| *** sdake has joined #openstack-ansible | 00:07 | |
| *** john51 has quit IRC | 00:10 | |
| *** cloudtrainme has quit IRC | 00:10 | |
| *** john51 has joined #openstack-ansible | 00:11 | |
| *** eil397 has joined #openstack-ansible | 00:16 | |
| *** keedya has quit IRC | 00:16 | |
| *** sdake has quit IRC | 00:23 | |
| openstackgerrit | Travis Truman (automagically) proposed openstack/openstack-ansible: Docs: Editing deployment config README for clarity https://review.openstack.org/283117 | 00:27 |
|---|---|---|
| *** weezS has quit IRC | 00:50 | |
| *** joseg has joined #openstack-ansible | 00:53 | |
| meteorfox | hi all, I think I'm almost there with configuring multi-region for Swift, but I have a question about swift-remote.yml. What IP address should I use when specifying the proxy nodes? Is it the host's storage ip, or the proxy_container IP address? | 01:02 |
| *** eil397 has quit IRC | 01:03 | |
| *** cloudtrainme has joined #openstack-ansible | 01:10 | |
| openstackgerrit | Michael Carden proposed openstack/openstack-ansible: Doc: Path error in Adding new Roles and Services https://review.openstack.org/284493 | 01:24 |
| *** jaypipes has quit IRC | 01:24 | |
| meteorfox | so, I was able to os-swift-sync.yml for my region1, I tried os-swift-sync for region 2, but I get this error: | 01:25 |
| meteorfox | https://www.irccloud.com/pastebin/ANivjHyq/ | 01:25 |
| meteorfox | I did propagate the deployment host public keys to the swift-remote_hosts in region1 | 01:25 |
| *** zhangjn has quit IRC | 01:32 | |
| *** abitha has quit IRC | 01:32 | |
| *** TheIntern has joined #openstack-ansible | 01:34 | |
| *** johnmilton has joined #openstack-ansible | 01:34 | |
| *** cloudtrainme has quit IRC | 01:37 | |
| *** zhangjn has joined #openstack-ansible | 01:37 | |
| openstackgerrit | Merged openstack/openstack-ansible: Store openstack_host_sysstat_enabled as lower case value for sysstat.default.j2 https://review.openstack.org/283337 | 01:38 |
| openstackgerrit | Merged openstack/openstack-ansible: Docs: Keystone AD/LDAP cleanup https://review.openstack.org/281883 | 01:38 |
| stevelle | meteorfox: which task was that error on | 01:45 |
| meteorfox | TASK: [os_swift_sync | Get public key contents and store as var] ************** | 01:46 |
| stevelle | meteorfox: running liberty? master branch? | 01:47 |
| meteorfox | 11.1 | 01:47 |
| *** izaakk has quit IRC | 01:47 | |
| meteorfox | that's kilo | 01:47 |
| *** Guest75 has quit IRC | 01:47 | |
| *** cloudtrainme has joined #openstack-ansible | 01:49 | |
| meteorfox | stevelle: 11.1.2 | 01:50 |
| stevelle | meteorfox: and you can verify that file doesn't exist on that host? | 01:52 |
| meteorfox | stevelle: right it doesn't exist but I don't understand why not. It worked from the primary region to the secondary region, but not the other way around. | 01:53 |
| stevelle | the swift user on that host should have /var/lib/swift as home dir | 01:53 |
| meteorfox | yes it's set | 01:55 |
| stevelle | ok, the meteorfox: os-swift-setup.yml playbook should have lay down that file under [os_swift_setup | Create the swift system user] | 01:56 |
| meteorfox | stevelle: I see. how safe is it to re-run that playbook if I already ran os-swift-sync.yml? | 01:57 |
| stevelle | meteorfox: if region 2 hasn't put together a ring yet, and doesn't have data, it should be ok | 01:58 |
| stevelle | with --tags=swift-key-create to just run the point fix | 01:58 |
| stevelle | looks like you failed on the first task of the sync playbook on region 2 so you should be fine | 01:59 |
| meteorfox | stevelle: well some of it when through, but I'm not sure if it's all or nothing | 01:59 |
| stevelle | after you do setup again on region 2, you should return to region 1 and sync before trying to sync region 2 | 01:59 |
| stevelle | but that is a recommendation from my fuzzy fuzzy memory. I need to find the docs on this | 02:00 |
| meteorfox | stevelle: ok, I see the ring files on region 2 | 02:01 |
| meteorfox | stevelle: I'm ok deleting it though, there's no data | 02:01 |
| openstackgerrit | Merged openstack/openstack-ansible: Do not setup messaging host and user when Rabbit is not installed https://review.openstack.org/284255 | 02:02 |
| openstackgerrit | Merged openstack/openstack-ansible: Move swift memcache conf into separate file https://review.openstack.org/281269 | 02:02 |
| openstackgerrit | Merged openstack/openstack-ansible: Docs: Manual nova_spice_console removal https://review.openstack.org/277686 | 02:02 |
| openstackgerrit | Merged openstack/openstack-ansible: Docs: Horizon ssl config for Kilo https://review.openstack.org/278512 | 02:02 |
| openstackgerrit | Merged openstack/openstack-ansible: Clarify user config loading in dynamic inventory https://review.openstack.org/283313 | 02:04 |
| stevelle | meteorfox: we designed the sync playbook to first do no harm, but then again we may have missed a use case. I expect the playbook would have failed rather than destroy an existing ring in region2 | 02:04 |
| meteorfox | stevelle: so, I ran the os-swift-setup.yml, and it didn't lay the file | 02:05 |
| stevelle | that task I mentioned, did it skip? | 02:06 |
| *** Guest75 has joined #openstack-ansible | 02:06 | |
| stevelle | (on the host in question) | 02:06 |
| meteorfox | https://www.irccloud.com/pastebin/S1eJNfoR/ | 02:08 |
| meteorfox | it's only laying the file on swift proxy containers, but not on the host of that container | 02:08 |
| stevelle | that's a paste of the task before the one I am looking for | 02:09 |
| stevelle | but if it's skipping the remove then that tells me the file isn't there | 02:09 |
| *** retreved_ has quit IRC | 02:10 | |
| meteorfox | stevelle: right, I could find the task you were saying, but I'm inferring that if didn't check for it, then it's probably not putting it there either | 02:10 |
| meteorfox | sorry I meant I could *not* | 02:11 |
| stevelle | meteorfox: playbooks/roles/os_swift/tasks/swift_pre_install.yml should contain "Create the swift system user" on line 37 by my reading. | 02:13 |
| stevelle | line 46 contains generate_ssh_key: "yes" which is where that key is generated | 02:13 |
| openstackgerrit | Merged openstack/openstack-ansible: Update PLUMgrid Appendix Doc paths https://review.openstack.org/283465 | 02:13 |
| *** galstrom_zzz is now known as galstrom | 02:14 | |
| meteorfox | stevelle: I do see the Create the swift system user task | 02:14 |
| meteorfox | https://www.irccloud.com/pastebin/l47GL6NL/ | 02:15 |
| openstackgerrit | Kevin Carter proposed openstack/openstack-ansible: Add condition to local IP for overlay net https://review.openstack.org/273793 | 02:15 |
| meteorfox | I think the problem is that my swift-remote.yml has the storage IP for the hosts with swift-proxy containers, but that doesn't have the ssh public key | 02:15 |
| stevelle | meteorfox: that sounds like a good lead to follow | 02:17 |
| stevelle | what you want in that swift-remote.yml is effectively every storage and proxy ip that exists in the other region | 02:20 |
| meteorfox | stevelle: right, but that's what confuses me, is it the proxy container's IP or its host's IP on the storage_network? | 02:20 |
| stevelle | meteorfox: both | 02:21 |
| stevelle | the storage is on the hosts | 02:21 |
| stevelle | it will want those, and for storage_network specifically | 02:22 |
| meteorfox | stevelle: in my case I run the proxy containers on infra nodes | 02:22 |
| *** Guest75 has quit IRC | 02:22 | |
| stevelle | for listing the proxies, you want to list the proxy container's IPs | 02:23 |
| stevelle | but everything has to talk to everything else | 02:23 |
| meteorfox | stevelle: so then, is it correct to say that I should be able to ping from the other region's host to the proxy container itself? | 02:23 |
| stevelle | the major distinction is that things in all other regions need to be in the swift-remote file | 02:23 |
| *** Bjoern_ has joined #openstack-ansible | 02:27 | |
| stevelle | meteorfox: I can definitely say you should expect to ping from a proxy container to another region's storage node and I am pretty sure you should be able to reach another region's proxy containers too | 02:28 |
| stevelle | I didn't do any of the work on network prerequisites though, so I'm at the edge of my familiarity with this. I would direct you at d34dh0r53 and rackertom | 02:29 |
| meteorfox | stevelle: so I can ping from the swift-proxy container to the other regions' storage node, but not the proxy-container on that other region | 02:29 |
| meteorfox | stevelle: also I can't ping from the storage node to a proxy-container on the other region | 02:30 |
| stevelle | meteorfox: yeah I'm guessing that is a problem but as I said, I would want to point you at one of them for verification and then follow up on more clarity in the docs for this | 02:30 |
| stevelle | I don't know where I have any of the scripts I used to rig up a network on public cloud that would give me clues on this | 02:31 |
| meteorfox | stevelle: ok, andymccr was helping me out earlier, but I'm not sure what his time zone | 02:31 |
| stevelle | he's back in EU | 02:32 |
| *** johnmilton has quit IRC | 02:32 | |
| meteorfox | ok so he's sleeping | 02:33 |
| *** johnmilton has joined #openstack-ansible | 02:33 | |
| stevelle | yeah. And it's well past time I found a refreshing beverage. | 02:33 |
| palendae | meteorfox, Yeah, UK time. rackertom is on EST US, d34dh0r53 is on vacation | 02:34 |
| meteorfox | palendae: thanks | 02:35 |
| *** johnmilton has quit IRC | 02:36 | |
| *** Guest75 has joined #openstack-ansible | 02:36 | |
| *** jcrst has quit IRC | 02:37 | |
| meteorfox | I guess the main question then is. Should a swift proxy container, its host, and the other storage nodes in one region be able to reach another swift proxy container in the other region? If yes, then I have to verify my network configuration | 02:38 |
| stevelle | meteorfox: you using a single deploy host for all regions? | 02:41 |
| meteorfox | stevelle: no | 02:41 |
| meteorfox | stevelle: by default the 2 regions can't talk to each other, we configured a VPN between them. | 02:42 |
| stevelle | meteorfox: I found some of the docs on this, rereading now | 02:43 |
| stevelle | "Static routes to the inter-region network from the containers should be defined in the openstack_user_variables.yml file" | 02:44 |
| *** galstrom is now known as galstrom_zzz | 02:44 | |
| meteorfox | stevelle: right I have that | 02:45 |
| meteorfox | it works from region1:proxy -> region2:host but not reverse | 02:45 |
| stevelle | yaml something like "group_binds: [ swift_proxy ]" in the docs here seems to be suggesting the containers should be on the repl mesh | 02:47 |
| stevelle | what I'm hearing is that region 1's pub keys were distributed to region 2 correctly, but the reverse is not true, and last we left things reigon2 doesn't even have pubkeys to distribute | 02:48 |
| meteorfox | stevelle: first part is correct. region 2 has pubkeys, but the region1 doesn't have the pubkeys on the host that has the proxy-container | 02:50 |
| stevelle | to be real specific, region1's proxys don't have the keys files that belong to region2, that is to say the region2 deploy host hasn't been able to put them onto the hosts listed in region2's conf.d/swift-remote.yml | 02:52 |
| stevelle | to return to where we started, the place where the region2 hosts and containers are supposed to be generating their keypairs (https://www.irccloud.com/pastebin/l47GL6NL/) is resulting in no keypairs on the proxys | 02:57 |
| stevelle | that's what I mean by no pubkeys to distribute. The file is missing: https://www.irccloud.com/pastebin/ANivjHyq/ | 02:58 |
| stevelle | the recap here is for my benefit meteorfox but if I'm wrong please correct me | 02:58 |
| *** joseg has quit IRC | 02:59 | |
| *** TheIntern has quit IRC | 03:01 | |
| meteorfox | stevelle: there are keypairs in the proxys but not in its hosts | 03:01 |
| meteorfox | stevelle: swift-remote.yml has the remote proxy's host IP, not the container's IP itself since I cannot reach it | 03:01 |
| stevelle | I don't *think* the infra hosts need keypairs here | 03:02 |
| stevelle | can't really recall and digging through the tasks now to see | 03:02 |
| meteorfox | stevelle: if that's the case, then my network is not correct which explains the problem | 03:02 |
| meteorfox | stevelle: that's what I'm trying to get confirmation for | 03:02 |
| stevelle | all my testing put storage on the infra hosts (3-nodes per region, each looked like an aio) | 03:05 |
| stevelle | so this is a question that all my materials are fuzzy on | 03:05 |
| meteorfox | stevelle: even our docs at Rackspace are not clear on this lol | 03:06 |
| stevelle | I blame the devs :) | 03:06 |
| meteorfox | :) | 03:06 |
| stevelle | docs did the best with the notes we gave them on stained napkins | 03:06 |
| meteorfox | yeah, pretty much. :) they had to piece it together | 03:07 |
| stevelle | ok, I'm going to guess this is a networking issue. | 03:07 |
| meteorfox | yeah, that's what I think as well | 03:07 |
| meteorfox | I sent an email to andymccr, I'll wait for his response | 03:08 |
| stevelle | but I'm also gonna make you verify that with andymccr or perhaps rackertom | 03:08 |
| stevelle | there ya go | 03:08 |
| meteorfox | stevelle: thanks for helping me out though, I appreciate it | 03:08 |
| stevelle | sorry I wasn't able to resolve | 03:09 |
| meteorfox | stevelle: no worries, you gave me so good leads, actually helped me out define my problem better | 03:09 |
| meteorfox | some* | 03:09 |
| *** Guest75 has quit IRC | 03:16 | |
| *** aslaen has joined #openstack-ansible | 03:18 | |
| *** cloudtrainme has quit IRC | 03:30 | |
| *** cemmason1 has quit IRC | 03:38 | |
| openstackgerrit | Merged openstack/openstack-ansible: Flatten deployment host section https://review.openstack.org/283814 | 03:46 |
| openstackgerrit | Ala Raddaoui proposed openstack/openstack-ansible: Add a function to check accuracy of user config https://review.openstack.org/282067 | 03:47 |
| *** Bjoern_ has quit IRC | 03:51 | |
| *** weshay has quit IRC | 04:42 | |
| *** phiche has joined #openstack-ansible | 05:16 | |
| *** sdake has joined #openstack-ansible | 05:17 | |
| *** phiche has quit IRC | 05:17 | |
| *** claco has joined #openstack-ansible | 05:18 | |
| *** phiche has joined #openstack-ansible | 05:18 | |
| *** phiche1 has joined #openstack-ansible | 05:19 | |
| claco | fyi, looks like a bad wheel has broken the requirements for 12.0.5, and maybe more: http://jenkins.propter.net/job/RPC-AIO/1810/console via SQLAlchemy-Utils | 05:21 |
| *** phiche has quit IRC | 05:22 | |
| *** sdake has quit IRC | 05:22 | |
| *** javeriak has joined #openstack-ansible | 05:38 | |
| *** javeriak has quit IRC | 05:39 | |
| *** javeriak has joined #openstack-ansible | 05:39 | |
| *** phiche1 has quit IRC | 05:45 | |
| *** phiche has joined #openstack-ansible | 05:46 | |
| *** cemmason has joined #openstack-ansible | 05:47 | |
| *** phiche has quit IRC | 05:58 | |
| *** javeriak_ has joined #openstack-ansible | 06:05 | |
| *** v1k0d3n has quit IRC | 06:05 | |
| *** v1k0d3n has joined #openstack-ansible | 06:05 | |
| *** javeriak has quit IRC | 06:07 | |
| claco | http://lists.openstack.org/pipermail/openstack-dev/2016-February/087486.html | 06:14 |
| *** aslaen has quit IRC | 06:16 | |
| *** galstrom_zzz is now known as galstrom | 06:23 | |
| *** miguelgrinberg has quit IRC | 06:25 | |
| *** miguelgrinberg has joined #openstack-ansible | 06:27 | |
| *** rgogunskiy has joined #openstack-ansible | 06:31 | |
| *** javeriak_ has quit IRC | 06:40 | |
| *** javeriak has joined #openstack-ansible | 06:45 | |
| *** fawadkhaliq has joined #openstack-ansible | 06:50 | |
| openstackgerrit | Javeria Khan proposed openstack/openstack-ansible: Update PLUMgrid Appendix Doc paths https://review.openstack.org/284564 | 07:38 |
| openstackgerrit | Javeria Khan proposed openstack/openstack-ansible: Update PLUMgrid Appendix Doc paths https://review.openstack.org/284565 | 07:41 |
| *** jiteka has quit IRC | 07:42 | |
| *** elopez has quit IRC | 07:44 | |
| *** galstrom is now known as galstrom_zzz | 07:51 | |
| *** fawadkhaliq has quit IRC | 07:51 | |
| *** jiteka has joined #openstack-ansible | 07:53 | |
| *** daneyon_ has quit IRC | 07:55 | |
| ralakus | quit | 08:09 |
| *** ralakus has quit IRC | 08:09 | |
| *** Oku_OS has quit IRC | 08:18 | |
| *** admin0 has joined #openstack-ansible | 08:22 | |
| *** admin0 has quit IRC | 08:22 | |
| *** mikelk has joined #openstack-ansible | 08:23 | |
| *** admin0 has joined #openstack-ansible | 08:24 | |
| *** admin0 has quit IRC | 08:24 | |
| *** admin0 has joined #openstack-ansible | 08:26 | |
| *** admin0 has quit IRC | 08:29 | |
| *** admin0 has joined #openstack-ansible | 08:35 | |
| jiteka | fail :D | 08:37 |
| jiteka | (08:09:07) ralakus: quit | 08:37 |
| jiteka | (08:09:27) ralakus left the room (quit: Quit: WeeChat 0.4.2). | 08:37 |
| *** admin0 has quit IRC | 08:47 | |
| *** jiteka has quit IRC | 08:48 | |
| *** jiteka has joined #openstack-ansible | 08:49 | |
| hughsaunders | mattt: you know you want to https://review.openstack.org/#/c/284057/1 | 08:51 |
| * mattt accidentally knocks his computer getting to https://review.openstack.org/#/c/284057 | 08:51 | |
| mattt | that is a horrible commit message :P | 08:52 |
| *** asettle has joined #openstack-ansible | 09:01 | |
| *** Oku_OS has joined #openstack-ansible | 09:10 | |
| *** fawadkhaliq has joined #openstack-ansible | 09:17 | |
| *** evrardjp has joined #openstack-ansible | 09:24 | |
| *** fawadkhaliq has quit IRC | 09:29 | |
| *** admin0 has joined #openstack-ansible | 09:32 | |
| *** mathlin has joined #openstack-ansible | 09:40 | |
| tiagogomes | It is a bit odd to suggest adding 3rd-party roles to ansible-role-requirements.yml | 09:46 |
| *** phiche has joined #openstack-ansible | 10:09 | |
| *** phiche has quit IRC | 10:09 | |
| *** phiche has joined #openstack-ansible | 10:14 | |
| mattt | tiagogomes: is it ? :) | 10:16 |
| tiagogomes | mattt I thought the recommended way to extend openstack-ansible was to use it as a submodule of a larger project, and not having to modify the sources of openstack-ansible. | 10:25 |
| tiagogomes | You could use the ANSIBLE_ROLE_FILE variable to specify a custom role requirements file, but then you'd have to duplicate the entries which are in the ansible-role-requirements.yml file in openstack-ansible | 10:26 |
| *** evrardjp has quit IRC | 10:27 | |
| tiagogomes | And with the plan of moving out roles of openstack-ansible to different repos, ansible-role-requirements.yml will grow (if it is the mechanism used for installing the roles) | 10:28 |
| *** electrofelix has joined #openstack-ansible | 10:32 | |
| odyssey4me | tiagogomes doing the submodule thing is not really the best way to do things | 10:38 |
| tiagogomes | I don't like submodules either. But creating a delta in openstack-ansible is worst IMO | 10:39 |
| odyssey4me | if you think there's a better way to facilitate things then I'd love to see suggestions | 10:39 |
| openstackgerrit | Hugh Saunders proposed openstack/openstack-ansible: Retry tempest cirros image creation https://review.openstack.org/284057 | 10:39 |
| odyssey4me | perhaps the bootstrap should have the option of a list of files, and the one file in repo is always included | 10:39 |
| *** pcaruana has joined #openstack-ansible | 10:42 | |
| *** admin0 has quit IRC | 10:46 | |
| *** arbrandes has quit IRC | 10:54 | |
| *** arbrandes has joined #openstack-ansible | 10:56 | |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible: Add Kilo to Liberty upgrade scaffold https://review.openstack.org/283823 | 10:56 |
| *** admin0 has joined #openstack-ansible | 10:57 | |
| *** sdake has joined #openstack-ansible | 10:58 | |
| *** slotti has joined #openstack-ansible | 11:00 | |
| *** mathlin has quit IRC | 11:02 | |
| *** admin0 has quit IRC | 11:02 | |
| *** admin0 has joined #openstack-ansible | 11:02 | |
| *** admin0 has quit IRC | 11:02 | |
| *** admin0 has joined #openstack-ansible | 11:03 | |
| *** admin0 has quit IRC | 11:05 | |
| odyssey4me | mattt hughsaunders https://etherpad.openstack.org/p/openstack-ansible-pip-conundrum | 11:14 |
| *** fawadkhaliq has joined #openstack-ansible | 11:27 | |
| *** sdake_ has joined #openstack-ansible | 11:29 | |
| *** sdake has quit IRC | 11:30 | |
| *** phiche has quit IRC | 11:31 | |
| *** admin0 has joined #openstack-ansible | 11:33 | |
| *** sdake_ has quit IRC | 11:37 | |
| *** asettle has quit IRC | 11:46 | |
| *** rgogunskiy has quit IRC | 11:51 | |
| *** Guest75 has joined #openstack-ansible | 11:57 | |
| *** neilus1 has joined #openstack-ansible | 12:02 | |
| *** MCoLo has quit IRC | 12:04 | |
| *** MCoLo has joined #openstack-ansible | 12:04 | |
| *** neilus has quit IRC | 12:05 | |
| *** Guest75 has quit IRC | 12:07 | |
| *** admin0_ has joined #openstack-ansible | 12:08 | |
| *** admin0 has quit IRC | 12:08 | |
| *** admin0_ is now known as admin0 | 12:08 | |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible: Gate: Use apt sources configured on the host https://review.openstack.org/279633 | 12:16 |
| openstackgerrit | git-harry proposed openstack/openstack-ansible: Upgrades: Cleanup RabbitMQ / vhost https://review.openstack.org/272652 | 12:19 |
| openstackgerrit | Merged openstack/openstack-ansible: Flatten the troubleshooting section https://review.openstack.org/284190 | 12:22 |
| *** johnmilton has joined #openstack-ansible | 12:31 | |
| *** phiche has joined #openstack-ansible | 12:35 | |
| *** javeriak has quit IRC | 12:42 | |
| *** fawadkhaliq has quit IRC | 12:42 | |
| odyssey4me | git-harry https://bootstrap.pypa.io/get-pip.py | 12:43 |
| *** fawadkhaliq has joined #openstack-ansible | 12:43 | |
| odyssey4me | search for 'pip<8' | 12:43 |
| *** fawadkhaliq has quit IRC | 12:47 | |
| *** pcaruana has quit IRC | 12:55 | |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible-pip_install: Ensure that the appropriate combination of setuptools, wheel and pip are installed https://review.openstack.org/284701 | 12:55 |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible-pip_install: Ensure that the appropriate combination of setuptools, wheel and pip are installed https://review.openstack.org/284701 | 12:58 |
| *** admin0 has quit IRC | 13:00 | |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible-pip_install: Ensure that the appropriate combination of setuptools, wheel and pip are installed https://review.openstack.org/284701 | 13:03 |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible-pip_install: Cap and install setuptools, wheel and pip together https://review.openstack.org/284701 | 13:05 |
| *** pcaruana has joined #openstack-ansible | 13:09 | |
| *** neilus1 has quit IRC | 13:11 | |
| *** os-tester has joined #openstack-ansible | 13:11 | |
| os-tester | hello | 13:11 |
| os-tester | Doing the "Setup hosts" in the ansible setup guide, but having some issues with null entries. | 13:12 |
| os-tester | openstack_hostname_ips.yml container entry 'container_address' are all null. | 13:13 |
| os-tester | any known and/or missing steps not documented in the current online version? | 13:14 |
| os-tester | Been following the guide step-by-step so far. | 13:15 |
| *** asettle has joined #openstack-ansible | 13:20 | |
| *** asettle has quit IRC | 13:20 | |
| *** asettle has joined #openstack-ansible | 13:21 | |
| *** neilus has joined #openstack-ansible | 13:23 | |
| *** evrardjp has joined #openstack-ansible | 13:33 | |
| odyssey4me | os-tester it sounds to me like you haven't completed the openstack_user_config correctly | 13:35 |
| odyssey4me | take a look at the example provided, and use the .aio file as an additional reference | 13:35 |
| os-tester | odyssey4me: i'll take at look at it, thanks. | 13:36 |
| *** rromans has quit IRC | 13:42 | |
| *** rromans has joined #openstack-ansible | 13:47 | |
| tiagogomes | Anyone seeing this: error in SQLAlchemy-Utils setup command: 'extras_require' must be a dictionary whose values are strings or lists of strings containing valid project/version requirement specifiers. | 13:49 |
| tiagogomes | It appeared today | 13:49 |
| *** jaypipes has joined #openstack-ansible | 13:51 | |
| *** fawadkhaliq has joined #openstack-ansible | 13:53 | |
| *** Guest75 has joined #openstack-ansible | 13:56 | |
| *** Guest75 has quit IRC | 13:56 | |
| mattt | tiagogomes: yep, it seems like a fair bit of openstack is broken by that | 14:01 |
| mattt | tiagogomes: it's due to a new release of setuptools | 14:01 |
| odyssey4me | tiagogomes yes, setuptools introduced a bug - we're waiting for an upstream fix | 14:02 |
| os-tester | odyssey4me: so, the setup playbook went all ok now. | 14:02 |
| tiagogomes | ok | 14:02 |
| odyssey4me | tiagogomes note that we'll be discussing the issues arising in the community meeting later: https://etherpad.openstack.org/p/openstack-ansible-pip-conundrum | 14:03 |
| os-tester | but all containers have an ip-address that does not exist anywhere in my playbook or configuration files... | 14:04 |
| os-tester | All ip's are within the 10.0.3.0/24 network | 14:04 |
| os-tester | That range is not specified anywhere in the configs. | 14:05 |
| *** retreved has joined #openstack-ansible | 14:05 | |
| *** slotti has quit IRC | 14:08 | |
| *** admin0 has joined #openstack-ansible | 14:09 | |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible: Cap and install setuptools, wheel and pip together https://review.openstack.org/284727 | 14:11 |
| odyssey4me | os-tester those addresses are the default lxc natted network address for the container | 14:11 |
| odyssey4me | I think | 14:12 |
| os-tester | But containers is not getting an ip address from br-mgmt network. The network is defined in the user config and the br-mgmt network is running. Should it use DHCP on that network? | 14:15 |
| *** claco has left #openstack-ansible | 14:19 | |
| *** fawadkhaliq has quit IRC | 14:20 | |
| *** keedya has joined #openstack-ansible | 14:23 | |
| *** mikelk has quit IRC | 14:25 | |
| *** Bofu2U has quit IRC | 14:26 | |
| *** mgoddard_ has joined #openstack-ansible | 14:27 | |
| hughsaunders | odyssey4me: lgtm | 14:27 |
| *** Bofu2U has joined #openstack-ansible | 14:29 | |
| admin0 | os-tester: maybe you hti somet bug like i did :D | 14:29 |
| admin0 | i have a big history of not getting ips :D | 14:29 |
| *** mgoddard has quit IRC | 14:30 | |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible: Cap and install setuptools, wheel and pip together https://review.openstack.org/284735 | 14:32 |
| os-tester | So, it seems like the containers get a random ip-address, it will be natted before going further. | 14:32 |
| *** b3rnard0_away is now known as b3rnard0 | 14:34 | |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible-pip_install: Cap and install setuptools, wheel and pip together https://review.openstack.org/284701 | 14:35 |
| *** mikelk has joined #openstack-ansible | 14:35 | |
| odyssey4me | os-tester your containers in the end will have multiple addresses - one is the address attached to the lxc default bridge | 14:35 |
| odyssey4me | others will attach to the other bridges (br-mgmt, br-storage, etc) | 14:35 |
| os-tester | yeah | 14:36 |
| os-tester | eth0 (connected to the container bridge) and eth1 (connected to host management) | 14:38 |
| openstackgerrit | Travis Truman (automagically) proposed openstack/openstack-ansible: Docs: Removing redundant config README https://review.openstack.org/283117 | 14:42 |
| os-tester | Step 4 (http://docs.openstack.org/developer/openstack-ansible/install-guide/configure-networking.html), container_interface, is that the interface seen by the container or how it should connect on the outside of the container? | 14:42 |
| odyssey4me | automagically I didn't do a thorough check of all the info in there - if you can validate that each bit of info that's useful is covered then it may be good to specify it in comments in the review so that reviewers can see and don't have to research | 14:44 |
| odyssey4me | automagically assuming you have the time to do the research, of course :) | 14:44 |
| automagically | odyssey4me: I think you are fundamentally correct | 14:44 |
| automagically | I have read and re-read most of the docs many times over | 14:45 |
| automagically | The only real missing piece is the list of host_groups, but that can be derived from reading through env.d/*.yml | 14:45 |
| *** phiche has quit IRC | 14:47 | |
| *** phiche has joined #openstack-ansible | 14:48 | |
| openstackgerrit | Travis Truman (automagically) proposed openstack/openstack-ansible: Docs: Removing redundant config README https://review.openstack.org/283117 | 14:48 |
| automagically | odyssey4me: Updated the commit message with the overall rationale | 14:48 |
| odyssey4me | automagically that sounds like something that might be useful to have somewhere in the docs so that people can implement different container->host allocations if they want to (or do on metal deployments) | 14:48 |
| automagically | Agreed, and I might find some time to add it | 14:48 |
| automagically | Since its something I’ve been digging into quite a bit as part of my deployment use cases | 14:48 |
| *** KLevenstein has joined #openstack-ansible | 14:50 | |
| *** os-tester has quit IRC | 14:50 | |
| openstackgerrit | Travis Truman (automagically) proposed openstack/openstack-ansible: Dynamic inventory: correct the inline commentary https://review.openstack.org/282452 | 14:52 |
| *** neerbeer has joined #openstack-ansible | 14:54 | |
| *** Mudpuppy has joined #openstack-ansible | 14:59 | |
| *** cemmason has quit IRC | 15:01 | |
| openstackgerrit | Travis Truman (automagically) proposed openstack/openstack-ansible: Docs: Explanation of dynamic inventory https://review.openstack.org/284457 | 15:02 |
| palendae | automagically: That edit's incorrect | 15:03 |
| automagically | How so? | 15:04 |
| palendae | automagically: Unless the --file parameter changed in another commit | 15:04 |
| automagically | Exactly: https://github.com/openstack/openstack-ansible/blob/master/playbooks/inventory/dynamic_inventory.py#L60 | 15:04 |
| palendae | Ah...is there a deprecation notice for that? | 15:04 |
| *** neerbeer has quit IRC | 15:04 | |
| automagically | No, didn’t realize one was desirable | 15:05 |
| odyssey4me | palendae should there be, considering that the dynamic inventory cannot be hooked into from outside and it's therefore an internally used function? | 15:05 |
| automagically | That was my assumption, that such a change was purely part of the “developer” internals | 15:06 |
| palendae | odyssey4me: It totally can, it's a python file, and that's the only argument | 15:06 |
| odyssey4me | hmm, I suppose the CLI parameter is something like an API | 15:06 |
| odyssey4me | ok, fair enough | 15:06 |
| palendae | Granted, calling it that way is more for developers | 15:06 |
| *** spotz_zzz is now known as spotz | 15:06 | |
| automagically | palendae: Yep, my assumption was that our dev community would figure it out reading source/doc or simply executing the inventory script with --help | 15:06 |
| palendae | There's a lot of weird baggage around that script, too, so I dunno if ansible ever calls it with the parameter. So actual deployers can't change it | 15:07 |
| palendae | automagically: I can see that, but even for developers I'd rather have a notice | 15:07 |
| palendae | There may be scripts out there depending on it, which is why I was hesitant to change it (I like the change fwiw, just thinking it may bite people) | 15:07 |
| odyssey4me | fair point - I think perhaps adding a release note to indicate the CLI parameter change would be good | 15:08 |
| automagically | K, makes sense. Looks I’ll be learning how to do a release note then | 15:08 |
| palendae | Thanks - it's fairly easy | 15:08 |
| *** jmccrory has quit IRC | 15:12 | |
| *** jmccrory has joined #openstack-ansible | 15:13 | |
| palendae | odyssey4me: Should we be doing a bunch of rechecks on gates failing because of the mysql utils thing? | 15:13 |
| palendae | Er, sqlalchemy utils | 15:13 |
| openstackgerrit | Travis Truman (automagically) proposed openstack/openstack-ansible: Release note for dynamic inventory args change https://review.openstack.org/284757 | 15:14 |
| palendae | e.g., http://logs.openstack.org/23/283823/4/check/gate-openstack-ansible-dsvm-commit/5a4833c/console.html#_2016-02-25_11_14_08_117 | 15:14 |
| *** sigmavirus24_awa is now known as sigmavirus24 | 15:15 | |
| odyssey4me | palendae nope, not yet - the issue has not yet been resolved so all branches are likely to fail builds | 15:15 |
| palendae | Ok | 15:15 |
| palendae | Will sit on my hands then :) | 15:15 |
| odyssey4me | palendae we need to discuss how we tackle this properly in the meeting | 15:15 |
| odyssey4me | palendae see https://etherpad.openstack.org/p/openstack-ansible-pip-conundrum | 15:15 |
| palendae | k. I'm not sure what the problem is other than hearing from our australians late last night | 15:16 |
| odyssey4me | palendae see the etherpad :) | 15:16 |
| palendae | Yeah, seeing now | 15:17 |
| odyssey4me | tl;dr setuptools released a broken version, everyone's waiting for a patch version | 15:17 |
| odyssey4me | bgmccollum note that gate failures since 3am UTC all relate to this bug | 15:17 |
| palendae | Not sure that anyone would read the topic, but that might be worth setting here :p | 15:18 |
| odyssey4me | lemme change the topic, and perhaps set a notice too | 15:18 |
| andymccr | id like to get the fallocate_reserve feature added to swift, by default, for OS-A: https://review.openstack.org/#/c/279589/ looking for more reviews/comments | 15:19 |
| *** neerbeer has joined #openstack-ansible | 15:20 | |
| *** galstrom_zzz is now known as galstrom | 15:20 | |
| *** ChanServ changes topic to "Launchpad: https://launchpad.net/openstack-ansible || Weekly Meetings: https://wiki.openstack.org/wiki/Meetings/openstack-ansible || Review Dashboard: https://goo.gl/tTmdgs" | 15:24 | |
| *** KLevenstein_ has joined #openstack-ansible | 15:24 | |
| *** tiagogomes___ has joined #openstack-ansible | 15:24 | |
| *** Mudpuppy_ has joined #openstack-ansible | 15:26 | |
| *** spotz is now known as spotz_zzz | 15:26 | |
| *** tiagogomes has quit IRC | 15:26 | |
| *** KLevenstein has quit IRC | 15:26 | |
| *** Mudpuppy has quit IRC | 15:26 | |
| *** KLevenstein_ is now known as KLevenstein | 15:27 | |
| *** neerbeer has quit IRC | 15:28 | |
| *** phiche1 has joined #openstack-ansible | 15:30 | |
| openstackgerrit | Ala Raddaoui proposed openstack/openstack-ansible: change host_containers group names in inventory https://review.openstack.org/283149 | 15:32 |
| *** raddaoui has joined #openstack-ansible | 15:32 | |
| *** phiche has quit IRC | 15:32 | |
| *** spotz_zzz is now known as spotz | 15:32 | |
| openstackgerrit | Ala Raddaoui proposed openstack/openstack-ansible: change host_containers group names in inventory https://review.openstack.org/283149 | 15:33 |
| *** mgoddard_ has quit IRC | 15:33 | |
| *** galstrom is now known as galstrom_zzz | 15:35 | |
| *** sdake has joined #openstack-ansible | 15:37 | |
| *** mgoddard has joined #openstack-ansible | 15:37 | |
| *** evrardjp has quit IRC | 15:37 | |
| *** phiche1 has quit IRC | 15:37 | |
| *** sdake_ has joined #openstack-ansible | 15:41 | |
| *** sdake has quit IRC | 15:42 | |
| *** rromans is now known as rromans_afk | 15:44 | |
| openstackgerrit | Ala Raddaoui proposed openstack/openstack-ansible: change host_containers group names in inventory https://review.openstack.org/283149 | 15:46 |
| *** izaakk has joined #openstack-ansible | 15:46 | |
| *** admin0 has quit IRC | 15:47 | |
| *** neerbeer has joined #openstack-ansible | 15:47 | |
| *** admin0 has joined #openstack-ansible | 15:49 | |
| *** admin0 has quit IRC | 15:54 | |
| *** Mudpuppy_ is now known as Mudpuppy | 15:54 | |
| *** daneyon has joined #openstack-ansible | 15:56 | |
| *** daneyon_ has joined #openstack-ansible | 15:57 | |
| *** sdake has joined #openstack-ansible | 15:57 | |
| *** sdake_ has quit IRC | 15:58 | |
| *** mgoddard_ has joined #openstack-ansible | 15:59 | |
| odyssey4me | community meeting in #openstack-meeting-4 cloudnull, mattt, andymccr, d34dh0r53, hughsaunders, b3rnard0, palendae, Sam-I-Am, odyssey4me, serverascode, rromans, erikmwilson, mancdaz, _shaps_, BjoernT, claco, echiu, dstanek, jwagner, ayoung, prometheanfire, evrardjp, arbrandes, mhayden, scarlisle, luckyinva, ntt, javeriak, automagically, spotz, vdo, jmccrory, alextricity25, jasondotstar, KLevenstein, admin0, | 16:00 |
| odyssey4me | michaelgugino, ametts, v1k0d3n, sever ion, bgmccollum | 16:00 |
| *** daneyon has quit IRC | 16:00 | |
| palendae | Is there a way to link to another rst document within the same tree? Would rather not use the rendered html URL if I can do an RST-style link | 16:01 |
| palendae | cloudnull: Is your Jenkins off completely, or just the emails? | 16:01 |
| cloudnull | off | 16:02 |
| palendae | k | 16:02 |
| hughsaunders | palendae: the RPC-AIO job should be a good start, that already deploys kilo and runs tempest. So you can add the upgrade bits to the end of that. | 16:02 |
| cloudnull | i nuked the vm running it | 16:02 |
| palendae | cloudnull: rgr | 16:02 |
| *** mgoddard has quit IRC | 16:03 | |
| *** sdake_ has joined #openstack-ansible | 16:11 | |
| *** sdake_ has quit IRC | 16:11 | |
| *** daneyon_ has quit IRC | 16:11 | |
| *** sdake has quit IRC | 16:11 | |
| *** sdake_ has joined #openstack-ansible | 16:12 | |
| *** phiche has joined #openstack-ansible | 16:17 | |
| *** aslaen has joined #openstack-ansible | 16:20 | |
| *** cemmason has joined #openstack-ansible | 16:23 | |
| *** aslaen has quit IRC | 16:25 | |
| *** jiteka has quit IRC | 16:31 | |
| *** MCoLo has quit IRC | 16:33 | |
| *** kencjohnston has joined #openstack-ansible | 16:33 | |
| *** MCoLo has joined #openstack-ansible | 16:34 | |
| *** Bjoern_ has joined #openstack-ansible | 16:35 | |
| *** Bjoern_ is now known as Bjoern_zZzZzZzZ | 16:35 | |
| *** sdake_ is now known as sdake | 16:36 | |
| *** cemmason has quit IRC | 16:36 | |
| *** cemmason1 has joined #openstack-ansible | 16:36 | |
| *** Bjoern_zZzZzZzZ is now known as Bjoern_ | 16:37 | |
| *** skamithi has joined #openstack-ansible | 16:41 | |
| *** weezS has joined #openstack-ansible | 16:41 | |
| *** cemmason has joined #openstack-ansible | 16:41 | |
| *** cemmason1 has quit IRC | 16:42 | |
| Nepoc | Good morning, well my day is off to a bad start. Suddenly I'm getting "ssl_.py:100: InsecurePlatformWarning" when running neutron on the utility container | 16:46 |
| *** admin0 has joined #openstack-ansible | 16:46 | |
| Nepoc | nevermind... it's all my fault | 16:47 |
| *** skamithi has quit IRC | 16:48 | |
| *** cemmason1 has joined #openstack-ansible | 16:49 | |
| *** cemmason has quit IRC | 16:50 | |
| openstackgerrit | Nolan Brubaker proposed openstack/openstack-ansible: Docs: Explanation of dynamic inventory https://review.openstack.org/284457 | 16:51 |
| *** jiteka has joined #openstack-ansible | 16:51 | |
| *** neerbeer has quit IRC | 16:53 | |
| *** mikelk has quit IRC | 16:54 | |
| *** cemmason has joined #openstack-ansible | 16:55 | |
| *** cemmason1 has quit IRC | 16:55 | |
| *** javeriak has joined #openstack-ansible | 16:56 | |
| openstackgerrit | Ala Raddaoui proposed openstack/openstack-ansible: change host_containers group names in inventory https://review.openstack.org/283149 | 17:00 |
| *** javeriak_ has joined #openstack-ansible | 17:00 | |
| cloudnull | sigmavirus24: i'd love to see a core group of stable maintianers setup to maintain the branches outside of master however at this point we've not seen that group come to be as of yet. | 17:00 |
| sigmavirus24 | cloudnull: I'm talking about openstack stable maintainers | 17:00 |
| sigmavirus24 | I think once a stable/codename branch is made and has been formally released we should pin SHAs when making tags | 17:01 |
| odyssey4me | Welcome jmccrory and automagically to the core reviewer group. Thank you for your service to this community and helping us make it better! | 17:01 |
| sigmavirus24 | Not when starting a new release cycle | 17:01 |
| sigmavirus24 | Good job jmccrory and automagically | 17:01 |
| *** javeriak has quit IRC | 17:01 | |
| odyssey4me | sigmavirus24 that's what's happening from liberty onwards as I understand it | 17:01 |
| odyssey4me | each project tags whenever they're ready | 17:01 |
| sigmavirus24 | odyssey4me: I'm talking about OSA updating SHAs at release | 17:02 |
| odyssey4me | personally I'd like us to switch from pinning sha's to rather using project tags - and let sha's be an option for deployers if they want to. | 17:02 |
| sigmavirus24 | odyssey4me: personally, I think we should be able to deploy any version of a stable branch with any version of our playbooks | 17:02 |
| odyssey4me | sigmavirus24 we pin SHA's at the beginning of every two week period to ensure that we get to test on that SHA for the whole period | 17:03 |
| odyssey4me | sigmavirus24 sure, the facility is there to do so - but we also have to test something and put a stake in the ground regarding what we tested | 17:03 |
| *** raddaoui has quit IRC | 17:03 | |
| sigmavirus24 | odyssey4me: no I get that, I still think it's wrong because the released branches are stable for a reason | 17:03 |
| odyssey4me | sigmavirus24 as far as I recall logan- uses different SHA's and some others use their own forks with their own SHA's | 17:04 |
| *** raddaoui has joined #openstack-ansible | 17:04 | |
| cloudnull | i think it would be a mistake to use the tag names instead of the SHAs from a branch . when using the SHA we're able to get fixes in faster. | 17:04 |
| odyssey4me | sigmavirus24 ah, so you're saying that our kilo tags should always deploy the head of stable/kilo upstream? | 17:04 |
| sigmavirus24 | odyssey4me: again, this is a decision I've disagreed with from the beginning so I'm not expecting this conversation to take a different turn | 17:04 |
| sigmavirus24 | odyssey4me: correct | 17:05 |
| sigmavirus24 | The volume of changes taht get backported upstream is a tiny fraction of what OSA backports | 17:05 |
| odyssey4me | sigmavirus24 I don't disagree - from liberty onwards a deployer may make that choice. | 17:05 |
| odyssey4me | as a project right now, though, that is not the convention | 17:05 |
| sigmavirus24 | In part because there are people dedicated upstream to ensuring the changes meet certain requirements and do not disrupt the stability of teh branch while fixing import user-facing issues | 17:05 |
| odyssey4me | anyway - I'd like to go back to the issue at hand | 17:05 |
| cloudnull | sigmavirus24: that'd work so long as we never tag using the branch names. | 17:06 |
| sigmavirus24 | cloudnull: right | 17:06 |
| sigmavirus24 | that's what I'm saying | 17:06 |
| sigmavirus24 | use branch up until a tag, the tagged version has SHA pins | 17:06 |
| sigmavirus24 | Same with python dependencies | 17:06 |
| sigmavirus24 | Uncapped until a tag, then a version with constraints of some sort | 17:06 |
| sigmavirus24 | And if the gate breaks in that two week period we can blacklist the version(s) that break us | 17:07 |
| sigmavirus24 | (unless we want to fix the underlying cause) | 17:07 |
| *** mgoddard_ has quit IRC | 17:07 | |
| sigmavirus24 | (assuming it is something we can fix) | 17:07 |
| *** mgoddard has joined #openstack-ansible | 17:07 | |
| cloudnull | the place where this has bit us in the past was when it broke on the day of the tag. | 17:08 |
| cloudnull | and thats happened more than a few times now. | 17:08 |
| *** neerbeer has joined #openstack-ansible | 17:09 | |
| cloudnull | which is why the two week sha set came to be | 17:09 |
| sigmavirus24 | So maybe the companies relying so heavily on OSA need to be convinced to pony up stable maintainers? | 17:09 |
| *** skamithi has joined #openstack-ansible | 17:09 | |
| cloudnull | we do need more stable maintainers. | 17:10 |
| sigmavirus24 | Again, i think fundamentally I have a different philosophy of how releases for software should be done | 17:10 |
| sigmavirus24 | And I don't think it will work well with the wounds of the times in the past where openstack has hurt us | 17:10 |
| cloudnull | wounds aside if its better and we can do it, lets. | 17:11 |
| sigmavirus24 | I think it will allow us to ship CVE backports faster and with less effort but that's my chief concern with our pinning | 17:12 |
| *** electrofelix has quit IRC | 17:12 | |
| *** admin0 has quit IRC | 17:12 | |
| cloudnull | so what are we doing with https://review.openstack.org/#/c/284727/ https://review.openstack.org/#/c/284735/ and https://review.openstack.org/#/c/284701/ ? | 17:13 |
| cloudnull | odyssey4me: ^ | 17:13 |
| odyssey4me | cloudnull I'm trying to get us back to that discussion, but the discussion digressed into how we do pinning and tagging in a space that is not relevant to the issue at hand. | 17:16 |
| odyssey4me | So I've been waiting for that tangent to complete. | 17:16 |
| cloudnull | i believe thats run its course for now | 17:16 |
| odyssey4me | we can do what's in those patches and it will band aid the situation for now, but it doesn't solve the fundamental problem | 17:17 |
| odyssey4me | we're already not using the same versions of things for our builds as all openstack testing | 17:17 |
| odyssey4me | that presents a risk to us that there may be issues experienced by our consumers which aren't issue if using the right versions of pip/setuptools/wheel | 17:18 |
| odyssey4me | this may be a risk that we're prepared to manage, but it needs to be made clear | 17:18 |
| odyssey4me | I would prefer us to figure out a better solution than pinning. | 17:18 |
| cloudnull | https://github.com/openstack/openstack-ansible/blob/master/requirements.txt -- i think those are the only requirements we have outside of os-g-r | 17:19 |
| odyssey4me | I prefer the option of using an upper constraints based on a known good set | 17:19 |
| odyssey4me | those are in g-r, mostly | 17:19 |
| odyssey4me | we have additional packages - ansible, and its deps | 17:19 |
| cloudnull | so which "other" requirements are we pulling in outside? | 17:20 |
| odyssey4me | also wheel, setuptools and some of https://github.com/openstack/openstack-ansible-repo_server/blob/master/defaults/main.yml#L83-L92 | 17:20 |
| cloudnull | those are un-pinned just as they are in os-g-r | 17:20 |
| odyssey4me | g-r only blacklists things that have come to be a problem - it doesn't cover all requirements of requirements | 17:20 |
| odyssey4me | our file http://rpc-repo.rackspace.com/os-releases/12.0.5/requirements_absolute_requirements.txt covers absolutely everything we have in the repo | 17:21 |
| odyssey4me | if a deployer has additional packages, it lists them too | 17:21 |
| odyssey4me | this is why I think it's the best upper bound we could wish for to ensure that our deployers get the same results every time they build a tag | 17:22 |
| odyssey4me | I agree that the use of it should be optional and that the URL for the file should be something that a deployer can override. | 17:23 |
| odyssey4me | I'm happy to put together a proposed patch. | 17:23 |
| odyssey4me | with the patches you listed above - we can implement those for now as a stop gap. | 17:23 |
| cloudnull | i like logan- have concerns with that approach because it tightly couples OSA to OS-infra | 17:24 |
| odyssey4me | you mean like we're now tightly coupled to rpc-repo? | 17:24 |
| cloudnull | yea | 17:24 |
| odyssey4me | we're not in either case | 17:24 |
| *** electrofelix has joined #openstack-ansible | 17:24 | |
| cloudnull | except we're adding yet another one | 17:24 |
| odyssey4me | both URL's can be changed to whichever URL you like | 17:24 |
| odyssey4me | and in time the rpc-repo link will shift to openstack infra instead | 17:25 |
| odyssey4me | as a default, at least | 17:25 |
| cloudnull | and rpc-repo is something that is being used in prod and rackspace is fine with making public, however os-infra may not appreciate folks depending on it for production | 17:26 |
| cloudnull | and it is yet another place to override for offline installs | 17:26 |
| odyssey4me | cloudnull they are expecting people to - I've had many discussions with them in that regard | 17:26 |
| odyssey4me | if that approach is such a problem, I'm all ears for an alternative solution | 17:27 |
| odyssey4me | note that part of my goal is also to eliminate caps we carry that override openstack constraints, and also to eliminate the manual work from keeping this stuff maintained | 17:28 |
| cloudnull | the fundemental problem is we require pip<8 and are not capping the other things it needs. so why isnt this the solution https://review.openstack.org/#/c/284701 so long as we're requireing pip<8 ? | 17:29 |
| odyssey4me | no, the fundamental problem is that we're capping pip at all - and that we have packages being installed on the repo server without constraints or any sort of repeatability | 17:29 |
| odyssey4me | if we cap pip and anything else openstack does not cap, then we assume greater testing responsiblity | 17:30 |
| cloudnull | no more so than we've already assumed | 17:30 |
| logan- | sorry--i am mostly afk--but I am not opposed to doing constraints or caps, I just think if any of that is implemented it needs to be done inside osa somehow so the tagged osa itself has the constraints rather than an upstream thing | 17:30 |
| logan- | especially if it is going to be done once per tag and not dynamically updated for past tagged releases | 17:31 |
| odyssey4me | cloudnull the point is that we should not be assuming it | 17:32 |
| odyssey4me | we implemented a stop-gap, thinking we were doing the right thing | 17:32 |
| odyssey4me | and it wasn't the right thing | 17:32 |
| odyssey4me | we can achieve the goal (repeatable results) in a different way | 17:32 |
| odyssey4me | ok, so taking logan-'s suggestion into account - I can make changes to the source branch updater | 17:33 |
| cloudnull | i have to run off to a meeting. sorry i cant carry on right now. | 17:33 |
| odyssey4me | it can ensure that our requirements are up to date with g-r, and that the repo-server role installs package versions constrained equal to the current sha's upper-constraints | 17:33 |
| *** sdake_ has joined #openstack-ansible | 17:34 | |
| cloudnull | however many of these solutions seem like to much machinery around this problem. maybe we simply need to re-order the repo-server tasks to always source the upstream constraints or we need to be able to live with pins at times. | 17:35 |
| odyssey4me | but that's creating a dep which you didn't want | 17:36 |
| odyssey4me | how is that different from my original suggestion? | 17:36 |
| cloudnull | idk whats best. but i'd like us to not go invent a bunch of things. | 17:36 |
| odyssey4me | it may be a different file | 17:36 |
| cloudnull | odyssey4me: its a dep that already exists | 17:36 |
| *** admin0 has joined #openstack-ansible | 17:36 | |
| *** sdake has quit IRC | 17:36 | |
| cloudnull | not yet another dep | 17:36 |
| odyssey4me | but we'd have to put new machinery down to use it | 17:37 |
| odyssey4me | and that machinery is more complex | 17:37 |
| cloudnull | is it ? | 17:37 |
| cloudnull | idk that it is | 17:37 |
| odyssey4me | meh, ok - I'm going to propose some options | 17:37 |
| odyssey4me | meanwhile - in order to unblock the gate and allow things to move on - I'll remove the WIP on those three caps | 17:37 |
| admin0 | odyssey4me: give me points/hints on how to add the new cluster as a 2nd endpoint .. and i can test and contrinute in making the documntation | 17:38 |
| cloudnull | i mean living with https://review.openstack.org/#/c/284701/5/defaults/main.yml seems like the simpelist solution if we're not going the blacklist route | 17:38 |
| admin0 | this is regarding the enable installation as an alternate region specs | 17:39 |
| odyssey4me | admin0 set the service region to whatever you want to name it, set the keystone endpoint URL's to the one you want, make sure you set the keystone admin token appropriately... that should take care of most of it | 17:40 |
| odyssey4me | admin0 if you want to be a little cleaner, you can also set the keystone container affinity to 0 to ensure that no keystone containers are created | 17:40 |
| odyssey4me | cloudnull that patch only takes care of part of the problem... I'm talking in circles here and code wins... so I'll propose options | 17:41 |
| odyssey4me | in review, not in theory | 17:43 |
| *** spotz is now known as spotz_zzz | 17:44 | |
| *** sdake_ is now known as sdake | 17:46 | |
| *** asettle has quit IRC | 17:47 | |
| *** electrofelix has quit IRC | 17:50 | |
| *** raddaoui has quit IRC | 17:51 | |
| *** electrofelix has joined #openstack-ansible | 17:51 | |
| *** abitha has joined #openstack-ansible | 17:55 | |
| openstackgerrit | Merged openstack/openstack-ansible: Set the storage address for cinder https://review.openstack.org/283234 | 17:55 |
| *** b3rnard0 is now known as b3rnard0_away | 18:01 | |
| *** sdake_ has joined #openstack-ansible | 18:01 | |
| *** sdake has quit IRC | 18:03 | |
| *** woodard has joined #openstack-ansible | 18:05 | |
| *** woodard has quit IRC | 18:06 | |
| *** woodard has joined #openstack-ansible | 18:07 | |
| *** woopstar has joined #openstack-ansible | 18:11 | |
| *** Bjoern_ is now known as Bjoern_zZzZzZzZ | 18:12 | |
| woopstar | Hi there. So i've been trying to setup an entire openstack environment now with ansible. When running chapter 5, everything runs fine and it seems to create alot of LXC containers. But the containers does not get an ip. I've added the br-mgmt network on the hosts, and configured the cidr container in user_config. But when attaching to a lxc contain | 18:12 |
| woopstar | er, there is no eth1, only a eth0 which is DHCP | 18:12 |
| *** spotz_zzz is now known as spotz | 18:14 | |
| skamithi | odyssey4me: i'm ready to upload my blueprint for vagrant AIO . I'm reading http://docs.openstack.org/infra/manual/developers.html but cannot find the specs directory. could you or someone help me for a few minutes to point me in the right direction. | 18:14 |
| odyssey4me | skamithi great! https://github.com/openstack/openstack-ansible-specs | 18:15 |
| *** spotz is now known as spotz_zzz | 18:15 | |
| woopstar | maybe a pastebin of my config and interface config would help? | 18:15 |
| *** Bjoern_zZzZzZzZ is now known as Bjoern_ | 18:17 | |
| woopstar | thing is when i'm done running step 4, all containers in openstack_hostnames_ips.yml is null, but servers are fine added with ip and such | 18:17 |
| *** phiche has quit IRC | 18:18 | |
| *** cemmason has quit IRC | 18:19 | |
| *** cemmason has joined #openstack-ansible | 18:19 | |
| woopstar | http://pastebin.com/rDV60F5L - http://pastebin.com/4cAGXZbr - http://pastebin.com/Mic2Zv39 | 18:20 |
| woopstar | Any ideas on why the containers would have a null ip | 18:21 |
| *** cemmason1 has joined #openstack-ansible | 18:25 | |
| *** neerbeer has quit IRC | 18:25 | |
| skamithi | odyssesy4me: thx cloned https://git.openstack.org/openstack-dev/openstack-ansible-specs.git. just read the spec/template.rst. I have to draw my diagram using asciiflow then will submit the blueprint for review. very new at using gerrit. played on the sandbox a little so hopeful i wont make any stupid mistakes the first time around. | 18:25 |
| *** cemmason has quit IRC | 18:26 | |
| *** woodard has quit IRC | 18:26 | |
| *** woodard has joined #openstack-ansible | 18:26 | |
| sigmavirus24 | skamithi: mistakes are fine :) | 18:27 |
| sigmavirus24 | skamithi: feel free to ping one of us if you run into problems though :D | 18:27 |
| *** cemmason has joined #openstack-ansible | 18:28 | |
| *** cemmason1 has quit IRC | 18:29 | |
| woopstar | Does the network maybe need a DHCP? | 18:30 |
| palendae | woopstar: Yep - indent your whole provider_networks stanza to be underneath global_overrides. https://review.openstack.org/#/c/282067/ is a patch for that | 18:31 |
| *** cemmason has quit IRC | 18:31 | |
| *** cemmason1 has joined #openstack-ansible | 18:31 | |
| woopstar | Oh, it must be a sub section to global_overrides ? | 18:32 |
| palendae | woopstar: Yep. I know it's hard to tell, but it's a list of dictionaries within global_overrides | 18:33 |
| woopstar | Okay.. Then it sure makes sense nothing works :D | 18:33 |
| palendae | woopstar: Yeah, someone else found that and raddaoui kindly contributed a patch, but that's not merged yet | 18:34 |
| woopstar | When setting everything up. There was a few thing i noticed. The official liberty docs uses chrony for ntp, but ansible uses ntp etc. There are alot of different things. Also this part with LXC containers for everything | 18:35 |
| *** woodard has quit IRC | 18:35 | |
| *** cemmason has joined #openstack-ansible | 18:35 | |
| *** woodard has joined #openstack-ansible | 18:35 | |
| *** cemmason1 has quit IRC | 18:35 | |
| *** neerbeer has joined #openstack-ansible | 18:39 | |
| *** cemmason1 has joined #openstack-ansible | 18:39 | |
| *** cemmason has quit IRC | 18:39 | |
| *** cemmason has joined #openstack-ansible | 18:42 | |
| *** cemmason1 has quit IRC | 18:42 | |
| palendae | For liberty upgrades, I want to go from https://review.openstack.org/#/c/283823/ to https://review.openstack.org/#/c/272182/ in a way that is digestable to reviewers. However, I'd like to get patches up without having to wait on reviews for next change in the line | 18:44 |
| palendae | Are people comfortable with me starting on the next one in sequence while that first patch is in review? | 18:45 |
| *** cemmason1 has joined #openstack-ansible | 18:46 | |
| *** cemmason has quit IRC | 18:46 | |
| *** b3rnard0_away is now known as b3rnard0 | 18:49 | |
| jmccrory | palendae sounds good to me | 18:52 |
| *** cemmason has joined #openstack-ansible | 18:52 | |
| palendae | Managing my own "branch" via gerrit dependencies isn't too appealing | 18:52 |
| palendae | But neither is just sitting on my hands | 18:52 |
| *** cemmason1 has quit IRC | 18:52 | |
| automagically | palendae: I’m a +1 on that as well | 18:53 |
| odyssey4me | ++ | 18:53 |
| admin0 | if i want to install a new cluster, but use a managed database ( by dba team ) .. how do I tell ansible to use an existing database ? | 18:53 |
| automagically | Thanks for all the work going into the upgrade stuff | 18:53 |
| odyssey4me | admin0 all the db connection string can be overridden | 18:53 |
| automagically | admin0: Check out the vars in galera_client role | 18:53 |
| palendae | Ok, I'll actually build off the vhost change, cause that's further along | 18:53 |
| palendae | Thanks | 18:53 |
| woopstar | palendae: They got ip now. Thanks.. Now the wait for ssh connection fails. Seems there is some connectivity issues. I can ping the container on the host, but not from other hosts, which i guess i should | 18:53 |
| odyssey4me | admin0 you can set the galera_server affinity to 0 as well | 18:54 |
| admin0 | odyssey4me: i know all can be overritten :D just exactly how to override them is the question :D | 18:54 |
| openstackgerrit | Nolan Brubaker proposed openstack/openstack-ansible: Docs: Explanation of dynamic inventory https://review.openstack.org/284457 | 18:54 |
| odyssey4me | admin0 just add the full string to whatever is needed to connect to the db in question | 18:54 |
| odyssey4me | admin0 alternatively, if you want to use some of the vars in place, then just override the bits you want | 18:54 |
| admin0 | odyssey4me: i use the playbooks to install clusters like 10+ times . but not a develper . so i need a bit of guidiance on that part :) | 18:55 |
| odyssey4me | admin0 an example: https://github.com/openstack/openstack-ansible/blob/master/playbooks/roles/os_keystone/templates/keystone.conf.j2#L58 | 18:55 |
| admin0 | aha | 18:55 |
| admin0 | hmm.. | 18:55 |
| odyssey4me | admin0 if it's a mysql back-end, then change the keystone_galera_address var in user_variables to the db server address you have | 18:56 |
| odyssey4me | admin0 if the db name is different to the default, then set keystone_galera_database in user_vars to the right one | 18:56 |
| *** woodard has quit IRC | 18:56 | |
| odyssey4me | and the other vars for the username/password | 18:56 |
| *** woodard has joined #openstack-ansible | 18:56 | |
| odyssey4me | if it's an entirely different database back-end, then you'd have to use a config_override mechanism to replace the whole string | 18:57 |
| odyssey4me | the DBA will give you what you need | 18:57 |
| odyssey4me | config_override guidance is here: http://docs.openstack.org/developer/openstack-ansible/install-guide/configure-openstack.html | 18:58 |
| *** cemmason has quit IRC | 18:58 | |
| admin0 | there is no keystone_galera_address in user_variables | 18:59 |
| *** spotz_zzz is now known as spotz | 18:59 | |
| odyssey4me | admin0 the stuff in iser_variables are simply there for convenience | 18:59 |
| odyssey4me | you can set any var you find in a default file in user variables | 18:59 |
| odyssey4me | any var you find in a role's defaults/main.yml file | 19:00 |
| admin0 | so i can pick any variable that i see in template files, and to override, i just place the value in user_variables :D ? | 19:00 |
| admin0 | that sounds great | 19:00 |
| odyssey4me | yep, and if there's no var then just use the config_override mechanism | 19:00 |
| *** daneyon has joined #openstack-ansible | 19:02 | |
| *** woopstar has quit IRC | 19:05 | |
| *** woodard has quit IRC | 19:05 | |
| *** woodard has joined #openstack-ansible | 19:06 | |
| *** daneyon has quit IRC | 19:06 | |
| odyssey4me | sigmavirus24 if I have the name of a pypi package, and want to determine all its dependencies - which python module should be used? the module will not be installed on the host doing the query | 19:12 |
| *** raddaoui has joined #openstack-ansible | 19:13 | |
| sigmavirus24 | odyssey4me: that's not something that is necessarily easily solvable | 19:17 |
| sigmavirus24 | odyssey4me: you need to download the package itself. Based on what kind of distribution it is, there are different ways to get the dependencies only some of which are deterministic and going to be correct | 19:18 |
| odyssey4me | sigmavirus24 do I have to pull something down, read it, then continue recursively? | 19:18 |
| sigmavirus24 | odyssey4me: if you're only handling wheels that's easier | 19:18 |
| sigmavirus24 | odyssey4me: correct | 19:18 |
| sigmavirus24 | this is how pip works | 19:18 |
| *** phalmos has joined #openstack-ansible | 19:18 | |
| odyssey4me | lovely - it sounds like this is going to be fun | 19:18 |
| sigmavirus24 | Download whatever is passed to "install", parse requirements to the best of its ability, download requirements, parse those | 19:18 |
| sigmavirus24 | odyssey4me: absolutely | 19:18 |
| sigmavirus24 | odyssey4me: sdists are hard to parse because you basically need to execute the setup.py to get information | 19:19 |
| sigmavirus24 | iirc | 19:19 |
| odyssey4me | bleh, yuk | 19:19 |
| odyssey4me | so I suppose I'll have to setup a venv, put things into it - then read all the bits out to produce my results | 19:20 |
| odyssey4me | that's the only way to automate this in the source branch updater, which seems to be the preferred method (which I hate, for the record) | 19:21 |
| sigmavirus24 | odyssey4me: oh I found a thing | 19:21 |
| odyssey4me | it's far better than maintaining caps which get forgotten though | 19:21 |
| odyssey4me | sigmavirus24 tell me about the thing :) | 19:22 |
| sigmavirus24 | A source dist, provided it was generated with setuptools, should have a {{package_name}}.egg_info directory which should have a requires.txt file | 19:22 |
| *** fawadkhaliq has joined #openstack-ansible | 19:22 | |
| sigmavirus24 | so if you do `tar xf downloaded.tar.gz` and then inside of the unpacked directory you'll find that .egg_info directory | 19:22 |
| *** phalmos has quit IRC | 19:23 | |
| sigmavirus24 | That said, if it doesn't exist, you should be able to do `python setup.py egg_info` to generate it | 19:23 |
| odyssey4me | so that still requires the download to happen, which I was hoping to avoid - but thanks... that could be useful | 19:23 |
| odyssey4me | let me get home, then I'll try and figure this out | 19:23 |
| *** raddaoui has quit IRC | 19:24 | |
| *** raddaoui has joined #openstack-ansible | 19:25 | |
| odyssey4me | palendae the very simplest way of doing the upgrade stuff to prevent you ending up with merge conflicts is to make each patch in order depend on the former one | 19:26 |
| palendae | odyssey4me: Right... | 19:26 |
| palendae | But then as each parent gets reviewed, you get conflicts down the chain | 19:26 |
| *** woodard has quit IRC | 19:26 | |
| palendae | Also my current one i might have to split even more than intended | 19:26 |
| palendae | Assuming reviewers find things that need to change, that is | 19:27 |
| odyssey4me | otherwise base everything on the scaffolding and we'll just have to deal with conflicts once the review process reaches the point of the patch being approved | 19:27 |
| *** woodard has joined #openstack-ansible | 19:27 | |
| palendae | Which I think is a fiar and valid | 19:27 |
| odyssey4me | the hard part is getting all the initial nits and corrections in documentation out the way - so perhaps the latter is the better approach | 19:27 |
| palendae | Actually starting my work off of Harry's rabbit vhost patch | 19:27 |
| odyssey4me | sure, that's fine as it was already approved | 19:28 |
| palendae | That said, I've not actually pushed any of this up | 19:28 |
| palendae | So it's not a problem til I do | 19:28 |
| palendae | I just feel uneasy having it just on my machine :p | 19:28 |
| openstackgerrit | Kevin Carter proposed openstack/openstack-ansible-repo_server: Removed pip requirements from the repo-server role https://review.openstack.org/284894 | 19:29 |
| openstackgerrit | Merged openstack/openstack-ansible: Doc: Path error in Adding new Roles and Services https://review.openstack.org/284493 | 19:30 |
| *** rgogunskiy has joined #openstack-ansible | 19:30 | |
| openstackgerrit | Kevin Carter proposed openstack/openstack-ansible: Added the pip packages for the repo-build process https://review.openstack.org/284896 | 19:32 |
| cloudnull | odyssey4me: ^ the last two commits we're what I was talking about using the cloned os-g-r upper constraints file | 19:32 |
| openstackgerrit | Nolan Brubaker proposed openstack/openstack-ansible: Docs: Explanation of dynamic inventory https://review.openstack.org/284457 | 19:33 |
| cloudnull | that should solve the constraint problem you were talking about without any additional machinery | 19:34 |
| cloudnull | or deps | 19:34 |
| *** electrofelix has quit IRC | 19:40 | |
| *** woodard has quit IRC | 19:47 | |
| *** woodard has joined #openstack-ansible | 19:47 | |
| *** krotscheck is now known as krotscheck_dcm | 19:50 | |
| openstackgerrit | Bjoern Teipel proposed openstack/openstack-ansible-rabbitmq_server: Adding ERLANG VM tuning parameters https://review.openstack.org/284899 | 19:53 |
| *** retreved has quit IRC | 19:54 | |
| *** eil397 has joined #openstack-ansible | 19:56 | |
| *** neerbeer has quit IRC | 20:00 | |
| openstackgerrit | Merged openstack/openstack-ansible: Update PLUMgrid Appendix Doc paths https://review.openstack.org/284565 | 20:03 |
| stevelle | palendae: just push the branch to your fork on gh and sleep easy? | 20:04 |
| *** eil397 has quit IRC | 20:04 | |
| *** eil397 has joined #openstack-ansible | 20:05 | |
| *** raddaoui has quit IRC | 20:05 | |
| palendae | stevelle: Yep, doing that | 20:05 |
| openstackgerrit | Nolan Brubaker proposed openstack/openstack-ansible: Docs: Explanation of dynamic inventory https://review.openstack.org/284457 | 20:05 |
| *** woodard has quit IRC | 20:08 | |
| *** woodard has joined #openstack-ansible | 20:08 | |
| openstackgerrit | Kevin Carter proposed openstack/openstack-ansible: Add support for the nova_api db https://review.openstack.org/274932 | 20:10 |
| openstackgerrit | Kevin Carter proposed openstack/openstack-ansible: Add support for the nova_api db https://review.openstack.org/274932 | 20:11 |
| *** fawadkhaliq has quit IRC | 20:13 | |
| *** fawadkhaliq has joined #openstack-ansible | 20:13 | |
| *** woodard has quit IRC | 20:17 | |
| *** raddaoui has joined #openstack-ansible | 20:18 | |
| openstackgerrit | Bjoern Teipel proposed openstack/openstack-ansible-rabbitmq_server: Adding ERLANG VM tuning parameters https://review.openstack.org/284899 | 20:21 |
| *** kencjohnston has quit IRC | 20:21 | |
| *** javeriak has joined #openstack-ansible | 20:22 | |
| *** javeriak has quit IRC | 20:23 | |
| *** javeriak_ has quit IRC | 20:25 | |
| *** cloudtrainme has joined #openstack-ansible | 20:34 | |
| *** daneyon has joined #openstack-ansible | 20:40 | |
| *** phalmos has joined #openstack-ansible | 20:44 | |
| *** phalmos has quit IRC | 20:45 | |
| *** phalmos has joined #openstack-ansible | 20:46 | |
| Nepoc | which endpoint is used when using the glance module you've created for ansible? I have secured the public and admin endpoints and left the internal non ssl and pointing to the internal lb address. When using the glance module to do "image-create" I'm getting SSL connection failures where I wasn't expecting it to communicate over ssl. | 20:51 |
| *** cloudtrainme has quit IRC | 20:51 | |
| palendae | raddaoui, izaakk, neillc: This looks like it's working it's way through Zuul's queue, but hopefully will provide you all with a reference/place to add changes as you look at the inventory pieces. https://review.openstack.org/#/c/284457/ | 20:52 |
| *** asettle has joined #openstack-ansible | 20:58 | |
| stevelle | Nepoc: look at the openrc file in the container to see which endpoint it is using (https://github.com/openstack/openstack-ansible-plugins/blob/master/library/glance#L143) | 20:58 |
| Nepoc | hmm yes I see the problem it has https | 20:59 |
| Nepoc | no wait | 20:59 |
| Nepoc | ignore that | 20:59 |
| *** cloudtrainme has joined #openstack-ansible | 21:00 | |
| Nepoc | internalURL | 21:00 |
| Nepoc | so it should be fine | 21:00 |
| Nepoc | glance | image | True | internal | http://10.140.0.10:9292 | 21:01 |
| *** asettle has quit IRC | 21:02 | |
| Nepoc | oh... actually my problem is keystone | 21:03 |
| Nepoc | Thank you for pointing me in the right direction | 21:03 |
| *** raddaoui has quit IRC | 21:04 | |
| *** eil397 has quit IRC | 21:04 | |
| *** eil397 has joined #openstack-ansible | 21:06 | |
| *** retreved has joined #openstack-ansible | 21:07 | |
| *** retreved has quit IRC | 21:08 | |
| *** retreved has joined #openstack-ansible | 21:08 | |
| *** eil397 has quit IRC | 21:11 | |
| openstackgerrit | Travis Truman (automagically) proposed openstack/openstack-ansible-lxc_hosts: Allow cert validation to be toggled by deployers https://review.openstack.org/284931 | 21:14 |
| izaakk | got it, thanks palendae | 21:17 |
| *** fawadkhaliq has quit IRC | 21:20 | |
| *** johnmilton has quit IRC | 21:20 | |
| openstackgerrit | Hector I Gonzalez Mendoza proposed openstack/openstack-ansible: Store openstack_host_sysstat_enabled as lower case value for sysstat.default.j2 https://review.openstack.org/284937 | 21:24 |
| *** phalmos has quit IRC | 21:27 | |
| *** woodard has joined #openstack-ansible | 21:28 | |
| *** phalmos has joined #openstack-ansible | 21:30 | |
| *** cloudtrainme has quit IRC | 21:33 | |
| cloudnull | odyssey4me: around ? | 21:33 |
| cloudnull | you have a chance to review the two PRs I dropped here https://etherpad.openstack.org/p/openstack-ansible-pip-conundrum | 21:33 |
| *** jwitko has joined #openstack-ansible | 21:34 | |
| *** sdake_ is now known as sdake | 21:35 | |
| jwitko | Hey All, I'm about to deploy my second openstack cluster in our production environment but its been a while since I deployed my first. I'm using the netapp plugin but I can't seem to remember or find where the IPs for the iSCSI interfaces on the netapp are configured | 21:37 |
| Nepoc | I'm not sure how only the glance container is having trouble connecting over ssl to keystone_service | 21:38 |
| *** mariognarly has quit IRC | 21:39 | |
| *** cloudtrainme has joined #openstack-ansible | 21:40 | |
| odyssey4me | cloudnull I saw them - unfortunately our dependencies between repositories don't work, so I'll have to try them manually | 21:41 |
| cloudnull | jwitko: those should be set in the openstack_user_config.yml file | 21:41 |
| odyssey4me | I like the idea though | 21:41 |
| odyssey4me | cloudnull I'll try and work out how to make the deps work properly for us tomorrow, with a fresh head. | 21:41 |
| cloudnull | is that a thing we can make go ? | 21:42 |
| cloudnull | I just kinda assumed it would always be busted and would be a good way to cross link | 21:42 |
| cloudnull | also in the meeting we never got to talk about https://review.openstack.org/#/c/284512/ | 21:43 |
| cloudnull | save it for another day ? | 21:43 |
| jwitko | cloudnull, so here is an example of the config I have from openstack_user_config.yml: http://pastebin.com/mPxRu5fD - but no where in that config are the Logical interfaces for iSCSI connections on the netapp specified | 21:43 |
| *** mariognarly has joined #openstack-ansible | 21:44 | |
| odyssey4me | cloudnull yeah, I think we can make it go - I have a pointer from infra and just need to try it out | 21:44 |
| *** woodard has quit IRC | 21:44 | |
| cloudnull | jwitko: hum... is it something that you've specified specifically for cinder or is that part of your nova config | 21:44 |
| *** woodard has joined #openstack-ansible | 21:45 | |
| cloudnull | maybe something youve set as an override in your user-variables.yml file ? | 21:45 |
| jwitko | cloudnull, negative. I've checked all my variable files. I basically grepped my entire OSAD configuration directory for the IPs of my existing working iSCSI LIFs, but they are no where to be found | 21:46 |
| *** cloudtrainme has quit IRC | 21:46 | |
| cloudnull | hum... | 21:46 |
| odyssey4me | jwitko check the netapp config guidance for openstack | 21:48 |
| jwitko | odyssey4me, looking it but not seeing anything there either! | 21:48 |
| jwitko | http://netapp.github.io/openstack-deploy-ops-guide/kilo | 21:48 |
| odyssey4me | ok, lemme peek - it's been a while :) | 21:48 |
| cloudnull | was there something set for the old cloud ? | 21:48 |
| *** Mudpuppy_ has joined #openstack-ansible | 21:49 | |
| jwitko | cloudnull, sorry not sure I understand what you mean | 21:49 |
| cloudnull | maybe the iscsi pathing is assumed. | 21:49 |
| automagically | later all | 21:49 |
| jwitko | cloudnull, I'm starting to think that may be the case? | 21:49 |
| *** eil397 has joined #openstack-ansible | 21:49 | |
| cloudnull | have a good one automagically. also congrats on core | 21:49 |
| odyssey4me | jwitko netapp_server_hostname: http://netapp.github.io/openstack-deploy-ops-guide/kilo/content/cinder.7mode.iscsi.configuration.html | 21:49 |
| jwitko | odyssey4me, thats the hostname of the netapp unit to SSH into | 21:49 |
| jwitko | not the IP for the logical interface to connect via iSCSI | 21:49 |
| odyssey4me | ah, but that is configured on the netapp itself as I recall | 21:50 |
| *** automagically is now known as automagically_aw | 21:50 | |
| cloudnull | looking at http://docs.openstack.org/kilo/config-reference/content/eseries-iscsi.html | 21:50 |
| odyssey4me | so basically the hostname needs to map to the IP which you bind to the right interface on the netapp | 21:50 |
| *** automagically_aw is now known as automagically_ | 21:50 | |
| cloudnull | i dont see anyting for specific iscsi pathing | 21:50 |
| jwitko | cloudnull, http://docs.openstack.org/kilo/config-reference/content/ontap-cluster-family.html -- this would be the appropriate one for me | 21:50 |
| cloudnull | yes . http://docs.openstack.org/kilo/config-reference/content/ontap-cluster-iscsi.html | 21:51 |
| openstackgerrit | Byron McCollum proposed openstack/openstack-ansible: Set the storage address for cinder https://review.openstack.org/284948 | 21:51 |
| cloudnull | liberty http://docs.openstack.org/liberty/config-reference/content/ontap-cluster-iscsi.html -- if thats what youre deploying | 21:51 |
| jwitko | deploying Kilo | 21:51 |
| cloudnull | kk | 21:51 |
| jwitko | so maybe an SVM can only have one set of iSCSI LIFs and it just finds them and uses them ? | 21:51 |
| *** woodard has quit IRC | 21:51 | |
| cloudnull | its sounds like yes. | 21:52 |
| jwitko | interesting... I'll check it out more | 21:52 |
| * cloudnull not a netapp expert though | 21:52 | |
| jwitko | as always thank you guys very much cloudnull, odyssey4me | 21:52 |
| *** Mudpuppy has quit IRC | 21:52 | |
| cloudnull | best of luck. | 21:52 |
| cloudnull | let us know how it goes | 21:52 |
| *** Mudpuppy_ has quit IRC | 21:52 | |
| *** woodard has joined #openstack-ansible | 21:53 | |
| odyssey4me | jwitko when I last touched all this two years back we would setup the 10gb interfaces in a bond and attach an address to the virtual interface | 21:53 |
| odyssey4me | that was the address to configure in DNS for access | 21:54 |
| jwitko | odyssey4me, pretty much doing the same thing. my nodes will have a bonded interface pair on the storage network with its own IP | 21:55 |
| jwitko | I believe it then does an iSCSI for the LUN or ID or something as a broadcast | 21:55 |
| jwitko | and thats how it connects... maybe | 21:55 |
| *** eil397 has quit IRC | 21:57 | |
| *** eil397 has joined #openstack-ansible | 21:57 | |
| bgmccollum | palendae the install guide don't make it clear those blocks should be indented...they're treated in isolation... | 21:58 |
| odyssey4me | jwitko I forget who the netapp cinder guys were, but they were extremely helpful when I approached them on IRC for help doing the config | 21:58 |
| odyssey4me | they should be noted as the authors of the whitepapers | 21:58 |
| bgmccollum | thinking about how to handle that better. maybe a skeleton file that show the correct structure before filling out each section. | 21:59 |
| jwitko | odyssey4me, great thanks. I think I understand it now | 21:59 |
| jwitko | I believe its done as a broadcast to the network range using the iSCSI initiator groups created by the netapp driver and reported back to openstack | 22:00 |
| *** woodard has quit IRC | 22:02 | |
| *** phalmos has quit IRC | 22:02 | |
| odyssey4me | jwitko heh, I just confirmed in my old configs - we used 'netapp_server_hostname=10.12.26.20' and that IP was setup like 'ifconfig data_1 10.12.26.20 netmask 255.255.254.0 partner data mtusize 9000' in the netapp | 22:04 |
| odyssey4me | in fact my old configs were the basis for http://docs.openstack.org/developer/openstack-ansible/install-guide/configure-cinder.html (see right at the bottom) | 22:05 |
| jwitko | odyssey4me, so you're actually setting the netapp server hostname to the iSCSI LIFs? I guess that could work | 22:07 |
| jwitko | but that is not how my currently operating cluster does it at all | 22:08 |
| odyssey4me | jwitko I figured removing a DNS query would be better for storage access :) | 22:09 |
| odyssey4me | jwitko that said, I basically learned how to configure the netapp and how to do the config for openstack in a saturday after spending six months asking vendors to do it | 22:10 |
| odyssey4me | so I may have done a bad job of it - rely on your better judgement :) | 22:11 |
| spotz | hehe | 22:11 |
| *** retreved has quit IRC | 22:12 | |
| *** fawadkhaliq has joined #openstack-ansible | 22:19 | |
| palendae | bgmccollum: Ah, yeah, the install guide doesn't help there | 22:20 |
| palendae | bgmccollum: I think the example yml files have it right, but you're correct... | 22:20 |
| palendae | bgmccollum: Maybe playing a global_overrides: line above in the provider_networks piece? | 22:20 |
| bgmccollum | if your following the install guide, and just copy-pasting, you're heading for fail whale | 22:20 |
| bgmccollum | will all the var file dicts get merged? cause i know the conf.d/*.yml files have a global-overrides too... | 22:21 |
| palendae | Yes | 22:21 |
| bgmccollum | maybe thats a special case | 22:21 |
| palendae | conf.d is merged | 22:21 |
| bgmccollum | but can you have global-overrides *in* the same file | 22:21 |
| palendae | Ah, you mean multiple references to it? | 22:22 |
| bgmccollum | yeah sorry | 22:22 |
| palendae | I think the last definition wins, but I have to test | 22:22 |
| palendae | ...sheeeit | 22:22 |
| bgmccollum | yeah, even the install-guide uses it multiple times | 22:22 |
| bgmccollum | well, provider-networks | 22:22 |
| palendae | Yeah | 22:23 |
| palendae | Probably cause they're broken out code blocks | 22:23 |
| bgmccollum | i think a skeleton dict for reference might be beneficial | 22:23 |
| bgmccollum | then just reword, to say "under section `foo`, add the following" | 22:23 |
| palendae | But still, I've now seen at least 2 people being bit by not indenting it | 22:23 |
| bgmccollum | or just bake the indentation in to the code blocks | 22:25 |
| palendae | Yeah, that's what I was thinking | 22:26 |
| palendae | Indentation with a comment that all this lives under global overrides | 22:26 |
| palendae | And is provided for explanation, copy/paste will be dangerous | 22:26 |
| bgmccollum | ill see if can put something together that is a little clearer | 22:27 |
| palendae | Ok | 22:27 |
| bgmccollum | because this wont be the last person to ask | 22:27 |
| palendae | I know automagically started in on some docs making more clear what groups are expected in conf | 22:27 |
| palendae | But yeah | 22:27 |
| palendae | The good news is master has an automated check for that on the way in | 22:27 |
| palendae | But clearer docs help from hitting that in the first place | 22:27 |
| bgmccollum | yeah can't wait for that too | 22:28 |
| *** cloudtrainme has joined #openstack-ansible | 22:28 | |
| *** cloudtra_ has joined #openstack-ansible | 22:34 | |
| *** john51_ has joined #openstack-ansible | 22:34 | |
| *** cloudtrainme has quit IRC | 22:35 | |
| bgmccollum | Bofu2U -- cinder storage network fixes merged to master -- https://review.openstack.org/#/c/283234/ | 22:35 |
| *** john51 has quit IRC | 22:38 | |
| jwitko | odyssey4me, so I've moved past that... but now I'm a bit confused on how netapp driver is chosing which volumes to stores the data to | 22:42 |
| *** KLevenstein has quit IRC | 22:58 | |
| *** eil397 has quit IRC | 23:00 | |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible: Use current, but pinned versions of pip, setuptools and wheel https://review.openstack.org/284977 | 23:02 |
| palendae | Zuul's catching up on our gate, but I think it's a little loaded right now. May be worth double checking http://status.openstack.org/zuul/ before you ask for another recheck, it just might be in queue | 23:03 |
| palendae | With that, good night :) | 23:03 |
| *** spotz is now known as spotz_zzz | 23:04 | |
| odyssey4me | jwitko I remember scratching my head on that one too, whcih is when I finally asked for help | 23:04 |
| admin0 | failing here: TASK: [galera_client | Install pip packages] ********************************** | 23:05 |
| admin0 | failed: [c11_galera_container-66834f37] => (item=MySQL-python) => {"attempts": 5, "cmd": "/usr/local/bin/pip install MySQL-python", "failed": true, "item": "MySQL-python"} msg: Task failed as maximum retries was encountered | 23:06 |
| admin0 | never failed here before | 23:06 |
| admin0 | trying to understand where/what could have gone wrong | 23:06 |
| jwitko | odyssey4me, yea I'm pinging in #openstack-netapp, no responses so far. | 23:07 |
| odyssey4me | jwitko as I recall, by default the driver uses a vserver (or maybe it's something else I can't remember the term) called 'openstack'... create that and the driver will automagically start creating LUN's in it | 23:07 |
| jwitko | odyssey4me, correct it uses a SVM (storage virtual machine) | 23:07 |
| jwitko | but those contain many volumes | 23:07 |
| jwitko | so its creating these LUNs, you're correct on that | 23:07 |
| odyssey4me | create one volume called 'openstack' and try it? | 23:07 |
| jwitko | and some luns go to "/vol/a" and some go to "/vol/b" | 23:07 |
| jwitko | there doesn't seem to be a rhyme or reason | 23:08 |
| odyssey4me | jwitko it's something rather simple which I laughed and had a stuff drink after finding out | 23:09 |
| odyssey4me | jwitko try #openstack-cinder too | 23:09 |
| odyssey4me | also note that they may take a few hours to respond | 23:10 |
| odyssey4me | jwitko one of these guys helped me eventually: http://netapp.github.io/openstack-deploy-ops-guide/kilo/content/authors-and-contributors.html | 23:11 |
| *** admin0 has quit IRC | 23:11 | |
| *** woodard has joined #openstack-ansible | 23:15 | |
| *** xek_ has joined #openstack-ansible | 23:19 | |
| *** xek has quit IRC | 23:19 | |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible: Use current, but pinned versions of pip, setuptools and wheel https://review.openstack.org/284977 | 23:22 |
| odyssey4me | cloudnull ^ tweaked a little to add support for updating the requirements files in-repo | 23:22 |
| bgmccollum | admin0 how man repo containers do you have? | 23:22 |
| bgmccollum | he gone | 23:22 |
| cloudnull | he needs a bouncer :) | 23:23 |
| bgmccollum | isn't there something broken with syncing between multiple repo containers? | 23:23 |
| *** sigmavirus24 is now known as sigmavirus24_awa | 23:23 | |
| bgmccollum | curl, 404, curl 404, curl 200 | 23:23 |
| cloudnull | bgmccollum: what are you seeing ? | 23:23 |
| odyssey4me | bgmccollum not usually, but it does sometimes happen | 23:23 |
| bgmccollum | was referring to admin0 | 23:23 |
| bgmccollum | cause ive seen it fail at that point too | 23:24 |
| cloudnull | check that lsync is running on the first node | 23:24 |
| odyssey4me | and that it's completed its sync | 23:24 |
| bgmccollum | recall a review out there in the ether...searches | 23:24 |
| odyssey4me | yep, hughsaunders has one in the list | 23:24 |
| cloudnull | bgmccollum: https://review.openstack.org/#/c/279452/ | 23:25 |
| bgmccollum | cloudnull thanks | 23:26 |
| *** woodard has quit IRC | 23:31 | |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible: Add support for the nova_api db https://review.openstack.org/274932 | 23:33 |
| odyssey4me | cloudnull tweaked your releasenote - now it's up to spotz_zzz or mcarden to do the final nits on it :) ^ | 23:34 |
| cloudnull | sounds good to me | 23:35 |
| *** Bjoern_ has quit IRC | 23:36 | |
| odyssey4me | cloudnull posted a Q on https://review.openstack.org/284896 | 23:38 |
| openstackgerrit | Kevin Carter proposed openstack/openstack-ansible: IRR - Implemented for os_keystone https://review.openstack.org/279710 | 23:38 |
| openstackgerrit | Kevin Carter proposed openstack/openstack-ansible: IRR - Implemented for all os_* and dep roles https://review.openstack.org/284987 | 23:38 |
| cloudnull | odyssey4me: replied | 23:40 |
| odyssey4me | cloudnull I think you misunderstood - there is no task to install those packages | 23:42 |
| odyssey4me | not that I see anyway | 23:42 |
| cloudnull | https://review.openstack.org/#/c/284896/1/playbooks/roles/repo_build/tasks/repo_clone_git.yml | 23:42 |
| cloudnull | right after the clone | 23:42 |
| cloudnull | it clones the repos needed which we know contains the global requirements repo and uses the upper constraint it provides. | 23:43 |
| odyssey4me | ah, I must've been tired - I mistook that for the repo-clone process instead of the git-clone process | 23:43 |
| *** fawadkhaliq has quit IRC | 23:47 | |
| mcarden | odyssey4me: Nitted as requested. :) | 23:48 |
| jwitko | odyssey4me, its the cinder scheduler. | 23:49 |
| jwitko | odyssey4me, since I had no filters it was defaulting to whichever volume was best suited according to filter defaults at the time | 23:49 |
| jwitko | so as one filled up, it would use the other | 23:49 |
| jwitko | hence why its using multiple volumes for me | 23:49 |
| * jwitko durrr | 23:49 | |
| cloudnull | odyssey4me: updated https://review.openstack.org/#/c/284512/ | 23:50 |
| cloudnull | it now has all of the repo upstreams | 23:50 |
| *** skamithi has quit IRC | 23:50 | |
| cloudnull | which are copy/paste version of the bits in the main repo | 23:50 |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!