes" to potentially dangerous things, like downgrading. | 19:13 |
| Apsu | That's why apt-get -y doesn't work. Just apt-get then saying "yes" does work | 19:14 |
| cloudnull | so are we good with the changes that have gone into juno https://review.openstack.org/#/c/176382/ ? | 19:15 |
| cloudnull | if so then we need to bp to icehosue . | 19:15 |
| *** stevemar has quit IRC | 19:15 | |
| cloudnull | *icehouse | 19:15 |
| claco | I guess the Q is around the force and downgrades | 19:16 |
| Apsu | cloudnull: I don't think so. Like I said, I think the forcing should be separated out and we need to discuss it in depth, since it has an upgrade/regression impact. | 19:16 |
| cloudnull | i tend to agree | 19:16 |
| Apsu | Namely, whether we want that to be automatic (and potentially break things) or manual, as release-notes | 19:16 |
| cloudnull | and while pinning this will once again for for a time, it is only a matter of time before its all broken again for some other reason. | 19:17 |
| Apsu | Sure | 19:17 |
| cloudnull | noting from the debian docs "...Debian does not encourage its use without thorough consideration." | 19:18 |
| Apsu | indeed. | 19:19 |
| *** javeriak has quit IRC | 19:19 | |
| Apsu | Downgrading in general is considered dangerous on Debian-based systems. | 19:19 |
| Apsu | Mostly because dpkg does not do dependency checking | 19:19 |
| cloudnull | ^ this ! | 19:19 |
| Apsu | Right now, that's probably ok. But at any moment, those packages could change which libraries they link against (soname bumps, their own regressions/bugfixes) | 19:20 |
| Apsu | And if they happen to drop them into the 3rd-party repo too, our pinning will pick them up | 19:20 |
| Apsu | Potentially breaking the entire system | 19:20 |
| claco | or as you mentioned, the monitoring stuff for omsa or hp doing dep things b4 we even get started | 19:20 |
| Apsu | Right | 19:20 |
| claco | the trisk is to get over the hump and I'll make sure we communicate up and outword that pinning is also volitile and repo tjhings need some prio love | 19:21 |
| Apsu | Commented and -1'd appropriately: https://review.openstack.org/#/c/176382/ | 19:22 |
| *** jaypipes has quit IRC | 19:24 | |
| *** javeriak has joined #openstack-ansible | 19:26 | |
| *** woodard has left #openstack-ansible | 19:27 | |
| *** sandywalsh has quit IRC | 19:30 | |
| *** sandywalsh has joined #openstack-ansible | 19:31 | |
| palendae | Yeah...it sounds like without some hardware that we have said agents on in automatic testing, it should be a big message to deployers | 19:35 |
| palendae | So juno has the force: yes, but master does not? | 19:36 |
| Apsu | Don't see the force in kevin's patch | 19:38 |
| Apsu | cloudnull: I see there's a urlfilters filter plugin combined in your patch. Was that intentional? | 19:38 |
| cloudnull | yes. that adds the origin type pin | 19:39 |
| Apsu | kk. I thought that's what it looked like, just making sure :) | 19:40 |
| *** javeriak_ has joined #openstack-ansible | 19:40 | |
| cloudnull | yea wanted to provide for origin, release and version type pins if needed. | 19:40 |
| Apsu | plus juan | 19:40 |
| palendae | Only one actual use of the filter in this current revision | 19:41 |
| palendae | That I see, anyway | 19:41 |
| *** javeriak has quit IRC | 19:43 | |
| cloudnull | palendae it should be in the pin file for both client and server | 19:43 |
| palendae | Yeah, sorry, 2 | 19:44 |
| palendae | netloc | 19:44 |
| *** stevemar has joined #openstack-ansible | 19:44 | |
| Apsu | -1, under loc quota | 19:45 |
| cloudnull | huh? | 19:45 |
| palendae | line of code quota | 19:46 |
| cloudnull | hahaha | 19:46 |
| Apsu | lolol | 19:48 |
| *** openstackgerrit has quit IRC | 19:54 | |
| *** openstackgerrit has joined #openstack-ansible | 19:55 | |
| *** subscope_ has quit IRC | 19:59 | |
| openstackgerrit | Kevin Carter proposed stackforge/os-ansible-deployment: Implement higher priority for mariadb packages https://review.openstack.org/176458 | 20:04 |
| openstackgerrit | Kevin Carter proposed stackforge/os-ansible-deployment: Fix the package pinning problem within the galera roles (juno) https://review.openstack.org/176458 | 20:06 |
| *** alextricity_r has joined #openstack-ansible | 20:08 | |
| *** alextricity_r has quit IRC | 20:13 | |
| *** fawadkhaliq has quit IRC | 20:15 | |
| openstackgerrit | Kevin Carter proposed stackforge/os-ansible-deployment: Fix the package pinning problem within the galera roles https://review.openstack.org/176134 | 20:16 |
| cloudnull | and now we wait for gerrit. | 20:16 |
| cloudnull | https://review.openstack.org/176458 <- Apsu pinning without the forcing | 20:17 |
| claco | who's on e the icehouse bp | 20:17 |
| cloudnull | these solution follow the upstream documentation from mariadb so they should go "https://blog.mariadb.org/dotdeb-repository-problems-with-mariadb-5-5-solution/" | 20:18 |
| cloudnull | im porting the juno fix to icehouse. | 20:18 |
| claco | we should lob that solution link towards docs | 20:19 |
| openstackgerrit | Kevin Carter proposed stackforge/os-ansible-deployment: Fix the package pinning problem within the galera roles (icehouse) https://review.openstack.org/176463 | 20:19 |
| openstackgerrit | Ian Cordasco proposed stackforge/os-ansible-deployment: Allow users to configure the Glance flavor in use https://review.openstack.org/176464 | 20:20 |
| cloudnull | Sam-I-Am: >"https://blog.mariadb.org/dotdeb-repository-problems-with-mariadb-5-5-solution/"< | 20:20 |
| * cloudnull lobbing | 20:20 | |
| openstackgerrit | Ian Cordasco proposed stackforge/os-ansible-deployment: Run the glance cache cron jobs more frequently https://review.openstack.org/176365 | 20:21 |
| openstackgerrit | Ian Cordasco proposed stackforge/os-ansible-deployment: Run the glance cache cron jobs more frequently https://review.openstack.org/176370 | 20:22 |
| cloudnull | claco https://review.openstack.org/#/c/176463/ <- icehouse fix | 20:22 |
| claco | we'll see :-) | 20:23 |
| openstackgerrit | Ian Cordasco proposed stackforge/os-ansible-deployment: Run the glance cache cron jobs more frequently https://review.openstack.org/176368 | 20:23 |
| openstackgerrit | Kevin Carter proposed stackforge/os-ansible-deployment: Fix the package pinning problem within the galera roles (icehouse) https://review.openstack.org/176463 | 20:28 |
| sigmavirus24 | I updated https://review.openstack.org/#/q/Ie7311f70656447382c2403e79453fe88401efca1,n,z because I realized that 5h 59m is what we wanted, not 6h 59m so I updated the changes for that | 20:28 |
| palendae | https://review.openstack.org/#/c/176134/ +2+W | 20:28 |
| openstackgerrit | Kevin Carter proposed stackforge/os-ansible-deployment: Fix the package pinning problem within the galera roles https://review.openstack.org/176134 | 20:29 |
| openstackgerrit | Kevin Carter proposed stackforge/os-ansible-deployment: Fix the package pinning problem within the galera roles (juno) https://review.openstack.org/176458 | 20:29 |
| cloudnull | Apsu palendae sorry updated commit messages to document why. | 20:29 |
| *** sandywalsh has quit IRC | 20:33 | |
| *** sandywalsh has joined #openstack-ansible | 20:34 | |
| mattt | palendae: yeah if you want to test this, first install libmysqclient18 before you run ansible | 20:35 |
| *** Mudpuppy has quit IRC | 20:37 | |
| *** alextricity_r has joined #openstack-ansible | 20:37 | |
| *** alextricity_r has quit IRC | 20:40 | |
| *** alextricity_r has joined #openstack-ansible | 20:41 | |
| *** stevemar has quit IRC | 20:41 | |
| *** Mudpuppy has joined #openstack-ansible | 20:42 | |
| *** KLevenstein has quit IRC | 20:42 | |
| *** stevemar has joined #openstack-ansible | 20:45 | |
| *** Mudpuppy_ has joined #openstack-ansible | 20:46 | |
| *** Mudpuppy_ has quit IRC | 20:47 | |
| *** Mudpuppy_ has joined #openstack-ansible | 20:47 | |
| *** Mudpuppy has quit IRC | 20:49 | |
| *** KLevenstein has joined #openstack-ansible | 20:50 | |
| *** Mudpuppy_ is now known as Mudpuppy | 20:53 | |
| openstackgerrit | Ian Cordasco proposed stackforge/os-ansible-deployment: Update openstack/requirements SHA for 11.0.0rc3 release https://review.openstack.org/176487 | 21:02 |
| claco | erwut | 21:05 |
| *** KLevenstein has quit IRC | 21:08 | |
| *** erikmwilson has quit IRC | 21:11 | |
| *** stevemar has quit IRC | 21:13 | |
| *** openstackgerrit has quit IRC | 21:29 | |
| *** openstackgerrit has joined #openstack-ansible | 21:30 | |
| *** sacharya has quit IRC | 21:38 | |
| *** sdake has quit IRC | 21:44 | |
| palendae | claco: I think mattt was referring to replicating the downgrading behavior on an AIO | 21:45 |
| openstackgerrit | Kevin Carter proposed stackforge/os-ansible-deployment: Fix the package pinning problem within the galera roles (icehouse) https://review.openstack.org/176463 | 21:53 |
| openstackgerrit | Kevin Carter proposed stackforge/os-ansible-deployment: Fix the package pinning problem within the galera roles (juno) https://review.openstack.org/176458 | 21:56 |
| *** yaya has quit IRC | 21:56 | |
| *** alextricity_r has quit IRC | 21:57 | |
| openstackgerrit | Kevin Carter proposed stackforge/os-ansible-deployment: Fix the package pinning problem within the galera roles https://review.openstack.org/176134 | 21:58 |
| cloudnull | so hopfully zuul is now going . . . | 21:59 |
| *** BjoernT has left #openstack-ansible | 22:03 | |
| *** Mudpuppy has quit IRC | 22:08 | |
| palendae | d34dh0r53: Those 3 reviews need another +2 | 22:09 |
| *** openstackgerrit has quit IRC | 22:11 | |
| *** openstackgerrit has joined #openstack-ansible | 22:11 | |
| claco | right about now is when they pull a maint window, because st murphy. | 22:15 |
| *** jaypipes has joined #openstack-ansible | 22:20 | |
| *** yaya has joined #openstack-ansible | 22:30 | |
| *** sdake has joined #openstack-ansible | 22:31 | |
| *** jwagner is now known as jwagner_away | 22:36 | |
| *** yaya has quit IRC | 22:39 | |
| *** JRobinson__ has joined #openstack-ansible | 22:44 | |
| openstackgerrit | Kevin Carter proposed stackforge/os-ansible-deployment: Fix the package pinning problem within the galera roles https://review.openstack.org/176134 | 22:45 |
| openstackgerrit | Ian Cordasco proposed stackforge/os-ansible-deployment: Remove the rpc_deployment directory https://review.openstack.org/176332 | 22:50 |
| *** alextricity_r has joined #openstack-ansible | 22:56 | |
| *** mnestheu1 has quit IRC | 22:56 | |
| *** alextricity_r has quit IRC | 22:57 | |
| cloudnull | so the ppa system seems to be down. so st murphy is winning here. | 22:59 |
| cloudnull | failed: [aio1_horizon_container-f93da9b5] => (item={'repo': 'ppa:adiscon/v8-stable', 'state': 'present'}) => {"attempts": 5, "failed": true, "item": {"repo": "ppa:adiscon/v8-stable", "state": "present"}} | 23:00 |
| claco | of course. never fails. | 23:00 |
| *** stevemar has joined #openstack-ansible | 23:01 | |
| *** fawadkhaliq has joined #openstack-ansible | 23:15 | |
| *** fawadkhaliq has quit IRC | 23:20 | |
| *** markvoelker has quit IRC | 23:26 | |
| sigmavirus24 | I'll be back on to help with the tagging of things in an hour or two | 23:27 |
| *** sigmavirus24 is now known as sigmavirus24_awa | 23:27 | |
| *** sdake has quit IRC | 23:39 | |
| palendae | Seems like Zuuls either really backed up, or f'ed | 23:40 |
| claco | not sure if laugh or cry | 23:41 |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!