| *** jascott1 has quit IRC | 00:00 | |
| *** yingjun has quit IRC | 00:05 | |
| *** salv-orlando has quit IRC | 00:14 | |
| *** yingjun has joined #openstack-kolla | 00:21 | |
| *** Pavo has quit IRC | 00:35 | |
| *** itlinux has quit IRC | 00:37 | |
| *** Pavo has joined #openstack-kolla | 00:43 | |
| *** zshi has joined #openstack-kolla | 00:50 | |
| *** hieulq has joined #openstack-kolla | 00:54 | |
| *** itlinux has joined #openstack-kolla | 00:56 | |
| *** tovin07_ has joined #openstack-kolla | 00:58 | |
| *** hieulq has quit IRC | 01:02 | |
| *** hieulq has joined #openstack-kolla | 01:04 | |
| *** itlinux has quit IRC | 01:06 | |
| masber | I tried deploy openstack with external ceph using kolla-ansible. Deployment is successful but I can't upload/create an image https://bpaste.net/show/77d2ab681ebc any thoughts? | 01:11 |
|---|---|---|
| *** salv-orlando has joined #openstack-kolla | 01:15 | |
| masber | ok, I found the error https://bpaste.net/show/65419b66706d | 01:28 |
| masber | so no permission to connect to ceph cluster, my question is which container is supposed to have the ceoh.client.glance.keyring and ceph.conf files? | 01:28 |
| *** itlinux has joined #openstack-kolla | 01:30 | |
| *** kiennt26 has joined #openstack-kolla | 01:33 | |
| *** caowei has joined #openstack-kolla | 01:41 | |
| *** salv-orlando has quit IRC | 01:49 | |
| *** zhangfei has joined #openstack-kolla | 01:56 | |
| masber | hi, it is me again with another question. Do I need to create 2 users cinder and cinder-backup or just 1 user called cinder and give him access to both volumes and backups volumes? | 02:17 |
| masber | this is for kolla-ansible ceph external | 02:17 |
| *** daidv has joined #openstack-kolla | 02:21 | |
| *** duonghq has joined #openstack-kolla | 02:22 | |
| *** Pavo has quit IRC | 02:25 | |
| Jeffrey4l | duonghq, please review https://review.openstack.org/521372 which blocks the gate. | 02:31 |
| lxdong | failed: [compute01] (item=/usr/share/kolla-ansible/ansible/roles/haproxy/templates/haproxy.cfg.j2) => {"changed": false, "failed": true, "item": "/usr/share/kolla-ansible/ansible/roles/haproxy/templates/haproxy.cfg.j2", "msg": "AnsibleUndefinedVariable: 'dict object' has no attribute 'api_interface'"} | 02:34 |
| duonghq | Jeffrey4l, done | 02:34 |
| Jeffrey4l | thx | 02:34 |
| duonghq | np | 02:34 |
| Jeffrey4l | lxdong, guess wrong configurations. no ip address on the api_interface on the copute01 node? | 02:34 |
| lxdong | configuration didn't change. | 02:36 |
| lxdong | both compute01 and compute02 came across this failure. | 02:38 |
| Jeffrey4l | lxdong, are you using a fact cache? or run some cli to check the whether there is a address on api_interface on these two nodes | 02:40 |
| *** salv-orlando has joined #openstack-kolla | 02:46 | |
| lxdong | What if it is using cache,reboot server? Do you think it's RAID card that is using cache? | 02:47 |
| openstackgerrit | Merged openstack/kolla-ansible master: Updated from global requirements https://review.openstack.org/520423 | 02:49 |
| Jeffrey4l | lxdong, if you never configure ansible.cfg, you are not using the ansible fact cache. | 02:52 |
| lxdong | I am stuck here. | 02:54 |
| spsurya | morning guys | 02:54 |
| spsurya | sup Jeffrey4l duonghq | 02:55 |
| Jeffrey4l | morning spsurya | 02:55 |
| Jeffrey4l | spsurya, could you review https://review.openstack.org/521372 | 02:55 |
| spsurya | sure | 02:55 |
| spsurya | Jeffrey4l: | 02:55 |
| Jeffrey4l | and https://review.openstack.org/508046 spsurya ;D | 02:55 |
| Jeffrey4l | lxdong, check the environment carefully. i am still thinking you are using the wrong configuration. may the interface name is different on the compute01 and 02 | 02:56 |
| lxdong | Is this due to update of kolla and kolla-ansible to the latest? | 02:56 |
| spsurya | Jeffrey4l: https://review.openstack.org/#/c/521372/1 done | 02:57 |
| Jeffrey4l | lxdong, do not think so. | 02:57 |
| Jeffrey4l | thx. | 02:58 |
| lxdong | exactly the same | 02:59 |
| openstackgerrit | Jeffrey Zhang proposed openstack/kolla stable/pike: Bump kafka version to 0.10.2.1 https://review.openstack.org/521405 | 03:02 |
| openstackgerrit | Jeffrey Zhang proposed openstack/kolla stable/ocata: Bump kafka version to 0.10.2.1 https://review.openstack.org/521376 | 03:03 |
| openstackgerrit | Jeffrey Zhang proposed openstack/kolla stable/ocata: Pin fluent-plugin-grok-parser to 2.1.4 https://review.openstack.org/521336 | 03:04 |
| openstackgerrit | Jeffrey Zhang proposed openstack/kolla stable/ocata: implement deploy jobs in kolla https://review.openstack.org/521333 | 03:04 |
| openstackgerrit | Jeffrey Zhang proposed openstack/kolla stable/ocata: Dockerhub publisher job https://review.openstack.org/521334 | 03:04 |
| openstackgerrit | Chason Chan proposed openstack/kolla stable/pike: Fix the Ubuntu Fluentd image https://review.openstack.org/521045 | 03:05 |
| lxdong | should #api_interface: "{{ network_interface }}" be commented out? | 03:06 |
| lxdong | which is commented out by default. | 03:07 |
| duonghq | hi spsurya | 03:15 |
| openstackgerrit | Duong Ha-Quang proposed openstack/kolla master: Modify the version of skydive to v0.14.0 https://review.openstack.org/520851 | 03:17 |
| *** ArminderSingh has quit IRC | 03:18 | |
| *** Pavo has joined #openstack-kolla | 03:26 | |
| *** ArminderSingh has joined #openstack-kolla | 03:28 | |
| *** dave-mccowan has quit IRC | 03:39 | |
| openstackgerrit | Heetae Ahn proposed openstack/kolla master: Add RHEL subscription registration https://review.openstack.org/495148 | 03:45 |
| *** mdnadeem has joined #openstack-kolla | 03:47 | |
| *** kolla-slack has quit IRC | 04:00 | |
| *** kolla-slack has joined #openstack-kolla | 04:00 | |
| lxdong | can two all-in-one nodes achieve high availability? | 04:10 |
| lxdong | toke down one node and instances can't be created. | 04:14 |
| *** coolsvap has joined #openstack-kolla | 04:22 | |
| *** yangyapeng has joined #openstack-kolla | 04:26 | |
| *** owalsh_ has joined #openstack-kolla | 04:57 | |
| *** Pavo has quit IRC | 05:01 | |
| *** owalsh has quit IRC | 05:01 | |
| *** janki has joined #openstack-kolla | 05:03 | |
| SamYaple | lxdong: no | 05:04 |
| SamYaple | lxdong: but not because of openstack | 05:04 |
| SamYaple | for ceph and galera, you need at least 3 nodes to do that safely | 05:05 |
| SamYaple | if you arent using ceph then *kind of* | 05:05 |
| SamYaple | a properly shutdown mariadb node thats connected to a galera cluster will *not* cause the galeracluster tofail, but an improperly shutdown or network paritioned node will cause the galera cluster to stop working | 05:06 |
| SamYaple | rabbitmq/memcached/openstack services are all fine with twonodes | 05:06 |
| masber | lxdong, If you have 2 nodes just do a hyperconverge installation cluster, that way you should be able to shut down 1 node and still work | 05:07 |
| SamYaple | masber: thats not quite correct, see my post above | 05:07 |
| SamYaple | and is entirely incorrect for ceph | 05:08 |
| masber | SamYaple, yes, that is not the right way of deploying openstack/ceph, I just got the impression he was just testing openstack | 05:10 |
| SamYaple | sure, but even in that case itwont workwith ceph | 05:10 |
| masber | so I am trying to make kolla-ansible to work with external ceph... so far I got glance working, now I would like to get cinder working... I get an error when I try to create a volume 2017-11-20 15:37:46.114 7 ERROR cinder.scheduler.flows.create_volume [req-07141267-75ab-4d1b-a29d-1316ccc4a693 e07dcfd877b34a5dafb3386231ef92f1 fe558596e7e542b080e0daa89ce56152 - default default] Failed to run task cinder.scheduler.flows.create_volume.ScheduleCreateVol | 05:11 |
| masber | umeTask;volume:create: No valid backend was found. No weighed backends available: NoValidBackend: No valid backend was found. No weighed backends available | 05:11 |
| SamYaple | sounds like the configmight be off | 05:11 |
| SamYaple | can you post the generated cinder.conf | 05:11 |
| masber | SamYaple, you mean cinder_api:/etc/cinder/cinder.conf? kolla deploys different cinder.conf depending on the container | 05:12 |
| SamYaple | yep | 05:13 |
| *** skramaja has joined #openstack-kolla | 05:13 | |
| lxdong | SamYaple will you post your post ? masber what is the correct way to create hyperconverged infrastrature, I meant to but seems incorrect by far. | 05:13 |
| masber | SamYaple, https://bpaste.net/show/c12aaf8a2f60 | 05:14 |
| lxdong | that't to say two nodes can never achieve HA ? | 05:15 |
| masber | SamYaple, rbd configuration is missing on cinder_api:/etc/cinder/cinder.conf but it does exists on cinder_volume containers. I don't know if that is expected? | 05:17 |
| masber | lxdong, sorry I got confused, I used 3 nodes in total, 1 nodes as controller and 2 of then hyperconverge for computing and storage | 05:19 |
| lxdong | In the case I stimulate one node power outrage which should be a improperly shutdown for galera and ceph -s timeout which suggest ceph is not functional , which is the root cause or they are independent? | 05:22 |
| lxdong | masber ,can your cluster achieve instance HA using ceph backend? | 05:24 |
| masber | lxdong, yes, but as SamYaple said that is not the right way to do that | 05:25 |
| SamYaple | masber: you arent configuring cinder ceph | 05:25 |
| SamYaple | lxdong: two nodes can never achieve HA for ceph.period | 05:25 |
| SamYaple | they can for galera, kind of. but its veryunstable | 05:25 |
| *** goutham has joined #openstack-kolla | 05:25 | |
| masber | SamYaple, ceph configuration needs to go to cinder_api, cinder_volume and cinder_backup? | 05:25 |
| SamYaple | and kolla doesnt properly shutdown galera cluster | 05:25 |
| lxdong | May I look at your multinode.yml masber? | 05:26 |
| SamYaple | masber: just for your info, while sometimes some confis do generate differently per service, i never got around to doing that in cinder.conf. so those configs should be identical | 05:27 |
| lxdong | sorry ,multinode | 05:27 |
| *** caowei has quit IRC | 05:28 | |
| *** jascott1 has joined #openstack-kolla | 05:29 | |
| SamYaple | masber: https://github.com/openstack/kolla-ansible/blob/master/ansible/roles/cinder/templates/cinder.conf.j2#L107-L121 you need to enable theoptions at the top of that if block | 05:29 |
| goutham | Hi all if at all if i want a stable openstack to be deployed can i go with master or should i go with pike?? | 05:29 |
| SamYaple | lxdong: if you want ceph with two nodes, youll never be able touse ceph if one goes down. no need to dig into that further | 05:29 |
| goutham | any idea?? after the recent fluentd issues?? | 05:29 |
| goutham | or is the newton a stable setup ?? if i have to deploy newton which tag should i be using?? any idea?? | 05:30 |
| lxdong | yes ,your HCI worked well right? I am doing the same thing and tried external ceph too, reference to your configuration might be a good idea. | 05:32 |
| SamYaple | masber: do you deploy 2 nodes as well? | 05:33 |
| *** jascott1 has quit IRC | 05:33 | |
| SamYaple | if so youllwant to revist you ha testing, because ceph is down when one of those nodes is down | 05:33 |
| masber | SamYaple, for testing yes, unfortunately I don't enough nodes to do everything :( | 05:34 |
| SamYaple | thats fine,im just saying, your ceph is 100% down when one node is down, its not ha | 05:34 |
| masber | SamYaple, even if the monitor is on the openstack controller? | 05:35 |
| masber | SamYaple, 2 nodes only for OSDs and the monitor running in a different node | 05:35 |
| lxdong | that's to say there are four nodes in production, one contoller ,three hyperconverged nodes? | 05:35 |
| SamYaple | masber: well thats not two nodes | 05:35 |
| SamYaple | masber: and also not ha, since ifone node goes down(controller) everything stops | 05:36 |
| masber | lxdong, 3 in total, 1 nodes for controller which also runs ceph monitor and another 2 nodes for OSDs and nova | 05:36 |
| lxdong | so masber got a minial of | 05:36 |
| masber | SamYaple, yes you are right kolla-ansible deploys monitor on controller but on the inventory file you only specify 2 nodes as storage... my bad ignoring that | 05:37 |
| masber | (if you deploy ceph through kolla-ansible) | 05:38 |
| SamYaple | sure. thats shouldmakeyour data resliant. but thats farfrom "higly-available" just to clear up confusion | 05:38 |
| lxdong | masber ,in your case external is the one meet your demand because you can't specify the node deploy ceph monitor to in inventory file? | 05:41 |
| SamYaple | lxdong: you can specify the monitor deploys in kolla-ansible too | 05:42 |
| masber | lxdong, yes plus I also want to learn ceph :) | 05:43 |
| lxdong | me too :) | 05:44 |
| lxdong | SamYaple ,how? | 05:48 |
| lxdong | change [ceph-mon:children] ceph to[ceph-mon:children] control ? | 05:50 |
| openstackgerrit | Merged openstack/kolla-ansible master: Remove setting of version/release from releasenotes https://review.openstack.org/521022 | 05:51 |
| SamYaple | lxdong: how ever you want to do it, sure. but the inventory file is 100% in your control, save for the naming of the groups | 05:55 |
| duonghq | Jeffrey4l, are you there? | 06:02 |
| Jeffrey4l | duonghq, yes. | 06:02 |
| *** caowei has joined #openstack-kolla | 06:03 | |
| duonghq | currently, we are using ceph image which is built by our? | 06:03 |
| lxdong | good point , at least my guess is right :) You mean two nodes with one controller is far from HA? | 06:03 |
| Jeffrey4l | duonghq, docker images? yes. | 06:03 |
| *** threestrands has quit IRC | 06:03 | |
| duonghq | Jeffrey4l, do we have plan switch to ceph one? | 06:03 |
| *** caowei has quit IRC | 06:04 | |
| Jeffrey4l | duonghq, this is talked and no any approach. | 06:04 |
| Jeffrey4l | duonghq, any reason why you want to ceph's? | 06:04 |
| duonghq | I just wonder why don't we just use image provided by ceph community, so that we do not need maintaining our image | 06:05 |
| SamYaple | duonghq: kolla currently builds *all* of its own images | 06:07 |
| SamYaple | it breaks thekolla model to use upstream images | 06:07 |
| SamYaple | to do that youd have to inject startup scripts at runtime (which i agree with) but others dont want to do that | 06:07 |
| Jeffrey4l | agree. | 06:08 |
| duonghq | SamYaple, can you elaborate the "model" here? is it we want to control the way image is built and work? | 06:08 |
| Jeffrey4l | the way we use the image. | 06:09 |
| Jeffrey4l | like configuration file, how BOOTSTRAP the ceph mon or ceph osd | 06:09 |
| SamYaple | duonghq: kolla has extend_start.sh and other scripts like that. at the time we all felt that wasa gooddirection | 06:09 |
| duonghq | I got the idea, thanks Jeffrey4l and SamYaple | 06:10 |
| SamYaple | as time moved on, itwas clear that the upstream moved a different direction. to align kolla with upstream, we would need to inject the extend_start.sh scripts at run time | 06:10 |
| Jeffrey4l | the better way, it integrate kolla with ceph official containerized deployed ceph. | 06:10 |
| SamYaple | im all for that if youcan getthe community to agree :) as of yet, i have not been able to | 06:10 |
| duonghq | Jeffrey4l, I think it's one of the external ceph option? | 06:10 |
| Jeffrey4l | duonghq, yes. external ceph should work. | 06:11 |
| Jeffrey4l | so deploy ceph by using whatever you want, and kolla just use it. | 06:11 |
| duonghq | afaik, we completed external ceph support? | 06:11 |
| Jeffrey4l | yes we are. some guys is using this feature. | 06:12 |
| Jeffrey4l | but tbh, there is no test to cover this. so not sure whether it works very well now. | 06:12 |
| duonghq | hmm, we do not have test about external ceph or we have not had tests? | 06:13 |
| Jeffrey4l | we do not have test/gate jobs about external ceph. | 06:13 |
| duonghq | okay | 06:14 |
| Jeffrey4l | btw, could you review https://review.openstack.org/521376 | 06:14 |
| Jeffrey4l | duonghq, ^^ | 06:14 |
| duonghq | Jeffrey4l, done | 06:15 |
| Jeffrey4l | thanks. | 06:15 |
| duonghq | Jeffrey4l, do we have network isolation for external ceph? | 06:16 |
| Jeffrey4l | duonghq, what that mean? | 06:16 |
| Jeffrey4l | when using external ceph, the only requirement is ceph publish network is accessible by nova-compute/glance-api | 06:16 |
| duonghq | Jeffrey4l, I mean some kind of storage interface configuration for access ceph public network | 06:17 |
| *** cliles has quit IRC | 06:17 | |
| Jeffrey4l | no. refer my previous sentence. | 06:18 |
| duonghq | okay, just want a little more detail on this | 06:18 |
| duonghq | thanks | 06:19 |
| Jeffrey4l | np. ;D | 06:19 |
| openstackgerrit | Tone Zhang proposed openstack/kolla-ansible master: Doc: update swift-guide in kolla-ansible https://review.openstack.org/520996 | 06:21 |
| *** janki is now known as janki|afk | 06:22 | |
| openstackgerrit | Tone Zhang proposed openstack/kolla-ansible master: Doc: update swift-guide in kolla-ansible https://review.openstack.org/520996 | 06:23 |
| *** lpetrut has joined #openstack-kolla | 06:27 | |
| *** jbadiapa has joined #openstack-kolla | 06:35 | |
| openstackgerrit | Chason Chan proposed openstack/kolla-ansible stable/pike: Fix the deprecated syntax in 01-rewrite.conf https://review.openstack.org/521433 | 06:37 |
| *** salv-orlando has quit IRC | 06:37 | |
| *** salv-orlando has joined #openstack-kolla | 06:38 | |
| masber | for external-ceph configuration, do I need to leave [storage] as empty in the inventory file? | 06:58 |
| Jeffrey4l | masber, i think yes. | 07:01 |
| *** cah_link has joined #openstack-kolla | 07:02 | |
| Jeffrey4l | spsurya, still around? please review https://review.openstack.org/521376 | 07:04 |
| *** cah_link has quit IRC | 07:05 | |
| *** cah_link has joined #openstack-kolla | 07:05 | |
| spsurya | Jeffrey4l: done | 07:05 |
| Jeffrey4l | thanks. | 07:06 |
| *** xinliang has quit IRC | 07:06 | |
| *** lpetrut has quit IRC | 07:08 | |
| *** lpetrut has joined #openstack-kolla | 07:08 | |
| openstackgerrit | Merged openstack/kolla-ansible master: Doc: update swift-guide in kolla-ansible https://review.openstack.org/520996 | 07:09 |
| *** lpetrut has quit IRC | 07:14 | |
| *** itlinux has quit IRC | 07:15 | |
| *** egonzalez has joined #openstack-kolla | 07:18 | |
| *** xinliang has joined #openstack-kolla | 07:18 | |
| *** lpetrut has joined #openstack-kolla | 07:22 | |
| *** janki|afk is now known as janki | 07:35 | |
| *** manheim has joined #openstack-kolla | 07:43 | |
| *** caowei has joined #openstack-kolla | 07:44 | |
| *** shardy has joined #openstack-kolla | 07:46 | |
| *** manheim has quit IRC | 07:50 | |
| *** manheim has joined #openstack-kolla | 07:51 | |
| *** dardelean has joined #openstack-kolla | 07:52 | |
| *** lpetrut has quit IRC | 07:53 | |
| *** manheim has quit IRC | 07:56 | |
| *** pcaruana has joined #openstack-kolla | 08:04 | |
| *** gfidente has joined #openstack-kolla | 08:15 | |
| openstackgerrit | Jeffrey Zhang proposed openstack/kolla master: Add squid cache proxy to speed up build https://review.openstack.org/521450 | 08:17 |
| *** dciabrin has joined #openstack-kolla | 08:19 | |
| *** zhangfei has quit IRC | 08:21 | |
| *** gkadam has joined #openstack-kolla | 08:26 | |
| openstackgerrit | Martin Mágr proposed openstack/kolla master: Clean fluentd_plugins_install block https://review.openstack.org/512598 | 08:26 |
| *** magicboiz has joined #openstack-kolla | 08:27 | |
| openstackgerrit | Jeffrey Zhang proposed openstack/kolla master: Add squid cache proxy to speed up build https://review.openstack.org/521450 | 08:33 |
| *** zhangfei has joined #openstack-kolla | 08:34 | |
| *** manheim has joined #openstack-kolla | 08:36 | |
| *** amoralej|off is now known as amoralej | 08:38 | |
| *** goldyfruit has joined #openstack-kolla | 08:41 | |
| *** ktibi has joined #openstack-kolla | 08:42 | |
| *** salv-orlando has quit IRC | 08:42 | |
| ktibi | morning all EU ;) | 08:42 |
| *** salv-orlando has joined #openstack-kolla | 08:43 | |
| *** manheim has quit IRC | 08:44 | |
| *** salv-orlando has quit IRC | 08:47 | |
| *** numans has joined #openstack-kolla | 08:49 | |
| *** rmart04 has joined #openstack-kolla | 08:53 | |
| openstackgerrit | Jeffrey Zhang proposed openstack/kolla master: Add squid cache proxy to speed up build https://review.openstack.org/521450 | 08:54 |
| *** salv-orlando has joined #openstack-kolla | 08:58 | |
| *** owalsh_ is now known as owalsh | 08:58 | |
| *** goldyfruit has quit IRC | 08:58 | |
| *** numans has quit IRC | 09:08 | |
| *** athomas has joined #openstack-kolla | 09:08 | |
| *** numans has joined #openstack-kolla | 09:10 | |
| *** manheim has joined #openstack-kolla | 09:11 | |
| *** goldyfruit has joined #openstack-kolla | 09:11 | |
| *** manheim has quit IRC | 09:15 | |
| openstackgerrit | Jeffrey Zhang proposed openstack/kolla master: Add squid cache proxy to speed up build https://review.openstack.org/521450 | 09:19 |
| *** mgoddard has joined #openstack-kolla | 09:20 | |
| bjolo | morning | 09:21 |
| bjolo | SamYaple: about injecting startup scripts... i (and a bunch of my colleagues) agree with you :) | 09:22 |
| openstackgerrit | Pierre Hanselmann proposed openstack/kolla-ansible master: Add horizon domain dropdown https://review.openstack.org/514676 | 09:23 |
| bjolo | duonghq: we use external ceph and it works just fine | 09:24 |
| *** magicboiz has quit IRC | 09:25 | |
| *** magicboiz has joined #openstack-kolla | 09:27 | |
| *** egonzalez has quit IRC | 09:27 | |
| *** egonzalez has joined #openstack-kolla | 09:28 | |
| *** jascott1 has joined #openstack-kolla | 09:29 | |
| *** magicboiz has quit IRC | 09:32 | |
| *** goldyfruit has quit IRC | 09:33 | |
| *** jascott1 has quit IRC | 09:33 | |
| duonghq | bjolo, thank you for your information | 09:35 |
| *** yingjun has quit IRC | 09:35 | |
| *** magicboiz has joined #openstack-kolla | 09:39 | |
| lxdong | SamYaple ,can three HCI nodes achieve the first level of HA? | 09:39 |
| lxdong | masber ,what's your opinion? | 09:41 |
| *** goldyfruit has joined #openstack-kolla | 09:42 | |
| goutham | Hi all nova console-log and get-vnc-console log commands doesnt work when triggered continuosly | 09:48 |
| goutham | https://hastebin.com/pixehakiko.rb | 09:49 |
| goutham | https://hastebin.com/omifirupuj.rb but nova list shows the vm launched | 09:49 |
| goutham | is this a *BUG* or is there any issue with my deployment?? | 09:50 |
| *** lpetrut has joined #openstack-kolla | 09:50 | |
| *** lpetrut has quit IRC | 09:56 | |
| *** tovin07_ has quit IRC | 10:08 | |
| *** lpetrut has joined #openstack-kolla | 10:13 | |
| *** duonghq has quit IRC | 10:15 | |
| lxdong | why don't you change your controller to the third HCI nodes? masber | 10:16 |
| goutham | the vm goes to error state when i reboot my vm | 10:17 |
| *** caowei has quit IRC | 10:17 | |
| *** caowei has joined #openstack-kolla | 10:18 | |
| *** magicboiz has quit IRC | 10:18 | |
| openstackgerrit | Kevin Zhao proposed openstack/kolla master: Creates directory /var/run/mysqld on aarch64 when building MariaDB docker image https://review.openstack.org/521479 | 10:21 |
| *** magicboiz has joined #openstack-kolla | 10:22 | |
| openstackgerrit | Kevin Zhao proposed openstack/kolla master: Add python-openstackclient to debian kolla-toolbox https://review.openstack.org/521481 | 10:22 |
| *** jiriprox has joined #openstack-kolla | 10:23 | |
| *** magicboiz has quit IRC | 10:26 | |
| *** jaosorior has joined #openstack-kolla | 10:30 | |
| goutham | https://bugs.launchpad.net/kolla/+bug/1733300 any one please help me with this | 10:32 |
| openstack | Launchpad bug 1733300 in kolla "nova service fails to find my VM regularly " [Undecided,New] | 10:32 |
| *** kiennt26 has quit IRC | 10:32 | |
| *** magicboiz has joined #openstack-kolla | 10:33 | |
| goutham | https://bugs.launchpad.net/kolla/+bug/1733302 help on this as well | 10:37 |
| openstack | Launchpad bug 1733302 in kolla "Vm fails to boot after reboot " [Undecided,New] | 10:37 |
| egonzalez | goutham, marked one as duplicated, both are same issue. nova mark the vm in error because one of the hosts is not finding the vm | 10:52 |
| egonzalez | add more info in the bug report, globals.yml, configs, inventory, logs, docker ps -a. etc | 10:52 |
| goutham | Ok sure thanks egonzalez i will add that | 10:53 |
| openstackgerrit | Merged openstack/kolla stable/ocata: Bump kafka version to 0.10.2.1 https://review.openstack.org/521376 | 10:59 |
| openstackgerrit | Vladislav Belogrudov proposed openstack/kolla-ansible master: Enable nfs driver volume snapshots by default https://review.openstack.org/521487 | 11:03 |
| *** zhangfei has quit IRC | 11:03 | |
| *** goldyfruit has quit IRC | 11:15 | |
| egonzalez | goutham, qemu/nova bug https://bugs.launchpad.net/ubuntu/+source/nova/+bug/1718133 | 11:16 |
| openstack | Launchpad bug 1718133 in nova (Ubuntu Artful) "Live block migration in Pike fails due to qemu-img" [Critical,Fix released] - Assigned to James Page (james-page) | 11:16 |
| *** goldyfruit has joined #openstack-kolla | 11:20 | |
| egonzalez | goutham, binary images will work with kolla 5.0.0, source wont because of the pinned nova version | 11:22 |
| goutham | ok now i have to download binary images right?? | 11:23 |
| goutham | will do it | 11:23 |
| *** caowei has quit IRC | 11:28 | |
| *** openstackgerrit has quit IRC | 11:32 | |
| spsurya | krtaylor: ping | 11:36 |
| *** yingjun has joined #openstack-kolla | 11:46 | |
| *** magicboiz has quit IRC | 11:47 | |
| *** magicboiz has joined #openstack-kolla | 11:48 | |
| *** gfidente^2nd has joined #openstack-kolla | 11:53 | |
| *** magicboiz has quit IRC | 11:55 | |
| *** magicboiz has joined #openstack-kolla | 11:55 | |
| *** gfidente has quit IRC | 11:56 | |
| *** dciabrin_ has joined #openstack-kolla | 12:01 | |
| *** dciabrin has quit IRC | 12:03 | |
| *** manheim has joined #openstack-kolla | 12:04 | |
| *** manheim has quit IRC | 12:08 | |
| *** jmccarthy has joined #openstack-kolla | 12:10 | |
| lxdong | I am using three HCI nodes ,ceph cluster worked with a downgrade stage when taking down one node while instance on the node can't be access any more. masber ,SamYaple ,Jeffrey4l , I need to achieve instance HA. | 12:13 |
| *** pbourke has quit IRC | 12:14 | |
| *** pbourke has joined #openstack-kolla | 12:16 | |
| *** unicell has quit IRC | 12:23 | |
| *** janki has quit IRC | 12:29 | |
| lxdong | I tried console ,snapshot ,volume backup ,migration ,no way to recover the instance. | 12:31 |
| lxdong | there are two replications, how to find and make use of another? | 12:33 |
| *** Pavo has joined #openstack-kolla | 12:33 | |
| *** Pavo has quit IRC | 12:36 | |
| *** ansmith has quit IRC | 12:36 | |
| *** rhallisey has joined #openstack-kolla | 12:57 | |
| *** salv-orlando has quit IRC | 12:59 | |
| *** salv-orlando has joined #openstack-kolla | 13:00 | |
| *** salv-orlando has quit IRC | 13:04 | |
| *** goldyfruit has quit IRC | 13:12 | |
| *** goldyfruit has joined #openstack-kolla | 13:14 | |
| *** coolsvap has quit IRC | 13:20 | |
| *** goldyfruit has quit IRC | 13:26 | |
| sdake | morning | 13:31 |
| *** ansmith has joined #openstack-kolla | 13:34 | |
| *** dave-mccowan has joined #openstack-kolla | 13:36 | |
| *** skramaja has quit IRC | 13:38 | |
| *** goldyfruit has joined #openstack-kolla | 13:43 | |
| *** dave-mcc_ has joined #openstack-kolla | 13:46 | |
| *** dave-mccowan has quit IRC | 13:48 | |
| rwellum | Morning sdake | 13:49 |
| *** jtriley has quit IRC | 13:57 | |
| *** openstackgerrit has joined #openstack-kolla | 13:58 | |
| openstackgerrit | Jeffrey Zhang proposed openstack/kolla master: Bump kafka version to 0.11.0.2 https://review.openstack.org/521372 | 13:58 |
| *** yangyapeng has quit IRC | 13:59 | |
| *** salv-orlando has joined #openstack-kolla | 13:59 | |
| *** milan_ has joined #openstack-kolla | 14:00 | |
| goutham | egonzalez: i still get the same error https://hastebin.com/udaqavorug.py BTW i m just booting up VM i think it has nothing to do with the Live migration right?? | 14:01 |
| milan_ | morning folks! :) I'd like to create an ironic-inspector container; where/how do I start best? | 14:01 |
| goutham | I used binary images this time egonzalez .. | 14:04 |
| *** serlex has joined #openstack-kolla | 14:15 | |
| *** serlex has left #openstack-kolla | 14:15 | |
| *** jmccarthy has quit IRC | 14:21 | |
| *** goldyfruit has quit IRC | 14:22 | |
| *** rhallisey has quit IRC | 14:22 | |
| *** jmccarthy has joined #openstack-kolla | 14:24 | |
| *** yangyapeng has joined #openstack-kolla | 14:24 | |
| *** rhallisey has joined #openstack-kolla | 14:25 | |
| egonzalez | goutham, ubuntu may not packaged the fix yet. dont know | 14:28 |
| *** yangyapeng has quit IRC | 14:29 | |
| *** benjamin_diaz has joined #openstack-kolla | 14:30 | |
| *** jmccarthy1 has joined #openstack-kolla | 14:31 | |
| *** pbourke has quit IRC | 14:32 | |
| *** jtriley has joined #openstack-kolla | 14:33 | |
| *** jmccarthy has quit IRC | 14:35 | |
| openstackgerrit | Doug Szumski proposed openstack/kolla-ansible master: Add missing hostnames to fluentd log messages https://review.openstack.org/514593 | 14:36 |
| *** yangyapeng has joined #openstack-kolla | 14:42 | |
| *** mdnadeem has quit IRC | 14:44 | |
| *** hachi_ has joined #openstack-kolla | 14:45 | |
| *** yangyapeng has quit IRC | 14:46 | |
| *** pbourke has joined #openstack-kolla | 14:47 | |
| *** dougsz has joined #openstack-kolla | 15:02 | |
| *** yangyapeng has joined #openstack-kolla | 15:04 | |
| *** dave-mcc_ is now known as dave-mccowan | 15:06 | |
| *** yangyapeng has quit IRC | 15:08 | |
| *** zshi has quit IRC | 15:10 | |
| *** hongbin has joined #openstack-kolla | 15:11 | |
| benjamin_diaz | Not kolla especific, but has anyone runned the Tempest tests required por the Opnestack Powered Platform? I am only failing 2 tests, the objectstore-container-quotas ones. | 15:12 |
| *** goldyfruit has joined #openstack-kolla | 15:12 | |
| krtaylor | spsurya, pong | 15:15 |
| *** yangyapeng has joined #openstack-kolla | 15:24 | |
| openstackgerrit | James McCarthy proposed openstack/kolla-ansible master: WIP: Update swift templates for rsync_module https://review.openstack.org/521133 | 15:25 |
| openstackgerrit | Jeffrey Zhang proposed openstack/kolla master: Add squid cache proxy to speed up build https://review.openstack.org/521450 | 15:26 |
| *** rhallisey has quit IRC | 15:28 | |
| *** yangyapeng has quit IRC | 15:28 | |
| *** rhallisey has joined #openstack-kolla | 15:29 | |
| *** ansmith has quit IRC | 15:30 | |
| *** hachi_ has quit IRC | 15:39 | |
| *** yingjun has quit IRC | 15:40 | |
| *** aagate has quit IRC | 15:46 | |
| *** jiriprox has quit IRC | 15:55 | |
| *** itlinux has joined #openstack-kolla | 15:59 | |
| *** egonzalez has quit IRC | 16:00 | |
| *** jtriley has quit IRC | 16:02 | |
| *** lrensing has joined #openstack-kolla | 16:02 | |
| *** jamesbenson has joined #openstack-kolla | 16:03 | |
| *** jtriley has joined #openstack-kolla | 16:04 | |
| *** gfidente^2nd is now known as gfidente | 16:08 | |
| *** rmart04 has quit IRC | 16:08 | |
| *** gfidente has quit IRC | 16:09 | |
| *** gfidente has joined #openstack-kolla | 16:09 | |
| *** lrensing has quit IRC | 16:21 | |
| *** lrensing has joined #openstack-kolla | 16:34 | |
| *** pcaruana has quit IRC | 16:35 | |
| *** salv-orlando has quit IRC | 16:38 | |
| *** salv-orlando has joined #openstack-kolla | 16:39 | |
| *** lebauce_ has joined #openstack-kolla | 16:43 | |
| *** salv-orlando has quit IRC | 16:43 | |
| *** jascott1 has joined #openstack-kolla | 16:44 | |
| *** jascott1 has quit IRC | 16:48 | |
| *** jtriley has quit IRC | 16:50 | |
| *** ntpttr_laptop has joined #openstack-kolla | 16:51 | |
| *** itlinux has quit IRC | 16:53 | |
| *** itlinux has joined #openstack-kolla | 16:54 | |
| *** jascott1 has joined #openstack-kolla | 17:00 | |
| *** jtriley has joined #openstack-kolla | 17:09 | |
| *** jamesbenson has quit IRC | 17:10 | |
| *** jamesbenson has joined #openstack-kolla | 17:10 | |
| *** cah_link has quit IRC | 17:11 | |
| jamesbenson | inc0 Ping | 17:15 |
| *** gkadam has quit IRC | 17:17 | |
| inc0 | good mornign | 17:18 |
| inc0 | just woke up, what's up James | 17:18 |
| *** dciabrin__ has joined #openstack-kolla | 17:22 | |
| *** lpetrut has quit IRC | 17:23 | |
| *** dciabrin_ has quit IRC | 17:24 | |
| inc0 | pbourke mandre egonzalez - https://review.openstack.org/#/c/519939/ can you review whenever get chance? this fixes gates... | 17:25 |
| kfox1111 | morning. | 17:25 |
| *** jamesbenson has quit IRC | 17:27 | |
| *** dougsz has quit IRC | 17:28 | |
| *** goldyfruit has quit IRC | 17:30 | |
| openstackgerrit | Merged openstack/kolla-kubernetes master: Specify unit (Gi) in common pvc https://review.openstack.org/519461 | 17:31 |
| jascott1 | yay | 17:32 |
| *** lrensing has quit IRC | 17:34 | |
| *** mgoddard has quit IRC | 17:38 | |
| *** mgoddard has joined #openstack-kolla | 17:38 | |
| *** salv-orlando has joined #openstack-kolla | 17:39 | |
| rwellum | Jeffrey4l: Is lokolla the correct repo? | 17:40 |
| sdake | Jeffrey4l around? | 17:40 |
| *** lrensing has joined #openstack-kolla | 17:40 | |
| sdake | rwellum yes it is | 17:40 |
| sdake | anyone know how to forward a connection from 127.0.0.1 to the machine's main IP address? | 17:41 |
| rwellum | I know sbezverk used this for the docker registry - didn't know it was official | 17:41 |
| sdake | rwellum its not official - its part of the build process iirc | 17:41 |
| *** ntpttr_laptop has quit IRC | 17:42 | |
| *** ntpttr_laptop__ has joined #openstack-kolla | 17:42 | |
| *** yangyapeng has joined #openstack-kolla | 17:43 | |
| *** salv-orlando has quit IRC | 17:45 | |
| *** rwsu has joined #openstack-kolla | 17:45 | |
| jascott1 | docker registry uses nodePort | 17:49 |
| *** yangyapeng has quit IRC | 17:50 | |
| kfox1111 | sdake: what do you mean? | 17:50 |
| kfox1111 | I mean, by forwarding from 127.0.0.1 | 17:50 |
| sdake | kfox1111 using kubectl port-forward | 17:50 |
| sdake | kfox1111 that only binds to 127.0.0.1 | 17:50 |
| kfox1111 | oh. you want to expose a port-forward to a different machine? | 17:51 |
| sdake | same machine | 17:51 |
| sdake | I just need to be able to access it by ip | 17:51 |
| kfox1111 | 127.0.0.1 is available on the machine. | 17:51 |
| sdake | 127.0.0.1 is not accesible from internets | 17:51 |
| sdake | ya sorry form a different machine | 17:51 |
| kfox1111 | not sure the easiest way.... socat might work in a pinch. | 17:51 |
| sdake | does that go both ways? | 17:52 |
| kfox1111 | I think so. | 17:52 |
| sdake | i'll check it out | 17:52 |
| *** ntpttr_laptop__ has quit IRC | 17:52 | |
| kfox1111 | its a tool in the networking swiss army knife. :) | 17:52 |
| *** mgoddard has quit IRC | 17:53 | |
| sdake | https://www.cyberciti.biz/faq/linux-unix-tcp-port-forwarding/ | 17:54 |
| sdake | unfortunately it looks like greek to me ;) | 17:54 |
| kfox1111 | time to read up on more networking. ;) | 17:54 |
| kfox1111 | the first example looks about right: | 17:55 |
| kfox1111 | socat TCP-LISTEN:80,fork TCP:202.54.1.5:80 | 17:55 |
| kfox1111 | might be able to do: socat TCP-LISTEN:3000,fork TCP:127.0.0.1:<proxieslocalport> | 17:55 |
| * kfox1111 just spent the weekend boning up on udp and the dhcp rfc's. | 17:57 | |
| sdake | sudo socat TCP-LISTEN:9090,reuseaddr,fork TCP:10.128.0.2:9090 | 17:57 |
| sdake | 2017/11/20 17:56:37 socat[3907] E bind(5, {AF=2 0.0.0.0:9090}, 16): Address already in use | 17:57 |
| sdake | i tred following | 17:57 |
| sdake | and get an addrinuse | 17:57 |
| kfox1111 | cant use the same port l | 17:58 |
| sdake | maybe i've got it | 17:58 |
| *** pcaruana has joined #openstack-kolla | 18:00 | |
| sdake | kfox1111 does this look correct: sudo socat TCP-LISTEN:127.0.0.9090,reuseaddr,fork TCP:10.128.0.2:9000 | 18:01 |
| sdake | 2017/11/20 18:01:25 socat[3943] E bind(5, {AF=2 0.0.0.0:127}, 16): Address already in use | 18:01 |
| kfox1111 | seems backwards | 18:02 |
| sdake | sudo socat TCP-LISTEN:127.0.0.1:9090 TCP:10.128.0.2:9001 | 18:03 |
| sdake | 2017/11/20 18:03:13 socat[3958] E TCP-LISTEN: wrong number of parameters (2 instead of 1) | 18:03 |
| sdake | so swap parameters? | 18:03 |
| sdake | oh right- listen | 18:03 |
| sdake | got it | 18:03 |
| sdake | sudo socat TCP:127.0.0.1:9090 TCP-LISTEN:10.128.0.2:9000 | 18:07 |
| sdake | 2017/11/20 18:05:01 socat[3991] E TCP-LISTEN: wrong number of parameters (2 instead of 1) | 18:07 |
| *** dciabrin__ has quit IRC | 18:07 | |
| sdake | curl 10.128.0.2:9000 | 18:07 |
| sdake | curl: (7) Failed to connect to 10.128.0.2 port 9000: Connection refused | 18:07 |
| sdake | seems like no worky | 18:07 |
| *** dciabrin__ has joined #openstack-kolla | 18:07 | |
| sdake | not sure what that wrong number of parameters are | 18:07 |
| kfox1111 | sudo socat TCP:127.0.0.1:9090 TCP-LISTEN:9000 | 18:07 |
| sdake | coool that works | 18:09 |
| sdake | kfox1111 = brillaint :) | 18:09 |
| kfox1111 | cool. | 18:09 |
| *** manheim has joined #openstack-kolla | 18:12 | |
| *** yangyapeng has joined #openstack-kolla | 18:12 | |
| *** yangyapeng has quit IRC | 18:17 | |
| openstackgerrit | James McCarthy proposed openstack/kolla-ansible master: Update swift templates for rsync_module https://review.openstack.org/521133 | 18:17 |
| *** manheim has quit IRC | 18:20 | |
| *** hachi_ has joined #openstack-kolla | 18:26 | |
| *** jamesbenson has joined #openstack-kolla | 18:27 | |
| jamesbenson | hey inc0, sorry multi-tasking this morning.... I had an issue but was able to figure it out.. it was related to the volume/instance creation .... apparently something i thought was a bug isn't upon finding some bug reports about the same thing. | 18:28 |
| inc0 | kfox1111: what do you use to deploy initial bare metal? | 18:28 |
| kfox1111 | inc0: cobbler. but am working on a replacement. | 18:29 |
| jamesbenson | inc0, we use maas and like it here... | 18:29 |
| inc0 | I'm playing with idea around something that would be effectively lightweight ironic | 18:29 |
| jamesbenson | for bare metal | 18:29 |
| *** lpetrut has joined #openstack-kolla | 18:29 | |
| inc0 | ipxe+mock metadata server | 18:29 |
| inc0 | yeah I looked at it too | 18:30 |
| *** yangyapeng has joined #openstack-kolla | 18:31 | |
| *** milan_ has quit IRC | 18:31 | |
| inc0 | so my goal is to get something that's extremally lightweight to provision seed node quickly and without big hassle | 18:31 |
| inc0 | and everything, including maas, seems to be this giant monolith of a software which is painful in many ways | 18:31 |
| inc0 | for cobbler - I hate kickstart scripts | 18:31 |
| inc0 | there is coreos matchbox, that's for terraform, but I don't see any non-coreos deployment there | 18:32 |
| inc0 | and coreos is wierd with it's ignition thingy | 18:33 |
| inc0 | ironic/bifrost and maas are big and much more than I really need | 18:34 |
| inc0 | you'd think it would be solved problem by now right?:) | 18:34 |
| *** yangyapeng has quit IRC | 18:34 | |
| *** vhosakot has joined #openstack-kolla | 18:36 | |
| *** dciabrin__ has quit IRC | 18:37 | |
| *** dciabrin has joined #openstack-kolla | 18:37 | |
| *** salv-orlando has joined #openstack-kolla | 18:40 | |
| *** salv-orlando has quit IRC | 18:42 | |
| *** salv-orlando has joined #openstack-kolla | 18:42 | |
| *** lpetrut has quit IRC | 18:51 | |
| *** lpetrut has joined #openstack-kolla | 18:55 | |
| jamesbenson | I agree inc0 | 18:55 |
| jamesbenson | if it helps at all, we have our maas sitting in a VM on our control plane | 18:56 |
| jamesbenson | but it isn't without it's faults either... | 18:56 |
| jamesbenson | you can really only deploy ubuntu nicely.. theoretically you can deploy cent, but has its own issues... and somehow, no docs mentioned anywhere you can deploy windows server, but they say it once, then nothing. | 18:57 |
| *** yangyapeng has joined #openstack-kolla | 18:57 | |
| jamesbenson | personally, I really liked how fuel deployed bare metal then deployed Openstack over it... wish kolla would do something similiar. | 18:57 |
| *** lpetrut has quit IRC | 19:01 | |
| *** yangyapeng has quit IRC | 19:01 | |
| *** lpetrut has joined #openstack-kolla | 19:01 | |
| kfox1111 | inc0: I'm playing with a provisioning system too. very light weight. | 19:01 |
| kfox1111 | basically using k8s to deploy k8s. | 19:01 |
| kfox1111 | or minikube to deploy k8s. | 19:01 |
| kfox1111 | see: https://hub.docker.com/r/kfox1111/provision-baseos-server/builds/ for the centos mirror part. | 19:02 |
| kfox1111 | self contained container that you can kickstart centos machines from. | 19:03 |
| kfox1111 | I'm working on some helm charts to deploy all the pieces so you can set it all up with one helm install. | 19:03 |
| *** itlinux has quit IRC | 19:03 | |
| *** manheim has joined #openstack-kolla | 19:05 | |
| *** lrensing has quit IRC | 19:05 | |
| jascott1 | kfox1111 have you ever seen the problem where you cannot exec into any pods that arent on master? | 19:12 |
| kfox1111 | yeah. | 19:12 |
| jascott1 | whatup with dat? | 19:12 |
| kfox1111 | tipically is a firewall issue on the compute node blocking access from the master. | 19:12 |
| kfox1111 | or a network connectivity issue. | 19:12 |
| jascott1 | yeah im reading about kubelet config but everything seems ok | 19:13 |
| jascott1 | it seems like everything is working except that | 19:13 |
| kfox1111 | for exec and logs, the master makes a http connection to the kubelet. | 19:13 |
| kfox1111 | which is backwards from all the other stuff. | 19:13 |
| kfox1111 | usually its the kubelet http accessing the apiserver. | 19:13 |
| jascott1 | i cant find where kubelet logs a request auth failure | 19:14 |
| kfox1111 | probably never saw the request? | 19:14 |
| kfox1111 | are you using centos? | 19:14 |
| *** yangyapeng has joined #openstack-kolla | 19:15 | |
| jascott1 | ubuntu | 19:15 |
| kfox1111 | not as familior. I think they are still using ufm? | 19:15 |
| jascott1 | so far googling suggests that it is the auth from api server to kubelet like you said but I seem to have all the flags for TLS client certs | 19:16 |
| kfox1111 | might need to disable it temporarily to validate. | 19:16 |
| jascott1 | ok thanks I do need to confirm that | 19:16 |
| kfox1111 | if you arn't seeing any logs its probably never seeing the packets. | 19:16 |
| *** yangyapeng has quit IRC | 19:20 | |
| *** rwsu has quit IRC | 19:30 | |
| *** itlinux has joined #openstack-kolla | 19:31 | |
| inc0 | kfox1111: what's your idea behind it? you're using pure dnsmasq+pxe or sth more? | 19:32 |
| inc0 | jascott1: I also looked at matchbox code a bit and while ipxe+dnsmasq thing is useful there, it's all the coreos on top of it that I don't want | 19:35 |
| jascott1 | right yeah thats unfortunate | 19:35 |
| inc0 | my idea goes like this: install first node, docker-compose containers with dnsmasq+pxe, and another with http server for 2 things | 19:36 |
| inc0 | 1. serve image contents for ipxe install | 19:36 |
| inc0 | 2. serve mock metadata service for cloud images | 19:36 |
| *** yangyapeng has joined #openstack-kolla | 19:36 | |
| inc0 | maybe spread ^ into 2 containers, dunno yet | 19:37 |
| inc0 | https://www.youtube.com/watch?v=gNqQL-1gZF8 | 19:37 |
| *** yangyapeng has quit IRC | 19:41 | |
| kfox1111 | inc0: I've seen 3 different ways to do this: | 19:42 |
| kfox1111 | 1. all in one systems. like cobbler, spacewalk, etc. very not cloud native. | 19:42 |
| kfox1111 | 2. single binary solutions. like digital-rebar-provision. compiles it all to a single go binary. | 19:43 |
| kfox1111 | 3. microservices based. only one I really know of is kind of ironic. but very heavy and still rather monolithic like #1. | 19:43 |
| inc0 | right | 19:44 |
| kfox1111 | I'm trying to build something very cloud native. | 19:44 |
| inc0 | well | 19:44 |
| inc0 | me too in a way, but I was thinking of aio use case | 19:44 |
| kfox1111 | dhcp conttainer, tftp container, http rpm server, rpm servers per yum repo, kickstart server, etc. | 19:44 |
| inc0 | however way I envision it will be fairly cloud native-y | 19:44 |
| kfox1111 | aio can be done via helm. :) | 19:44 |
| inc0 | dnsmasq can do both tftp and dhcp | 19:45 |
| kfox1111 | so no reason to do aio and big cluster be different. | 19:45 |
| kfox1111 | yeah. but still mixes use cases. | 19:45 |
| inc0 | sure | 19:45 |
| kfox1111 | no reason not to do tftp and dhcp seperate when you can easily do so. | 19:45 |
| inc0 | well, however to break this up, we're thinking of same thing effectively | 19:45 |
| kfox1111 | I'm thinking pxiecore too. | 19:45 |
| kfox1111 | yeah. | 19:45 |
| kfox1111 | i have a prototype dhcp server that serves entries out of kube crd's too. :) | 19:46 |
| *** absubram has joined #openstack-kolla | 19:46 | |
| kfox1111 | so, totally stateless. :) | 19:46 |
| inc0 | so I wanted to do bare metal first, k8s later | 19:46 |
| inc0 | separate these two | 19:46 |
| kfox1111 | thats where you get forced into #2. | 19:46 |
| kfox1111 | I want to not worry about day 2, by archetecting it as microservices. | 19:47 |
| kfox1111 | then day 1 is just bundling it all into one vm. | 19:47 |
| kfox1111 | k8s is great at microservices, so we just reuse that. | 19:47 |
| inc0 | well what I'm saying is to be able to stop when you just install say ubuntu | 19:47 |
| kfox1111 | thats a kickstart thingy. | 19:48 |
| inc0 | rigth | 19:48 |
| kfox1111 | just dont kubeadm join on it. | 19:48 |
| inc0 | or lack of thereof | 19:48 |
| kfox1111 | right. | 19:48 |
| kfox1111 | (preseed in deb terms) | 19:48 |
| kfox1111 | still all the same plumbing though. | 19:48 |
| inc0 | so that can be done with metadata server | 19:48 |
| inc0 | and cloud images | 19:48 |
| kfox1111 | yeah, some server to do templating. | 19:49 |
| kfox1111 | no on cloud images. | 19:49 |
| inc0 | I think it's much more elegant than kickstarts/preseed | 19:49 |
| inc0 | and not distro specific | 19:49 |
| kfox1111 | I wana provision natively. much much more flexible. | 19:49 |
| inc0 | well, image with cloud-init baked in | 19:49 |
| kfox1111 | the big problem is, cloud images are built with the assumption of vm's. | 19:49 |
| kfox1111 | vms are nice and pretty hardware wise. | 19:49 |
| kfox1111 | bare metal is fugly. | 19:49 |
| inc0 | yueah I know there are issues | 19:49 |
| inc0 | we've seen them | 19:49 |
| kfox1111 | very few machines work out of the box. | 19:49 |
| inc0 | that being said, ironic will need this kind of thing | 19:50 |
| kfox1111 | this is why coreos isn't so prtty a bare metal solution. | 19:50 |
| kfox1111 | yeah, but tahts ironic's problem. | 19:50 |
| *** amoralej is now known as amoralej|off | 19:50 | |
| inc0 | or distro problems | 19:50 |
| kfox1111 | I want something very simple/light weight. | 19:50 |
| inc0 | it's just adding one more pkg to distro | 19:50 |
| kfox1111 | its just a 'cloud-init' in a kickstart file too. | 19:51 |
| kfox1111 | image building is much harder. | 19:51 |
| inc0 | I know, but what I'm saying is kickstart is distro specific | 19:51 |
| kfox1111 | kind of. | 19:51 |
| inc0 | cloud init is not | 19:51 |
| kfox1111 | oh. our saying, template out most of the install stuff in cloud init instead of kickstart but stil have a kickstart to provision initially? | 19:52 |
| kfox1111 | (btw, cloud init's prettty crappy for cross platform. :( ) | 19:52 |
| kfox1111 | every distro configures it differently. :( | 19:52 |
| kfox1111 | I'd rather kickstart it and run something like ansible standalone. | 19:52 |
| kfox1111 | much more portable. | 19:53 |
| inc0 | right, all I need is pubkey on machine | 19:53 |
| kfox1111 | can drop that on with the kickstart/preseed. | 19:53 |
| *** jmccarthy1 has left #openstack-kolla | 19:54 | |
| inc0 | right, that's what I was thinking of using cloud-init for | 19:55 |
| kfox1111 | you gotta kickstart/preseed anyway. cloud init just adds complication. | 19:55 |
| kfox1111 | as it doesn't really exist on bare metal. | 19:55 |
| inc0 | why tho? | 19:56 |
| kfox1111 | you have to ship a cloud init metadata file somehow just as much as pushing a key file. | 19:56 |
| inc0 | I mean why you need kickstart anyway | 19:56 |
| kfox1111 | how does cloud-init find data? | 19:56 |
| kfox1111 | oh. | 19:56 |
| kfox1111 | because I need to do stuff like: | 19:56 |
| inc0 | 169.254.169.254 | 19:56 |
| kfox1111 | 1. choose my partition layouts. | 19:56 |
| kfox1111 | 2. choose network drivers. | 19:56 |
| kfox1111 | 169.254.169.254 doesn't exist with bare meta. | 19:56 |
| kfox1111 | unless you are running ironic. | 19:56 |
| kfox1111 | I'm trying to avoid it. | 19:56 |
| *** yangyapeng has joined #openstack-kolla | 19:57 | |
| inc0 | or mock it on gateway in the way neutron does | 19:57 |
| inc0 | it's one iptables rule and http server running | 19:57 |
| kfox1111 | still need kickstart to customize partition layout, nics, storage controllers, etc. | 19:57 |
| inc0 | yeah, true to that | 19:57 |
| kfox1111 | so already gotta payt that cost | 19:58 |
| kfox1111 | a metadata server is then extra cost. | 19:58 |
| inc0 | right I guess | 19:59 |
| kfox1111 | so one of the few parts I don't have currently is a template engine. | 19:59 |
| inc0 | well, I just want to minimze kickstart/preseed stuff | 19:59 |
| inc0 | as it's ugly | 19:59 |
| kfox1111 | thinking of writing a simple go webserver thta pulls content from crd's. | 19:59 |
| kfox1111 | me too in a way... | 19:59 |
| kfox1111 | I mostly want to use this to provision k8s nodes. | 19:59 |
| kfox1111 | which doesn't need very much. docker does the rest. | 19:59 |
| kfox1111 | so like 3 repo servers. | 20:00 |
| kfox1111 | 1 baseos, 2 security updates, 3 k8s repo. | 20:00 |
| inc0 | right what I'd want is output barebone ansible inv and let ansible do the rest | 20:00 |
| kfox1111 | if all the content were crd's in k8s, it would e easy to inventory to ansible. :) | 20:00 |
| inc0 | anyway, let's work together on it because well, we work on it anyway;) | 20:01 |
| kfox1111 | sure. :) | 20:01 |
| *** yangyapeng has quit IRC | 20:01 | |
| inc0 | do you have any code already? | 20:01 |
| kfox1111 | I was hoping to get a working manually installed via pixie prototype up tomorrow. | 20:02 |
| kfox1111 | got a fair chunk. :) | 20:02 |
| inc0 | mind sharing? | 20:02 |
| kfox1111 | I'll try and post a non working prototype later today? its close though. | 20:03 |
| inc0 | cool thanks | 20:03 |
| inc0 | I was planning on focusing on this stuff this week | 20:03 |
| kfox1111 | the dhcp stuff is crap for sure. want to get aio working before I tackle proper dhcp. got a co-worker working on a much more elegant solution. | 20:04 |
| kfox1111 | cool. how fortunate. :) | 20:04 |
| inc0 | yeah, that's cool coincidence:) | 20:04 |
| kfox1111 | I spent the last 2 weeks debugging dhcp udp flows, figuring out dhcp relaying, etc. | 20:04 |
| kfox1111 | for this kind of thing. :) | 20:04 |
| inc0 | I promised our demo team that I'll help them figure out quick bare metal provisioning for all the demos they do | 20:04 |
| kfox1111 | cool. | 20:05 |
| kfox1111 | if this is done with crd's, you could even put your hardware inventory in as a helm chart. | 20:05 |
| kfox1111 | burn your minikube down, and helm install the provisioner all the way back up. :) | 20:05 |
| inc0 | crd? | 20:05 |
| kfox1111 | custom resource definition. | 20:05 |
| benjamin_diaz | If someone has time, please take a look at https://review.openstack.org/#/c/520761/ | 20:05 |
| inc0 | ah ok | 20:06 |
| kfox1111 | used to be called 3rd party resource. | 20:06 |
| kfox1111 | so, you have kind=mythingy | 20:06 |
| kfox1111 | :) | 20:06 |
| kfox1111 | kind=dhcp-host.... ;) | 20:06 |
| kfox1111 | kind=kickstart ... | 20:06 |
| inc0 | benjamin_diaz: well....it shouldn't destroy images on deploy host anyway | 20:07 |
| kfox1111 | declaratively tell k8s cluster your wishes, and the provisioning system relaizes it. :) | 20:07 |
| inc0 | right, my first goal is quickest setup possible on aio | 20:07 |
| kfox1111 | mine's slightly different, | 20:07 |
| kfox1111 | in that I want to be able to start up a system, then sluff off management of the system to the nodes installed. | 20:08 |
| inc0 | from bunck of nuc's -> working env (k8s or not) in least time | 20:08 |
| kfox1111 | ie, minikube start a provisioner, install first node, move all the crd's/helm charrts to the spawned cluster. now can provision more nodes. | 20:08 |
| inc0 | but however we run it, k8s or docker-compose, components are the saem | 20:08 |
| kfox1111 | trueish. | 20:08 |
| kfox1111 | most of the work is in the pod descriptions though I think. | 20:09 |
| kfox1111 | and crd's. | 20:09 |
| kfox1111 | can't exactly replicate that with docker-compose... | 20:09 |
| inc0 | right, for me that's step 2;) | 20:09 |
| inc0 | well you can go simple with it | 20:09 |
| inc0 | net=host | 20:09 |
| inc0 | and that's pretty much it | 20:09 |
| kfox1111 | I guess what I'm saying is, where do you store your data? I want to not manage that myself so k8s crds are awesome. | 20:09 |
| inc0 | you use containers as glorified packages, kolla-ansible style | 20:09 |
| kfox1111 | with docker-compose you'd have to use files I guess. | 20:10 |
| kfox1111 | might still work. | 20:10 |
| inc0 | right | 20:10 |
| kfox1111 | I'm mostly thinking on relying on k8s service tokens and crds. | 20:10 |
| kfox1111 | ok. | 20:10 |
| benjamin_diaz | inc0 It didn't in ocata, it does in pike. Suffered a lot when I discovered it. There is a bug related, where in the comments, egonzalez argued in favor of mantaining the destruction of images in deployer host. So I thought of leaving that feature, but making it so it is disabled by default. | 20:10 |
| kfox1111 | so, the real missing piece is the template engine. | 20:10 |
| kfox1111 | for consistency with the rest of the stack, was thinking gotl.. | 20:11 |
| kfox1111 | I guess that engine could work either from crds or from files on disk. | 20:11 |
| inc0 | benjamin_diaz: as in, kolla-ansible destroy without --include images destroys stuff? | 20:11 |
| inc0 | if it has include-images, I'd say it should | 20:12 |
| kfox1111 | I guess technically though, we can just use static kickstart files too. | 20:12 |
| inc0 | but that being said, you can just run destroyu on all nodes besides deploy | 20:12 |
| kfox1111 | just way less good. | 20:12 |
| inc0 | as deploy node doesn't have any running containers or volmes anyway | 20:12 |
| kfox1111 | I could prototype it out static, and then we could make it dynamic later. | 20:12 |
| inc0 | well I'm still more in favor of kickstart just laying down pubkey | 20:13 |
| inc0 | and let ansible do the rest | 20:13 |
| kfox1111 | yeah. totally should be doable. | 20:13 |
| benjamin_diaz | Originally, --include-images only destroyed images in target hosts, but not on the machine where you executed the command. Now, because localhost was added to the inventory file, it also destroys the image on the deployment host. | 20:13 |
| kfox1111 | kickstart still needs templating for mac addresses, partition info sharing, etc. | 20:13 |
| kfox1111 | ie, making classes of machiens and just pointing at a partition info template chunk. | 20:14 |
| kfox1111 | so you don't have to duplicate the partition info per host. | 20:14 |
| inc0 | right but you can modify ansible-playbook run | 20:14 |
| kfox1111 | not much more then a glorified include. | 20:14 |
| inc0 | and disable localhost | 20:14 |
| kfox1111 | hmm? | 20:14 |
| kfox1111 | oh. for later things, yeah. | 20:14 |
| inc0 | sorry kfox1111 was talking to benjamin_diaz ;) | 20:14 |
| kfox1111 | ok. :) | 20:14 |
| inc0 | well I can quickly write piece of python for kickstart templating | 20:15 |
| kfox1111 | bbiaf. gotta catch up with someone. | 20:15 |
| kfox1111 | yeah. trying to use gotl though. | 20:15 |
| inc0 | if you're hell bend on it can do that too;) | 20:15 |
| kfox1111 | since users will be seeing much more gotl in my use case. | 20:15 |
| kfox1111 | well, true. we could support both. :) | 20:15 |
| kfox1111 | just lilke the deployment images, since its microservices we c an just run both. :) | 20:16 |
| inc0 | well, whatever, I can do golang as well, python is just faster for me | 20:16 |
| kfox1111 | gota go. something broke. bbiab. | 20:16 |
| inc0 | sure | 20:16 |
| kfox1111 | true. has the benifit of being more cobbler compatible too. | 20:16 |
| kfox1111 | though they are cheetah. which is close to jinja, but a little different. | 20:16 |
| kfox1111 | bbiab | 20:16 |
| *** yangyapeng has joined #openstack-kolla | 20:17 | |
| *** yangyapeng has quit IRC | 20:22 | |
| *** salv-orlando has quit IRC | 20:24 | |
| *** salv-orlando has joined #openstack-kolla | 20:25 | |
| jascott1 | https://opensource.googleblog.com/2017/11/container-diff-for-comparing-container-images.html | 20:26 |
| openstackgerrit | Benjamin Diaz proposed openstack/kolla-ansible master: Added var to disable deploy host image destruction https://review.openstack.org/520761 | 20:27 |
| *** salv-orlando has quit IRC | 20:28 | |
| *** hachi_ has quit IRC | 20:29 | |
| *** lrensing has joined #openstack-kolla | 20:30 | |
| *** ktibi has quit IRC | 20:34 | |
| *** salv-orlando has joined #openstack-kolla | 20:36 | |
| *** pcaruana has quit IRC | 20:39 | |
| *** yangyapeng has joined #openstack-kolla | 20:41 | |
| jamesbenson | inc0: I had some pretty crazy kickstart files in a straight up pxe server... not sure if I read the whole history there, but I used to put my ssh keys in the kickstart file... | 20:43 |
| inc0 | right...;) | 20:44 |
| inc0 | as I said, seems to be unsolved issue still;) | 20:44 |
| *** yangyapeng has quit IRC | 20:45 | |
| kfox1111 | jascott1: cool. I wrote my own recently. :) | 20:49 |
| jascott1 | yeah I thought you might be interested in that link, could be useful ;) | 20:49 |
| kfox1111 | https://github.com/kfox1111/hubbuildtools | 20:49 |
| kfox1111 | check this out: https://github.com/kfox1111/dhcp :) | 20:50 |
| SamYaple | o/ jascott1 | 20:50 |
| SamYaple | sup buddy | 20:50 |
| jascott1 | SamYaple whatsup! | 20:50 |
| *** rwsu has joined #openstack-kolla | 20:54 | |
| *** rwsu has quit IRC | 21:00 | |
| *** lrensing has quit IRC | 21:00 | |
| *** rwsu has joined #openstack-kolla | 21:01 | |
| *** retr0h has quit IRC | 21:02 | |
| *** retr0h has joined #openstack-kolla | 21:02 | |
| *** krtaylor has quit IRC | 21:03 | |
| *** lrensing has joined #openstack-kolla | 21:04 | |
| *** rwsu has quit IRC | 21:05 | |
| *** rwsu has joined #openstack-kolla | 21:05 | |
| *** rwsu has quit IRC | 21:06 | |
| *** jamesbenson has quit IRC | 21:06 | |
| *** rwsu has joined #openstack-kolla | 21:06 | |
| *** gfidente has quit IRC | 21:08 | |
| *** jamesbenson has joined #openstack-kolla | 21:10 | |
| sdake | kfox1111 have a moment? | 21:10 |
| *** rhallisey has quit IRC | 21:15 | |
| *** jamesbenson has quit IRC | 21:16 | |
| *** krtaylor has joined #openstack-kolla | 21:16 | |
| *** benjamin_diaz has quit IRC | 21:16 | |
| *** dave-mccowan has quit IRC | 21:18 | |
| *** jamesbenson has joined #openstack-kolla | 21:19 | |
| *** jamesbenson has quit IRC | 21:21 | |
| *** jamesbenson has joined #openstack-kolla | 21:25 | |
| *** jamesbenson has quit IRC | 21:28 | |
| *** rwsu has quit IRC | 21:29 | |
| rwellum | has anyone seen this? keystone container complaining: "sh: line 3: /usr/bin/openstack: No such file or directory" - Pike generated from source. | 21:32 |
| rwellum | Sorry neutron - not keystone | 21:32 |
| rwellum | jascott1: ^^ ? | 21:33 |
| jascott1 | rwellum idk. i did have an issue with from source horizon container not having symlink for actual dashbaord content but that was a known bug | 21:34 |
| rwellum | Thanks jascott1 - yes this looks like neutron missing an install step | 21:35 |
| inc0 | rwellum: check if you have this change https://review.openstack.org/#/c/519971/ | 21:38 |
| inc0 | dependency mgmt is hard | 21:39 |
| *** mchlumsky has joined #openstack-kolla | 21:41 | |
| rwellum | inc0: ah crud - I was a bit out of date - I'll rety | 21:42 |
| rwellum | retry | 21:42 |
| inc0 | yeah we got hit by this painfully :( | 21:43 |
| rwellum | Definitely missing this - thanks! | 21:46 |
| *** jamesbenson has joined #openstack-kolla | 21:48 | |
| rwellum | inc0: Am I safe to just rebuild neutron? | 21:49 |
| inc0 | rwellum: you want to rebuild kolla-toolbox | 21:49 |
| inc0 | but yeah you should be ok with just this one | 21:49 |
| *** jamesbenson has quit IRC | 21:52 | |
| inc0 | any centos ppl around? | 21:57 |
| *** jamesbenson has joined #openstack-kolla | 21:58 | |
| *** jamesbenson has quit IRC | 21:58 | |
| *** jamesbenson has joined #openstack-kolla | 21:59 | |
| *** itlinux has quit IRC | 22:01 | |
| *** itlinux has joined #openstack-kolla | 22:04 | |
| *** shardy has quit IRC | 22:06 | |
| *** jmccarthy has joined #openstack-kolla | 22:07 | |
| *** rwsu has joined #openstack-kolla | 22:10 | |
| *** yangyapeng has joined #openstack-kolla | 22:23 | |
| *** mikal has joined #openstack-kolla | 22:25 | |
| mikal | I'm following the kolla setup instructions, and canal-2tckp is in a crashloop. Is that something I should talk to you about here, or should I talk to the canal people instead? | 22:26 |
| mikal | Sorry, kolla-kubernetes that is | 22:26 |
| *** yangyapeng has quit IRC | 22:28 | |
| rwellum | mikal: openstack version? | 22:30 |
| mikal | I haven't gotten as far as installing openstack | 22:30 |
| rwellum | mikal: I think first debug locally, because lot of folks using canal with no issues. | 22:30 |
| mikal | I'm still at the step just before "verify installation" for kube at https://docs.openstack.org/kolla-kubernetes/latest/deployment-guide.html | 22:31 |
| rwellum | Sorry kubectl version etc. | 22:31 |
| mikal | user@lab8:~$ kubectl version | 22:31 |
| mikal | Client Version: version.Info{Major:"1", Minor:"8", GitVersion:"v1.8.4", GitCommit:"9befc2b8928a9426501d3bf62f72849d5cbcd5a3", GitTreeState:"clean", BuildDate:"2017-11-20T05:28:34Z", GoVersion:"go1.8.3", Compiler:"gc", Platform:"linux/amd64"} | 22:31 |
| mikal | Server Version: version.Info{Major:"1", Minor:"8", GitVersion:"v1.8.4", GitCommit:"9befc2b8928a9426501d3bf62f72849d5cbcd5a3", GitTreeState:"clean", BuildDate:"2017-11-20T05:17:43Z", GoVersion:"go1.8.3", Compiler:"gc", Platform:"linux/amd64"} | 22:31 |
| mikal | It looks like one of the containers in that pod isn't working, but I'm still trying to learn enough kube to work out which one... | 22:31 |
| rwellum | Haven't tried 1.8.4 yet - 1.8.2 yes. | 22:32 |
| mikal | kube-system canal-2tckp 2/3 CrashLoopBackOff 6 10m 192.168.50.32 lab8 | 22:32 |
| rwellum | So "kubectl describe pod canal-2tckp -n kube-system" | 22:32 |
| mikal | Oh, its the calico node | 22:32 |
| rwellum | yup | 22:34 |
| mikal | Ahhh, this is the thing I think: | 22:34 |
| mikal | time="2017-11-20T22:33:17Z" level=error msg="Hit error initializing TPR" error="resource does not exist: {{ } {global-config.projectcalico.org kube-system %!s(int64=0) 0001-01-01 00:00:00 +0000 UTC <nil> %!s(*int64=<nil>) map[] map[] [] [] } Calico Global Configuration [{v1}]}" | 22:34 |
| mikal | At least that's an error I can google | 22:34 |
| rwellum | Also: kubectl log canal-2tckp -n kolla | 22:34 |
| rwellum | Running from VM? | 22:35 |
| mikal | No, this is a baremetal install | 22:35 |
| rwellum | Ah tricky - you're brave :) | 22:35 |
| rwellum | You can ping google.com right? | 22:36 |
| rwellum | centos or ubuntu and version? | 22:36 |
| mikal | https://github.com/samsung-cnct/kraken-lib/issues/901 | 22:36 |
| mikal | Ubuntu 16.04 | 22:36 |
| mikal | Yeah, the machine has working internet connectivity | 22:36 |
| mikal | I think this is a canal version incompatability | 22:37 |
| mikal | So if I change the canal version in canal.yaml and then kubectl apply the new version, does that do the right thing? | 22:39 |
| rwellum | So which version did you download? | 22:42 |
| mikal | Of canal? | 22:43 |
| mikal | The install guide uses a YAML file which specifies 2.4.1 | 22:43 |
| mikal | That bug I found implies I should use 2.6.1 instead | 22:43 |
| mikal | Giving it a try now | 22:43 |
| *** yangyapeng has joined #openstack-kolla | 22:44 | |
| inc0 | mikal: you can try different network overlays | 22:45 |
| inc0 | canal is quite sophiscicated one | 22:45 |
| rwellum | Yeah I am using: " image: quay.io/calico/node:v2.6.2" | 22:45 |
| inc0 | and unless you want calico I'd stick to flannel or weave | 22:45 |
| mikal | inc0: oh, I had flannel working before, but the docs strongly suggested canal | 22:45 |
| mikal | inc0: so I rebuilt with canal | 22:45 |
| inc0 | well canal is calico-in-flannel | 22:46 |
| inc0 | so it's another layer of complexity out there | 22:46 |
| inc0 | cool tech, but I wouldn't call it simplest | 22:46 |
| *** rwsu has quit IRC | 22:47 | |
| mikal | Hmmm, do I have to reinstall the OS to make canal go away? I can tear down the kube and rebuild, but I don't know what else (i.e. network config) that might have been done | 22:47 |
| rwellum | We use it in the gates, and the deployment guide though inc0 | 22:47 |
| mikal | Yeah, I'm just tyring to follow the deployment guide | 22:47 |
| inc0 | rwellum: yeah I know, it's cool, just saying that there are simpler ones | 22:47 |
| inc0 | and it doesn't affect opesntack itself that much | 22:47 |
| *** jamesbenson has quit IRC | 22:48 | |
| rwellum | You may be able to 'kubectl delete pod <name> -n kube-system - after you make your changes mikal | 22:48 |
| mikal | Where is the gate job defined? It would be interesting to compare that with the deployment guide and see if there's a reason this isn't working | 22:48 |
| rwellum | But might be safer to restart. | 22:48 |
| *** jamesbenson has joined #openstack-kolla | 22:48 | |
| *** yangyapeng has quit IRC | 22:49 | |
| inc0 | mikal: https://github.com/openstack/kolla-kubernetes/tree/master/tests/bin | 22:49 |
| *** itlinux has quit IRC | 22:49 | |
| rwellum | They are a little hard to read mikal | 22:49 |
| rwellum | You can also try: https://github.com/openstack/kolla-kubernetes/blob/master/contrib/orchestration/ko.py | 22:49 |
| rwellum | For a quick start to test your environment etc. | 22:49 |
| *** itlinux has joined #openstack-kolla | 22:50 | |
| mikal | Oh, you're using a different version of canal in the gate... | 22:51 |
| mikal | 1.7 vs 1.6 in the deploy guide | 22:51 |
| inc0 | https://github.com/openstack/kolla-kubernetes/blob/1d1a9c2177b87866685249a5599ea95263174a21/tools/setup_gate_ceph.sh if you stat from this file you should be able to follow this | 22:51 |
| inc0 | yeah deploy guide wasn't updated I guess | 22:51 |
| rwellum | Deployment guides always get out of date | 22:51 |
| rwellum | But the tool above is up-todate :) | 22:52 |
| mikal | Yep, I have to go to a meeting now but will continue to play later | 22:55 |
| mikal | If I can identify the updates the deployment guide needs I'll send a patch | 22:55 |
| *** lrensing has quit IRC | 22:55 | |
| rwellum | Sure mikal - I have a few too :) | 22:55 |
| mikal | Oh, bumping to 1.7 worked! | 22:59 |
| mikal | That's a win | 22:59 |
| *** kristaps_ has joined #openstack-kolla | 23:00 | |
| rwellum | nice! | 23:02 |
| mikal | Now to see if busybox runs | 23:02 |
| mikal | OMG, it works! | 23:02 |
| *** lvdombrkr has quit IRC | 23:02 | |
| mikal | I am a genius! | 23:02 |
| *** jamesbenson has quit IRC | 23:03 | |
| *** pc_m has quit IRC | 23:11 | |
| *** pc_m has joined #openstack-kolla | 23:12 | |
| *** jrist is now known as jrist-afk | 23:13 | |
| mikal | Oh course, then I get weird errors from the next step... | 23:14 |
| mikal | :~$ kubectl apply -f rbac-kolla.yaml | 23:14 |
| mikal | error: error validating "rbac-kolla.yaml": error validating data: unknown object type schema.GroupVersionKind{Group:"rbac.authorization.k8s.io", Version:"v1alpha1", Kind:"ClusterRoleBinding"}; if you choose to ignore these errors, turn validation off with --validate=false | 23:14 |
| masber | good morning, do I need to specify a storage node in the inventory file if I choose to use external ceph? I am using kolla-ansible | 23:16 |
| inc0 | masber: yes | 23:26 |
| *** lrensing has joined #openstack-kolla | 23:27 | |
| inc0 | you still need cinder-volume to land somewhere | 23:27 |
| masber | inc0, you mean cinder-volume container? | 23:27 |
| inc0 | yes | 23:27 |
| inc0 | cinder-volume service is deploymet on [storage] | 23:27 |
| masber | ok, so I still need to put ceph nodes as storage nodes in the ansible inventory file to deploy cinder-volume container there? or do i only need 1 ceph node there? | 23:29 |
| *** lpetrut has quit IRC | 23:32 | |
| masber | inc0, ^^^ | 23:33 |
| inc0 | no | 23:33 |
| inc0 | you just neeed a storage node | 23:33 |
| inc0 | or nodes | 23:33 |
| inc0 | doesn't need to be same as ceph nodes | 23:33 |
| inc0 | what cinder-volume will do is it'll conenct to ceph-mons to deal with volumes | 23:34 |
| masber | ok, thank you very much for the clarification | 23:34 |
| inc0 | no worries | 23:34 |
| masber | I will give it a go :) | 23:34 |
| *** lrensing has quit IRC | 23:34 | |
| *** lrensing has joined #openstack-kolla | 23:41 | |
| *** gfhellma has joined #openstack-kolla | 23:49 | |
| *** lrensing has quit IRC | 23:55 | |
| *** yingjun has joined #openstack-kolla | 23:56 | |
| *** Pavo has joined #openstack-kolla | 23:58 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!