| *** hongbin has joined #openstack-kolla | 01:10 | |
| *** hongbin has quit IRC | 01:28 | |
| *** gfidente|afk has quit IRC | 01:41 | |
| *** hongbin has joined #openstack-kolla | 01:43 | |
| *** zhanglong has joined #openstack-kolla | 02:43 | |
| *** jcmdln has quit IRC | 02:57 | |
| *** hongbin has quit IRC | 04:50 | |
| *** evrardjp has quit IRC | 05:34 | |
| *** evrardjp has joined #openstack-kolla | 05:34 | |
| *** jcmdln has joined #openstack-kolla | 05:49 | |
| *** bjolo has quit IRC | 06:46 | |
| *** k_mouza has joined #openstack-kolla | 06:53 | |
| *** k_mouza has quit IRC | 06:58 | |
| *** pbing19 has joined #openstack-kolla | 07:57 | |
| openstackgerrit | Dincer Celik proposed openstack/kolla stable/train: [train] Bump versions https://review.opendev.org/706659 | 08:33 |
|---|---|---|
| openstackgerrit | Dincer Celik proposed openstack/kolla stable/stein: [stein] Bump versions https://review.opendev.org/706660 | 08:35 |
| openstackgerrit | Dincer Celik proposed openstack/kolla stable/rocky: [rocky] Bump versions https://review.opendev.org/706661 | 08:40 |
| *** zhanglong has quit IRC | 08:42 | |
| *** zhanglong has joined #openstack-kolla | 09:29 | |
| *** pbing19 has quit IRC | 09:38 | |
| *** pbing19 has joined #openstack-kolla | 09:39 | |
| *** zhanglong has quit IRC | 09:57 | |
| *** zhanglong has joined #openstack-kolla | 09:58 | |
| *** zhanglong has quit IRC | 10:19 | |
| *** jopdorp has joined #openstack-kolla | 10:20 | |
| jopdorp | hi all | 10:30 |
| jopdorp | I know it's the weekend, but I've been facing this issue since yesterday and don't know what to do with it | 10:31 |
| jopdorp | this step fails for a multinode deployment with ubuntu as base os rabbitmq : Waiting for rabbitmq to start on first node | 10:31 |
| jopdorp | I'm using private key for ssh | 10:32 |
| jopdorp | I saw that there used to be some problems with the rabbitmq container constantly restarting | 10:32 |
| jopdorp | could that still be an issue? | 10:32 |
| jopdorp | yeah, it's restarting | 10:35 |
| jopdorp | it looks like this issue, https://bugs.launchpad.net/kolla-ansible/+bug/1837699 | 10:39 |
| openstack | Launchpad bug 1837699 in kolla-ansible train "[ubuntu host] Handle duplicated hostnames in /etc/hosts file (127.0.1.1 problem)" [High,Fix released] - Assigned to Dincer Celik (osmanlicilegi) | 10:39 |
| jopdorp | but i think it should have been fixed? | 10:40 |
| jopdorp | hH | 10:40 |
| jopdorp | haha | 10:40 |
| jopdorp | yeah so exactly that one | 10:40 |
| jopdorp | should i remove the entries in etc hosts in all my nodes or just in control? | 10:41 |
| jopdorp | i removed them just from control, think it's enough, thanks! | 10:50 |
| *** jopdorp has quit IRC | 11:13 | |
| *** vesper11 has joined #openstack-kolla | 12:08 | |
| *** vesper has quit IRC | 12:08 | |
| *** mloza has quit IRC | 12:52 | |
| *** negronjl has quit IRC | 12:54 | |
| *** negronjl has joined #openstack-kolla | 12:55 | |
| *** admin0 has joined #openstack-kolla | 13:44 | |
| admin0 | quick question .. how big ( in terms of cpu/ram/disk ) resource would be required to test kolla all-in-one with all services enabled .. where the goal is to test 1 service at a time and not run 100s of it | 13:48 |
| admin0 | just to say "ok.. it works" | 13:48 |
| admin0 | trying to run kolla aio trein .. got error on: TASK [mariadb : Copying over galera.cnf] The error was: TypeError: unicode argument expected, got 'str' .. i did not even touch this file .. | 13:56 |
| admin0 | is this a fault in our QA :) ? | 14:00 |
| *** pbing19 has quit IRC | 14:35 | |
| yoctozepto | admin0: this sounds like using python2 on deployment machine and python3 on target (or vice versa, never remember) | 15:04 |
| yoctozepto | ;-) | 15:04 |
| r3ap3r | yoctozepto: so does that mean at this point in time, we can drop python2 and strictly use python3 for Kolla deployments? | 15:05 |
| yoctozepto | r3ap3r: ansible works reliably with the same python version on both sides | 15:06 |
| r3ap3r | yoctozepto: Ok cool. I will try a python3 deployment sometime in the next couple of weeks and see how my mileage goes. Thanks. | 15:07 |
| yoctozepto | r3ap3r: feature-wise it ends up the same because containers are separate | 15:09 |
| *** vesper11 has quit IRC | 15:09 | |
| *** vesper11 has joined #openstack-kolla | 15:11 | |
| r3ap3r | yoctozepto: +1 | 15:12 |
| openstackgerrit | Radosław Piliszek proposed openstack/kolla-ansible master: Add --clean to kolla-mergepwd https://review.opendev.org/706668 | 15:28 |
| openstackgerrit | Radosław Piliszek proposed openstack/kolla-ansible master: Docs: fix cells upgrade guide https://review.opendev.org/706669 | 15:30 |
| openstackgerrit | Dincer Celik proposed openstack/kolla stable/train: [train] Bump versions https://review.opendev.org/706659 | 15:50 |
| openstackgerrit | Dincer Celik proposed openstack/kolla stable/train: [train] Bump versions https://review.opendev.org/706659 | 15:54 |
| openstackgerrit | Dincer Celik proposed openstack/kolla stable/stein: [stein] Bump versions https://review.opendev.org/706660 | 15:56 |
| openstackgerrit | Dincer Celik proposed openstack/kolla stable/rocky: [rocky] Bump versions https://review.opendev.org/706661 | 15:58 |
| admin0 | yoctozepto, its a all in one to test first .. i followed steps on https://docs.openstack.org/kolla-ansible/latest/user/quickstart.html | 16:01 |
| admin0 | the document does not say anywhere regarding python versions | 16:02 |
| yoctozepto | admin0: well, it's hard to get wrong python versions in an aio - did you use virtualenv with custom python? | 16:08 |
| yoctozepto | i.e. it's a different python than if you now ssh'ed back to the machine | 16:08 |
| admin0 | all i did was apt-get install python-virtualenv on ubuntu 18.04 and then source it | 16:23 |
| admin0 | and proceeded with the rest of the copy/paste commands | 16:23 |
| admin0 | let me destroy this box and do it again one more time .. will document what i did | 16:24 |
| *** pbing19 has joined #openstack-kolla | 16:44 | |
| *** evrardjp has quit IRC | 17:34 | |
| *** evrardjp has joined #openstack-kolla | 17:34 | |
| *** devfaz has quit IRC | 17:49 | |
| *** devfaz has joined #openstack-kolla | 17:49 | |
| *** pbing19 has quit IRC | 17:54 | |
| *** born2bake has joined #openstack-kolla | 17:56 | |
| admin0 | yoctozepto, this time different output: https://gist.github.com/a1git/788efba8948d5695e19b39c51442e7e2 --- please suggest where I missed following the documentation | 18:03 |
| yoctozepto | admin0: the bad part is where you did the git clone and left it at the default branch - you got master which is for ussuri | 18:10 |
| yoctozepto | this is also the reason for step duplication | 18:10 |
| yoctozepto | they are duplicated because you did both the "deployment/eval" part and "devel" part | 18:10 |
| admin0 | so #1. we need to add git checkout (correct branch) in the documentation ? | 18:11 |
| admin0 | the deployment/eval an devel part, i did only in the gen-passwords | 18:11 |
| yoctozepto | I think we need to rewrite it so that nobody thinks these are the next steps | 18:11 |
| yoctozepto | well, first you installed k-a from pypi | 18:12 |
| yoctozepto | and then from git | 18:12 |
| yoctozepto | so you rewrote train stable release with ussuri master | 18:12 |
| yoctozepto | I agree that docs are not the best, I'll see to having them more clear | 18:13 |
| admin0 | i see it | 18:13 |
| admin0 | i will correct this, move forward and document where i get stuck again | 18:14 |
| *** pbing19 has joined #openstack-kolla | 18:14 | |
| admin0 | i still need to be root for the all-in-one which is on localhost when doing the precheck, bootstrap etc right ? | 18:15 |
| dking_desktop | I notice that when k-a setups up ovs, all of the internal ports are down. I can bring them up by manually creating the ifcfg-* files with IP and netmask added and then ifup, but is there some place in the configuration where that could be done? | 18:16 |
| yoctozepto | admin0: thanks, that's helpful the way you give feedback | 18:16 |
| yoctozepto | admin0: or have passwordless sudo | 18:16 |
| yoctozepto | admin0: this is general ansible knowledge but it might be wise to repeat it here, indeed | 18:16 |
| yoctozepto | dking_desktop: why do you want them to be up? | 18:16 |
| yoctozepto | dking_desktop: these are nothing to worry about | 18:16 |
| admin0 | i started the same way when starting with osa also (many years ago) .. making sure the documentation is easy to follow for everyone to get the basics up and running | 18:17 |
| dking_desktop | yoctozepto: I'm still working on getting Ironic to work with my untagged networks. Maybe I'm going about it the wrong way, though. | 18:17 |
| yoctozepto | dking_desktop: http://paste.openstack.org/show/789321/ <- with these STATES all networking works fine | 18:18 |
| yoctozepto | so you've got DOWN on 4 common | 18:19 |
| yoctozepto | and UNKNOWN on specific | 18:19 |
| yoctozepto | and life is good | 18:19 |
| dking_desktop | I have two NICs which for redundancy need to be using a bond (bond0). All of the rest of my connections use VLANs, which is working fine. I set bond0 as neutron_external_interface in globals.yml, but saw that this overwrites any IP and I can't use that interface for ironic_dnsmasq. However, if I manually bring br-ex up, it seemed to work. | 18:20 |
| yoctozepto | admin0: yeah, the thing with docs is that people perceive them differently and it's no easy task to get them really right | 18:20 |
| dking_desktop | ...or maybe I should just be making a separate port? | 18:20 |
| admin0 | yoctozepto, but we can also separate this to an all-in-one .. like its done in ansible: https://docs.openstack.org/openstack-ansible/latest/user/aio/quickstart.html | 18:21 |
| admin0 | there is just one set of instructions to follow .. avoiding confusion | 18:22 |
| admin0 | yoctozepto, maybe i will contribute and create this document and submit a request | 18:22 |
| yoctozepto | admin0: yeah, I was thinking about separate aio/multinode and separate eval+deploy/dev | 18:22 |
| yoctozepto | admin0: plus mentioning definite release, not any | 18:23 |
| yoctozepto | dking_desktop: bond0 is fine; the thing with neutron_external_interface it gets captured as bridge for physnet1 | 18:23 |
| yoctozepto | so yeah, it loses any ip addressing | 18:23 |
| yoctozepto | ironic_dnsmasq is for provisioning | 18:24 |
| yoctozepto | afair ;-) | 18:24 |
| admin0 | how do i contribute to the docs ? is it the same way as in openstack-ansible ? | 18:24 |
| yoctozepto | so you probably want to separate a tagged interface for it and use it | 18:24 |
| yoctozepto | bond0.xxx | 18:24 |
| yoctozepto | e.g. bond0.123 | 18:24 |
| dking_desktop | Okay. I'm sure I'm asking the wrong question, then. What should I use for ironic_dnsmasq_interface? | 18:24 |
| yoctozepto | well, that xxx was most unfortunate as it is probably some pr0n site, oh well | 18:25 |
| yoctozepto | ;D | 18:25 |
| dking_desktop | Well, the problem there is that my servers can't handle VLANs for PXE, and the servers need to be on trunk ports. | 18:25 |
| yoctozepto | admin0: yeah, the gerrit review | 18:25 |
| admin0 | ok | 18:25 |
| yoctozepto | dking_desktop: yeah, you don't need to tag these for pxe | 18:26 |
| dking_desktop | lol. Maybe that's IEEE 802.1xxx? | 18:26 |
| yoctozepto | it could be untagged on them | 18:26 |
| yoctozepto | but tagged on port to neutron | 18:26 |
| yoctozepto | lemme find usage of ironic_dnsmasq in the code | 18:26 |
| dking_desktop | It looks like ironic_dnsmasq_interface sets interface in dnsmasq.conf inside of the ironic_dnsmasq container. | 18:27 |
| yoctozepto | oh, it's for inspector | 18:27 |
| yoctozepto | yeah | 18:27 |
| yoctozepto | so it's for initial enrollment | 18:28 |
| dking_desktop | Correct. And once they are up, they will need to be able to communicate through multiple VLANs, so they are trunked. | 18:30 |
| yoctozepto | yeah, trunked but the native/default/whatever should be this one | 18:31 |
| dking_desktop | Yep. I've tried about every other combination I could think of. Our switches have VLAN 1 set for the native VLAN, so I tried bond0.1. I see that sending from that becomes untagged, which is great, but if I set dnsmasq's interface to bond0.1, it still doesn't see the packets. | 18:32 |
| yoctozepto | dking_desktop: they need to be tagged on that switch's port to that server with dnsmasq | 18:33 |
| dking_desktop | The only way I've been able to get the controller to accept untagged frames so far has been to use an internal port. | 18:33 |
| yoctozepto | what is an internal port? | 18:34 |
| dking_desktop | I meant the port type internal through ovs, such as br-ex. Really, that's the only one I tried so far, but that worked fine. | 18:34 |
| yoctozepto | ah, you mean you hacked on kolla ovs internals, not nice! :D | 18:36 |
| yoctozepto | what you need to do is configure the switch port to dnsmasq to tag that vlan 1 | 18:37 |
| dking_desktop | lol. Maybe not, but it wasn't being too nice to me. And since the documentation hasn't been helping me, I thought I would try to dig until I could find something that worked, and then work backwards from that. | 18:37 |
| yoctozepto | because otherwise you can't separate that traffic | 18:37 |
| yoctozepto | dking_desktop: I was joking :-) but that does not change the fact that it's far from a supported solution | 18:37 |
| dking_desktop | True. I just haven't figured out a supported solution that works. The documentation looks to me like it assumes that I have a spare NIC. Would you know where I could find the proper documentation? | 18:40 |
| admin0 | yoctozepto, https://gist.github.com/a1git/9fc701e4d2aaec41f46499cefe2ede78 ... looks like i am still missing something | 18:43 |
| dking_desktop | I'm hoping to avoid too many custom changes to the switch port. I might try that, but it would be Monday before I could give it a shot. | 18:44 |
| yoctozepto | dking_desktop: yeah, spare, but .123 is also a spare | 18:46 |
| yoctozepto | the issue really is linux does not differentiate between "any" and "untagged" | 18:46 |
| yoctozepto | so it's best done on switch side | 18:46 |
| dking_desktop | It's a spare VLAN, though. But perhaps other folks aren't attempting to provision baremetal nodes that use trunked ports. | 18:47 |
| yoctozepto | admin0: very weird | 18:49 |
| yoctozepto | dking_desktop: trunk is a must | 18:49 |
| admin0 | yoctozepto, on my 8 months old attempt, documented here: https://www.openstackfaq.com/openstack-kolla-all-in-one/ i see a line where it says pip install kolla-ansible docker ( when not using virualenv) | 18:51 |
| yoctozepto | admin0: could you sahre the logs from bootstrap? | 18:51 |
| admin0 | sure | 18:51 |
| admin0 | let me run it again | 18:51 |
| yoctozepto | admin0: yeah, it does install it in bootstrap | 18:51 |
| yoctozepto | no idea why it did not now ;p | 18:52 |
| admin0 | it now installed docker, on the 2nd run | 18:53 |
| admin0 | first run: https://gist.github.com/a1git/06ed6534203e41c0af2456991a2b126b second run: https://gist.github.com/a1git/0d0916c52d16423767d67264fa62f4a1 | 18:55 |
| admin0 | i did not used -v in first, used -v in 2nd | 18:55 |
| admin0 | running prechecks now :) | 18:55 |
| admin0 | which failed again in the same point .. failed in checking docker sdk version | 18:56 |
| admin0 | ImportError: No module named docker | 18:56 |
| admin0 | i think inside the virtualenv, its needed to do pip install docker | 18:56 |
| admin0 | which is in my old documentation, but unable to find in new one | 18:56 |
| admin0 | it returned with requirement already satisfied | 18:57 |
| *** vmixor has joined #openstack-kolla | 18:59 | |
| dking_desktop | yoctozepto: Thanks for the tips. I'm going to think about that and maybe re-visit the switch configuration on Monday. In the meantime, I might push ahead with what I've got manually. Thank you. | 19:01 |
| yoctozepto | dking_desktop: yw, good luck! | 19:01 |
| admin0 | i am going to retry wtihout using virtualenv | 19:01 |
| yoctozepto | admin0: hmm, odd, it installed it in the venv | 19:02 |
| yoctozepto | admin0: I wonder if ansible started doing us a "favor" there | 19:02 |
| admin0 | well, i am still nowhere to start :) and test | 19:03 |
| yoctozepto | admin0: https://docs.openstack.org/kolla-ansible/latest/reference/deployment-and-bootstrapping/bootstrap-servers.html#initial-bootstrap-considerations | 19:04 |
| yoctozepto | well, this mentions it | 19:04 |
| yoctozepto | so quickstart is bad at suggesting venv for beginners :-) | 19:05 |
| *** pbing19 has quit IRC | 19:07 | |
| yoctozepto | well, no, it does not make sense still | 19:14 |
| yoctozepto | it explicitly says it's not using virtualenv there | 19:14 |
| yoctozepto | why did it pick up this pip... | 19:14 |
| *** vmixor has quit IRC | 19:23 | |
| openstackgerrit | Radosław Piliszek proposed openstack/kolla-ansible master: DNM: Test venv https://review.opendev.org/706673 | 19:25 |
| yoctozepto | admin0: I've set up some poc in CI ^ | 19:25 |
| yoctozepto | I wonder what happens in there | 19:25 |
| openstackgerrit | Radosław Piliszek proposed openstack/kolla-ansible master: DNM: Test venv https://review.opendev.org/706673 | 19:36 |
| admin0 | "so quickstart is bad at suggesting venv for beginners :-)" -- yes .. hours lost | 19:48 |
| admin0 | trying to follow the docs | 19:48 |
| yoctozepto | admin0: well, the poc actually works so I'm still unsure what happens in your case... | 20:04 |
| yoctozepto | "/usr/local/bin/pip2", | 20:12 |
| yoctozepto | despite using venv | 20:12 |
| yoctozepto | hmm | 20:13 |
| *** goldyfruit has quit IRC | 20:17 | |
| *** goldyfruit has joined #openstack-kolla | 20:18 | |
| *** goldyfruit has quit IRC | 20:18 | |
| *** goldyfruit has joined #openstack-kolla | 20:19 | |
| *** goldyfruit has quit IRC | 20:20 | |
| *** goldyfruit has joined #openstack-kolla | 20:20 | |
| openstackgerrit | Radosław Piliszek proposed openstack/kolla-ansible master: DNM: Test venv https://review.opendev.org/706673 | 20:29 |
| *** negronjl has quit IRC | 20:33 | |
| *** negronjl has joined #openstack-kolla | 20:37 | |
| *** goldyfruit has quit IRC | 21:07 | |
| *** bjolo has joined #openstack-kolla | 21:17 | |
| yoctozepto | admin0: oddly all are passing | 21:38 |
| yoctozepto | admin0: I thought it could be due to localhost but it passed anyways | 21:38 |
| admin0 | i will redo a fresh ubuntu install tomorrow and retry | 21:39 |
| yoctozepto | admin0: and I'll do a poc for train - could be ussuri is fine but train is not - since you move farther with ussuri previously | 21:43 |
| yoctozepto | good night | 21:44 |
| admin0 | good night .. thanks .. will test tomororw | 22:10 |
| *** goldyfruit has joined #openstack-kolla | 23:46 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!