| *** amotoki_away is now known as amotoki | 01:03 | |
| *** yamamoto has joined #openstack-lbaas | 02:16 | |
| *** sanfern has joined #openstack-lbaas | 02:48 | |
| *** fnaval has joined #openstack-lbaas | 03:29 | |
| *** links has joined #openstack-lbaas | 03:35 | |
| *** links has quit IRC | 03:43 | |
| *** links has joined #openstack-lbaas | 03:45 | |
| *** aojea has joined #openstack-lbaas | 04:26 | |
| *** Dinesh_Bhor has joined #openstack-lbaas | 04:29 | |
| *** aojea has quit IRC | 04:30 | |
| *** krypto has joined #openstack-lbaas | 04:35 | |
| *** krypto has quit IRC | 04:35 | |
| *** krypto has joined #openstack-lbaas | 04:35 | |
| *** armax has quit IRC | 04:47 | |
| *** bzhao has quit IRC | 05:00 | |
| *** bzhao has joined #openstack-lbaas | 05:02 | |
| *** yamamoto has quit IRC | 05:02 | |
| *** gcheresh has joined #openstack-lbaas | 05:26 | |
| *** krypto has quit IRC | 05:29 | |
| *** yamamoto has joined #openstack-lbaas | 05:45 | |
| *** fnaval has quit IRC | 06:07 | |
| *** rcernin has joined #openstack-lbaas | 06:33 | |
| *** aojea has joined #openstack-lbaas | 07:20 | |
| *** aojea has quit IRC | 07:20 | |
| *** aojea has joined #openstack-lbaas | 07:20 | |
| *** tesseract has joined #openstack-lbaas | 07:33 | |
| *** krypto has joined #openstack-lbaas | 07:53 | |
| *** krypto has quit IRC | 07:58 | |
| *** krypto has joined #openstack-lbaas | 08:00 | |
| *** fnaval has joined #openstack-lbaas | 08:08 | |
| *** fnaval has quit IRC | 08:13 | |
| nmagnezi | rm_work o/ | 08:14 |
|---|---|---|
| *** gans has joined #openstack-lbaas | 08:14 | |
| gans | anyone facing issues with amphora image created via disk-image-create utility | 08:16 |
| gans | for me it is giving error "Internal Server Error" while creating LB. I am using stable/newton | 08:17 |
| nmagnezi | gans, hi there, where do you see the "Internal Server Error" ? | 08:18 |
| nmagnezi | gans, also, which image are you using, Ubuntu? | 08:18 |
| gans | in octavia-worker log | 08:18 |
| gans | while plugging VIP in the amphora, it is making the REST call, but failing with code 500 | 08:18 |
| gans | yeah, ubuntu-xenial | 08:19 |
| nmagnezi | gans, i suggest you'll login the the amphora instance before you try to plug the vip and tail the amphora-agent log (/var/log/amphora-agent.log IIRC) to get more details | 08:19 |
| gans | it is deleting the VM once LB goes in ERROR state | 08:20 |
| gans | how can i defer that | 08:20 |
| nmagnezi | gans, i can't recall at the moment, i know some cores here know a way but I don't at the moment. anyhow, it won't matter since you'll tail the log *before* it is deleted so it will get printer to your terminal | 08:21 |
| gans | 2017-05-29 08:20:34.186 13099 WARNING octavia.controller.worker.controller_worker [-] Task 'octavia.controller.worker.tasks.amphora_driver_tasks.AmphoraePostVIPPlug' (7e9fa09c-5047-46f6-b8be-b2c978e5c95c) transitioned into state 'FAILURE' from state 'RUNNING' | 08:22 |
| gans | thanks, i will try that | 08:22 |
| nmagnezi | gans, yup. that log snippet does not say much :-) | 08:22 |
| nmagnezi | gans, np | 08:22 |
| gans | i was just trying to point out the flow in which it is failing. Interesting thing, i am able to create LB with images generated on Saturday. | 08:23 |
| gans | Here is the log from amphora-agent | 08:45 |
| gans | http://paste.openstack.org/show/610830/ | 08:45 |
| nmagnezi | gans, strange.. i don't see anything going on there | 08:46 |
| nmagnezi | gans, did you attach a vip? | 08:47 |
| nmagnezi | gans, maybe try to create a listener? | 08:47 |
| gans | LB is going in error state and VM is getting deleted. | 08:48 |
| gans | i am not manually adding the VIP, it does automatically as part of loadbalancer-create | 08:48 |
| nmagnezi | gans, yeah i know, but nothing happens in that log so I was wondering if that part works for you now | 08:49 |
| nmagnezi | nevermind | 08:49 |
| gans | this time i tried stable/ocata branch | 08:49 |
| nmagnezi | hmm... | 08:49 |
| *** belharar has joined #openstack-lbaas | 08:49 | |
| nmagnezi | gans, maybe you need to set a pdb in the worker to allow yourself time to login and also set a pdb in the agent for in the vip plug part so you get the exception | 08:51 |
| nmagnezi | gans, if you know how, I can point you to the relevant parts in the code | 08:51 |
| nmagnezi | gans, for the amphora vm you'll need to shutdown the agent from systemd and start it in a bash prompt so pdb will work | 08:52 |
| gans | nmagnezi, i know how | 08:52 |
| nmagnezi | gans, that's what i would do to debug this | 08:52 |
| nmagnezi | gans, ack | 08:52 |
| nmagnezi | gans, shall I point you in stable/newton? | 08:53 |
| gans | i had done pdf for haproxy config on agent | 08:53 |
| gans | pdb* | 08:53 |
| gans | now i am on stable/ocata | 08:53 |
| nmagnezi | gans, just tell me which version you prefer and I'll provide the code entry points for you to place pdb :) | 08:54 |
| gans | stable/ocata please, sorry for confusion | 08:54 |
| nmagnezi | gans, for the controller worker i'd place a pdb right after the request to nova: https://github.com/openstack/octavia/blob/stable/ocata/octavia/compute/drivers/nova_driver.py#L145 | 08:56 |
| nmagnezi | gans, at the amphora agent side, you can place a pdb here: https://github.com/openstack/octavia/blob/stable/ocata/octavia/amphorae/backends/agent/api_server/server.py#L146 | 08:56 |
| gans | okay | 08:56 |
| gans | thanks, will try that now | 08:57 |
| nmagnezi | gans, so place a pdb, restart the worker process and create a loadbalancer | 08:57 |
| nmagnezi | gans, when the instance booted, login and place the breakpoint in the agent | 08:57 |
| nmagnezi | gans, start the agent yourself (ps -ef | grep amphora-agent to get the command line for it), but don't forget to 'systemctl stop amphora-agent.service' first | 08:58 |
| gans | awesome, doing it now | 08:58 |
| nmagnezi | gans, one last thing before you start the agent | 08:59 |
| nmagnezi | gans, enable debug mode in /etc/octavia/amphora-agent.conf | 08:59 |
| nmagnezi | can't hurt | 08:59 |
| gans | okay | 08:59 |
| gans | it will never :-) | 08:59 |
| *** krypto has quit IRC | 09:01 | |
| gans | http://paste.openstack.org/show/610833/ | 09:12 |
| gans | amphora interface failed to come up | 09:12 |
| gans | nmagnezi, any clues | 09:14 |
| *** krypto has joined #openstack-lbaas | 09:34 | |
| *** krypto has quit IRC | 09:34 | |
| *** krypto has joined #openstack-lbaas | 09:34 | |
| *** fnaval has joined #openstack-lbaas | 10:08 | |
| *** fnaval has quit IRC | 10:14 | |
| xgerman | https://www.irccloud.com/pastebin/cy9h6pPO/ | 10:17 |
| nmagnezi | gans, check xgerman's url, he pointed you to the problem | 10:24 |
| nmagnezi | xgerman, hi btw o/ :D | 10:24 |
| gans | xgerman, nmagnezi, i will try that now | 10:25 |
| *** sanfern has quit IRC | 10:47 | |
| *** yamamoto has quit IRC | 10:55 | |
| gans | i put breakpoint here : https://github.com/openstack/octavia/blob/stable/ocata/octavia/amphorae/backends/agent/api_server/osutils.py#L186 | 11:03 |
| gans | on pdf prompt, fetched the command and executed on separate window, got same error | 11:03 |
| gans | ubuntu@amphora-65f4000d-c620-48b0-a0b9-dd5f7814b831:~$ sudo su | 11:03 |
| gans | sudo: unable to resolve host amphora-65f4000d-c620-48b0-a0b9-dd5f7814b831 | 11:03 |
| gans | root@amphora-65f4000d-c620-48b0-a0b9-dd5f7814b831:/home/ubuntu# ip netns exec amphora-haproxy ifup eth1 | 11:03 |
| gans | Cannot find device "eth1" | 11:03 |
| gans | Failed to bring up eth1. | 11:03 |
| *** gans has quit IRC | 11:05 | |
| *** aojea has quit IRC | 11:08 | |
| *** circ-user-TeTGj has joined #openstack-lbaas | 11:13 | |
| *** openstackgerrit has joined #openstack-lbaas | 11:23 | |
| openstackgerrit | Nir Magnezi proposed openstack/octavia master: Devstack plugin should mark the HM ovs port for cleanup skip https://review.openstack.org/460528 | 11:23 |
| nmagnezi | xgerman, around? | 11:35 |
| xgerman | my train has Wifi… | 11:35 |
| xgerman | for a few more minutes ;-) | 11:35 |
| nmagnezi | xgerman, lol :-) | 11:35 |
| xgerman | yeah, will be back in the states Friday night… | 11:36 |
| nmagnezi | xgerman, something strange is going on. i set up a new devstack nodes (on Ubuntu, after it fails to create loadbalancers in Centos based devstack -> this is what I try to debug). anyhow it fails to create a loadbalancer for me | 11:36 |
| nmagnezi | xgerman, now i get the same internal server error as gans | 11:36 |
| nmagnezi | xgerman, *but*, the worker service refuses to stop at the breakpoint I set | 11:37 |
| nmagnezi | xgerman, not sure why | 11:37 |
| xgerman | mmh, if eth1 is missing neutron either didn’t plug the port or it is under a different name | 11:37 |
| nmagnezi | xgerman, http://paste.openstack.org/show/610850/ | 11:37 |
| nmagnezi | xgerman, i didn't even got the the point that I'm able to create an instance and debug it | 11:38 |
| nmagnezi | xgerman, a Ubuntu image for the amphora btw | 11:38 |
| xgerman | ah | 11:39 |
| nmagnezi | xgerman, i just rebased one of my patches to see if anything comes up in zuul, maybe something is broken.. | 11:39 |
| xgerman | yeah, it looks like the queue RPC part has issues… | 11:39 |
| nmagnezi | yay :< | 11:40 |
| xgerman | so they use rabbit and there are rabbit CLI commands but I expect that maybe that new systemd stuff is causing trouble | 11:41 |
| nmagnezi | xgerman, i'm still using screens | 11:41 |
| xgerman | ok | 11:41 |
| nmagnezi | xgerman, not sure if that's an issue or not | 11:41 |
| nmagnezi | as if devstack is not bad enough.. | 11:42 |
| xgerman | neither — so the command gets send ok from the controller? | 11:42 |
| xgerman | check the controller logs… | 11:42 |
| nmagnezi | xgerman, from what i go so far (and it's hard to debug when my breakpoints won't work) it spawns a nova instance but it fails not much afterwards and deletes that instance immediately | 11:43 |
| nmagnezi | it looks like it just won't wait for the instance to boot | 11:44 |
| *** yamamoto has joined #openstack-lbaas | 11:44 | |
| *** yamamoto_ has joined #openstack-lbaas | 11:44 | |
| nmagnezi | xgerman, maybe something changed in the image creation and broke the agent? | 11:47 |
| *** yamamoto has quit IRC | 11:48 | |
| xgerman | Ok. Usually in debug logging it will have the nova error | 11:49 |
| nmagnezi | xgerman, checking.. i managed to connect the the instance | 11:50 |
| nmagnezi | xgerman, now the worker tries to communicate with the agent, which is how it should be but didn't happen up until now | 11:51 |
| *** gans has joined #openstack-lbaas | 11:51 | |
| nmagnezi | xgerman, yup. I got the same issue as gans. looking into this. | 11:52 |
| gans | Awesome..So I am not alone | 11:52 |
| nmagnezi | xgerman, i think it happens in the gates as well http://logs.openstack.org/28/460528/5/check/gate-octavia-v1-dsvm-py3x-scenario-ubuntu-xenial/db044e8/logs/screen-o-cw.txt.gz#_May_29_11_49_39_289152 | 11:52 |
| *** chandankumar is now known as chkumar|hackersp | 11:52 | |
| nmagnezi | gans, indeed | 11:53 |
| nmagnezi | rm_work, ^ | 11:53 |
| *** links has quit IRC | 11:57 | |
| *** gans has quit IRC | 11:57 | |
| *** gans has joined #openstack-lbaas | 12:00 | |
| xgerman | Ok. Let's see how good the Swiss cell data is... | 12:02 |
| nmagnezi | xgerman, ha.. i'm looking at this as well. will update you if/when i find anything | 12:03 |
| xgerman | Ok. The gate can't connect to the amp | 12:04 |
| xgerman | So it shuts down | 12:05 |
| nmagnezi | xgerman, something is wrong with the agent | 12:06 |
| nmagnezi | xgerman, until my instance boot i'm looking for recent merges in that area | 12:07 |
| nmagnezi | xgerman, but it's possible that changes in diskimage-builder break us as well | 12:07 |
| xgerman | Ok. There is a setting to prevent the vm from being deleted | 12:07 |
| nmagnezi | please share :-) | 12:08 |
| nmagnezi | btw it's worth documenting this somewhere | 12:08 |
| nmagnezi | johnsom once showed me how, but i forgot | 12:08 |
| xgerman | Same! | 12:08 |
| *** gans has quit IRC | 12:09 | |
| xgerman | Ok. AFK for a while... | 12:10 |
| nmagnezi | xgerman, ack | 12:10 |
| openstackgerrit | Evgeny Fedoruk proposed openstack/neutron-lbaas master: Adding configuration flag for skipping SSL certificate verification https://review.openstack.org/468866 | 12:10 |
| *** gans has joined #openstack-lbaas | 12:12 | |
| *** pcaruana has joined #openstack-lbaas | 12:20 | |
| *** gans has quit IRC | 12:20 | |
| *** gans has joined #openstack-lbaas | 12:22 | |
| *** gans has quit IRC | 12:23 | |
| *** yamamoto_ has quit IRC | 12:37 | |
| *** yamamoto has joined #openstack-lbaas | 12:37 | |
| *** belharar has quit IRC | 12:43 | |
| *** belharar has joined #openstack-lbaas | 12:45 | |
| *** aojea has joined #openstack-lbaas | 12:52 | |
| *** sanfern has joined #openstack-lbaas | 12:54 | |
| *** belharar has quit IRC | 13:01 | |
| *** gans has joined #openstack-lbaas | 13:04 | |
| *** gans_ has joined #openstack-lbaas | 13:08 | |
| *** gans has quit IRC | 13:10 | |
| *** gans_ has quit IRC | 13:10 | |
| *** gans has joined #openstack-lbaas | 13:10 | |
| nmagnezi | xgerman, i suspect the interface naming in the image was changed | 13:11 |
| nmagnezi | xgerman, and i think we have eth1 hardcoded for the namespace | 13:11 |
| nmagnezi | xgerman, so when it tries to ifup eth1 if just fails to locate it | 13:11 |
| xgerman | Likely - though we compare mac addresses-- | 13:13 |
| *** gans has quit IRC | 13:13 | |
| xgerman | Not sure if that logic does something.. | 13:13 |
| *** gans has joined #openstack-lbaas | 13:14 | |
| nmagnezi | xgerman, | 13:14 |
| nmagnezi | xgerman, http://paste.openstack.org/show/610876/ | 13:14 |
| nmagnezi | xgerman, and btw noticed another interesting thing | 13:15 |
| nmagnezi | xgerman, we used to run the agent in python3 using the dib-python executable | 13:15 |
| nmagnezi | xgerman, but now it just uses python2 | 13:15 |
| nmagnezi | very strange.. | 13:15 |
| xgerman | Yeah, that is weird as well... | 13:16 |
| xgerman | https://github.com/openstack/octavia/blob/master/octavia/amphorae/backends/agent/api_server/plug.py | 13:17 |
| xgerman | It searches the name by the mac | 13:18 |
| xgerman | So maybe an old version? | 13:18 |
| nmagnezi | old version of what? | 13:18 |
| nmagnezi | xgerman, we still have references to a hardcoded eth1 https://github.com/openstack/octavia/blob/master/octavia/amphorae/backends/agent/api_server/plug.py#L178 | 13:19 |
| nmagnezi | xgerman, i need to go AFK, will reconnect a bit later to continue working on this | 13:19 |
| *** gans has quit IRC | 13:21 | |
| xgerman | Yeah, that's bad. But we "name" it in the interface file... but who knows... | 13:21 |
| johnsom | We assign those interface names when we add them to the namespace, so it is fine. | 13:40 |
| johnsom | To stop reverts, edit common/taskflow_base and add the taskflow engine option alway_resolve =False (something like that). | 13:41 |
| johnsom | Back in town and working tomorrow | 13:42 |
| openstackgerrit | Evgeny Fedoruk proposed openstack/neutron-lbaas master: Adding configuration flag for skipping SSL certificate verification https://review.openstack.org/468866 | 13:53 |
| *** cpuga has joined #openstack-lbaas | 14:01 | |
| *** cpuga_ has joined #openstack-lbaas | 14:02 | |
| *** cpuga has quit IRC | 14:05 | |
| *** fnaval has joined #openstack-lbaas | 14:10 | |
| *** krypto has quit IRC | 14:10 | |
| *** krypto has joined #openstack-lbaas | 14:11 | |
| *** krypto has quit IRC | 14:11 | |
| *** krypto has joined #openstack-lbaas | 14:11 | |
| *** fnaval has quit IRC | 14:15 | |
| *** cpuga_ has quit IRC | 14:22 | |
| *** leitan has joined #openstack-lbaas | 14:28 | |
| leitan | Hi guys i was looking at manual rescheduling of lbaas to another agent on Mitaka, i know that automatic rescheduling has been implemented on Ocata when a down agent is detected, but is there any sort of manual evacuation on Mitaka ? | 14:29 |
| leitan | maybe rm_work xgerman ? | 14:30 |
| *** cpuga has joined #openstack-lbaas | 14:35 | |
| *** gcheresh has quit IRC | 14:39 | |
| *** catintheroof has joined #openstack-lbaas | 14:47 | |
| *** catinthe_ has joined #openstack-lbaas | 14:48 | |
| *** catintheroof has quit IRC | 14:52 | |
| *** krypto has quit IRC | 15:10 | |
| *** belharar has joined #openstack-lbaas | 15:16 | |
| *** armax has joined #openstack-lbaas | 15:22 | |
| *** cpuga has quit IRC | 15:28 | |
| *** cpuga has joined #openstack-lbaas | 15:29 | |
| *** cpuga has quit IRC | 15:34 | |
| *** chkumar|hackersp is now known as chandankumar | 15:36 | |
| *** krypto has joined #openstack-lbaas | 15:37 | |
| *** belharar has quit IRC | 15:48 | |
| *** JudeC has joined #openstack-lbaas | 15:50 | |
| *** tesseract has quit IRC | 15:53 | |
| *** rcernin has quit IRC | 16:01 | |
| *** aojea has quit IRC | 16:01 | |
| *** aojea has joined #openstack-lbaas | 16:02 | |
| *** aojea has quit IRC | 16:06 | |
| *** fnaval has joined #openstack-lbaas | 16:11 | |
| *** fnaval has quit IRC | 16:15 | |
| *** pcaruana has quit IRC | 16:17 | |
| *** gans has joined #openstack-lbaas | 16:21 | |
| *** JudeC has quit IRC | 16:21 | |
| *** fnaval has joined #openstack-lbaas | 16:22 | |
| *** cpuga has joined #openstack-lbaas | 16:29 | |
| *** armax has quit IRC | 16:36 | |
| *** cpuga has quit IRC | 16:37 | |
| *** cpuga has joined #openstack-lbaas | 16:38 | |
| *** armax has joined #openstack-lbaas | 16:41 | |
| *** fnaval_ has joined #openstack-lbaas | 16:42 | |
| *** cpuga has quit IRC | 16:43 | |
| *** fnaval has quit IRC | 16:45 | |
| *** pcaruana has joined #openstack-lbaas | 16:45 | |
| *** armax has quit IRC | 16:49 | |
| *** gans has quit IRC | 16:50 | |
| *** yamamoto has quit IRC | 17:06 | |
| leitan | somebody ? rm_work xgerman | 17:12 |
| *** krypto has quit IRC | 17:12 | |
| *** aojea has joined #openstack-lbaas | 17:14 | |
| *** armax has joined #openstack-lbaas | 17:14 | |
| *** aojea has quit IRC | 17:18 | |
| rm_work | leitan: it has been a weekend / holiday in the IS | 17:22 |
| rm_work | *US | 17:22 |
| rm_work | and xgerman / johnsom were on vacation | 17:22 |
| rm_work | today is also a holiday | 17:23 |
| rm_work | err, today is the holiday | 17:23 |
| leitan | ooo | 17:52 |
| leitan | im sorry | 17:52 |
| leitan | btw, made octavia work, just today | 17:52 |
| leitan | the create_certificates.sh was messing with the client.pem at the end | 17:53 |
| leitan | rm_work: thanks, if you have it in mind, what i asked in the morning, let me know, if dont ill keep searching / wait | 17:54 |
| rm_work | leitan: hmm, curious what it was doing | 17:57 |
| rm_work | leitan: if there's a bug in that script (it is old and a little wonky, would love a ... better version of it) submit a bug report or patch! :) | 17:57 |
| rm_work | leitan: i really don't know about rescheduling in the agents, or really ANYTHING about neutron-lbaas unfortunately | 17:58 |
| rm_work | I'm focused almost 100% on octavia | 17:58 |
| leitan | rm_work: i imagine, thanks anyway, ill submit a bug report right away | 17:59 |
| leitan | i need to get into the oepsntack dev circuit so i can push patches | 17:59 |
| *** fnaval has joined #openstack-lbaas | 17:59 | |
| rm_work | heh yeah there is a guide... | 18:00 |
| rm_work | umm | 18:00 |
| rm_work | https://docs.openstack.org/infra/manual/developers.html | 18:01 |
| rm_work | I think that | 18:01 |
| rm_work | make sure you follow it exactly | 18:01 |
| rm_work | there are some places where I skipped a step before and things don't work right T_T | 18:02 |
| *** fnaval_ has quit IRC | 18:03 | |
| leitan | rm_work: will do thanks ! | 18:03 |
| *** yamamoto has joined #openstack-lbaas | 18:06 | |
| *** armax has quit IRC | 18:14 | |
| *** yamamoto has quit IRC | 18:19 | |
| *** fnaval_ has joined #openstack-lbaas | 18:45 | |
| *** fnaval has quit IRC | 18:49 | |
| *** aojea has joined #openstack-lbaas | 18:52 | |
| *** gcheresh has joined #openstack-lbaas | 19:23 | |
| *** fnaval has joined #openstack-lbaas | 19:29 | |
| *** fnaval_ has quit IRC | 19:33 | |
| leitan | rm_work: one quick question about octavia security groups, im managing them on the tenant service since i used this credentials to the amphora part, but when a user creates a listener, it will update the secgroups or the admin need to change the secgroup generated for the LB ? | 19:50 |
| *** fnaval_ has joined #openstack-lbaas | 19:53 | |
| *** fnaval has quit IRC | 19:57 | |
| johnsom | They are automatically updated | 19:59 |
| leitan | johnsom: lovely thanks ! and ssh only listens on the mgmt interface of the mgmt network on the image correct ? | 20:11 |
| *** pcaruana has quit IRC | 20:24 | |
| *** fnaval_ has quit IRC | 20:29 | |
| johnsom | Correct, and only if you specify a keypair in the config | 20:30 |
| leitan | johnsom: roger that, thanks !" | 20:35 |
| *** fnaval has joined #openstack-lbaas | 20:37 | |
| *** fnaval has quit IRC | 20:43 | |
| *** gcheresh has quit IRC | 20:44 | |
| nmagnezi | johnsom, o/ | 20:49 |
| *** aojea has quit IRC | 20:58 | |
| *** aojea has joined #openstack-lbaas | 20:59 | |
| *** JudeC has joined #openstack-lbaas | 20:59 | |
| *** JudeC has quit IRC | 21:00 | |
| *** aojea has quit IRC | 21:03 | |
| nmagnezi | leitan, re: your question about auto-reschedule | 21:03 |
| nmagnezi | leitan, to the best of my knowledge there was never a manual evacuation | 21:03 |
| nmagnezi | leitan, i recall seeing a patch for such a thing on either ocata or pike but i don't think it was eventually merged | 21:04 |
| leitan | nmagnezi: hi ! i know theres a feature (i guess since ocata) to auto-reschedule balancers from dead agents to alive ones, cause i saw the rewiew of the commit, but i wanted to know if previous to this, was a manual res-cheduling version | 21:04 |
| leitan | or if prior to acata if agent dies, all lb dies with hime | 21:05 |
| leitan | him | 21:05 |
| nmagnezi | leitan, to the best of my knowledge (i coded the auto-reschedule patch), prior to stable/ocata if the agent dies the loadbalancers will not get to other agents | 21:06 |
| nmagnezi | leitan, that being said | 21:06 |
| nmagnezi | leitan, keep in mind that the actual loadbalancing it done in the haproxy process | 21:06 |
| nmagnezi | leitan, meaning that if the agent died, as long as the haproxy process is up loadbalancing will work | 21:06 |
| nmagnezi | leitan, you will not be able to modify that loadbalancer with new configs | 21:07 |
| nmagnezi | as long at that agent is dead | 21:07 |
| leitan | nmagnezi: yes thats totally clear, i was thinking about a dead server that hosts the LB more than the agent itself | 21:07 |
| leitan | thinking about backporting that to mitaka, in the meantime i play with octavia | 21:07 |
| nmagnezi | leitan, so yeah.. that is not a recoverable scenario in mitaka (at least until you revive the server) | 21:08 |
| leitan | to offer an ""HA"" to the users in the meantime | 21:08 |
| leitan | nmagnezi: so great, that totally answer my question, thanks nmagnezi | 21:08 |
| nmagnezi | leitan, well, please keep in mind that an official backport to Mitaka (or even newton) will not get accepted as it is against the openstack backports policy | 21:09 |
| nmagnezi | leitan, obviously, on your machines you can modify the code to your liking but i just wanted you to know this | 21:09 |
| leitan | nmagnezi: yes, it will be an "internal" backport | 21:09 |
| nmagnezi | leitan, keep in mind that in order for this to work as an internal backport for mitaka you'll need to backport some things in Neutron as well | 21:10 |
| nmagnezi | leitan, i was actually writing a blog post about it, but didn't get the chance to finalize and publish it at the moment | 21:11 |
| leitan | nmagnezi: i didnt analized the complexity of the backport yet, im planning to do that tomorrow to measure the effort of doing that or launching octavia into production sooner | 21:11 |
| nmagnezi | leitan, anyhow, Octavia does work with HA in stable/mitaka : | 21:11 |
| leitan | nmagnezi: yes, works ACTIVE_STANDBY, im currently testing that setup, it took me a while to get it working, but i got my first ACTIVE load balancer today :) | 21:12 |
| nmagnezi | leitan, well done! | 21:12 |
| leitan | nmagnezi: haha thanks, reading the ansible-playbooks was the best choice i made to finally understand everything | 21:14 |
| nmagnezi | leitan, if there are any gaps you were able to fill by doing so, that's great. it would be awesome if you could file a bug against octavia with specific things missing from the docs so we can improve | 21:18 |
| leitan | nmagnezi: yes, i filled a lot already, and also an RFE to "simplify" the mgmt network part | 21:19 |
| *** catinthe_ has quit IRC | 21:20 | |
| *** leitan has quit IRC | 21:24 | |
| *** leitan has joined #openstack-lbaas | 21:25 | |
| *** circ-user-TeTGj has quit IRC | 21:33 | |
| openstackgerrit | Nir Magnezi proposed openstack/octavia master: The VIP NIC should not be hard-coded to eth1 https://review.openstack.org/468963 | 21:52 |
| openstackgerrit | Nir Magnezi proposed openstack/octavia master: The VIP NIC should not be hard-coded to eth1 https://review.openstack.org/468963 | 22:49 |
| openstackgerrit | Nir Magnezi proposed openstack/octavia master: The VIP NIC should not be hard-coded to eth1 https://review.openstack.org/468963 | 23:20 |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!