| *** longkb has joined #openstack-lbaas | 00:41 | |
| *** yamamoto has joined #openstack-lbaas | 01:48 | |
| *** yamamoto has quit IRC | 01:51 | |
| *** SumitNaiksatam has joined #openstack-lbaas | 01:56 | |
| *** hongbin has joined #openstack-lbaas | 02:49 | |
| bzhao__ | Hi, @Team, may I add our project into stackalytics? I'm so sorry to bring it up as my company wants to use stackalytics records for showing to customers the contribution status, because company treats Octavia as the standard. But now, I don't see our project in stackalytics repo, it just gets the neutron official group from openstack/governance, that means stackalytics collect the contribution or merged patches based | 03:10 |
|---|---|---|
| bzhao__ | on its first post time, not the merge time. So the proposal is changing the collected merged patches based on the merged time. I'm OK if we think that's not necessary. Just want to collect some feedbacks from our @Team members. If we are disagree, please feel free to say "No". Thanks. Then, the slogan again, Fight for Octavia! ;-) | 03:10 |
| johnsom | bzhao__: we are there as “Octavia official” | 03:12 |
| johnsom | http://stackalytics.com/?module=octavia-group | 03:15 |
| johnsom | The merges are listed as “commits” | 03:16 |
| johnsom | Note though, stackalytics can get behind, it maybe tomorrow before UDP shows up there | 03:17 |
| bzhao__ | johnsom: Thanks, let me think a while how to suggest to marketing. Yeah, thanks johnsom for explaining, just a marketing people catch me about this thing this morning. | 03:19 |
| johnsom | Ok, let me know if I don’t understand your request | 03:20 |
| johnsom | bzhao__: Is that what you were looking for or something else? I am concerned I don’t know the request | 03:29 |
| bzhao__ | johnsom: They want the advantage records. ;-( . But some Metrics had already matched, except review, commits, personal effort. Hmm, I think I can suggest them about other metrics. All things are that want to make customers attention. | 03:30 |
| bzhao__ | johnsom: Sorry, just thinking and searching something. | 03:30 |
| bzhao__ | johnsom: Yeah, that's it. | 03:32 |
| bzhao__ | johnsom: That's I was looking for. | 03:32 |
| johnsom | bzhao__: ok, our Stackalytics should be setup the same as Neutron official, just for our project and repositories. | 03:33 |
| bzhao__ | johnsom: Yeah, I search our octavia in Neutron official in previous time. ;-). | 03:35 |
| johnsom | bzhao__: yeah, we are not a neutron subproject anymore. We are a top level project like neutron | 03:36 |
| *** ramishra has joined #openstack-lbaas | 03:36 | |
| *** abaindur has joined #openstack-lbaas | 03:38 | |
| *** abaindur has quit IRC | 03:38 | |
| bzhao__ | johnsom: Correct. Ha, that's my mistake. OK, I remember that you suggests that the udp healthmonitor is good to using notify_up/notify_down and store the member status in a specific file. Then I wish that you could give the kind advice about whether I need to refactor that part, from this point, I just want to make it better and make our team member comfortable when see that part code. ;-). | 03:39 |
| *** abaindur has joined #openstack-lbaas | 03:39 | |
| *** hongbin has quit IRC | 03:41 | |
| johnsom | bzhao__: your fix from the DRAIN issue works for me. | 03:41 |
| bzhao__ | johnsom: And the future features, maybe I will try to discuss with you first, then implement as you and our team members suggests. So I'm afraid that you guys may kick me if feel that bad guy(me) make you uncomfortable. ;-) | 03:42 |
| johnsom | bzhao__: I think that is a good fix at least for now. If we need we can change it later. | 03:42 |
| bzhao__ | johnsom: ok. That sounds good. ;-) | 03:43 |
| johnsom | bzhao__: Ha no, we are nice people. We just want to help and make it easier | 03:43 |
| johnsom | bzhao__: most features should go through a specification or an RFE if it is a small patch. This helps the team know it is being worked on and can give comments | 03:44 |
| bzhao__ | johnsom: Yeah. I'm prety sure that all of you and team members are very nice peoples. Sorry for that, it just a joke...;-) | 03:45 |
| johnsom | bzhao__: ok, I hope we get to continue to work together on Octavia | 03:46 |
| bzhao__ | johnsom: Right. I will follow our guideline. | 03:46 |
| bzhao__ | johnsom: It must be a long time. ;-) | 03:46 |
| johnsom | Ok, I am signing off for the night. Have a good day with the satisfaction of a job well done | 03:47 |
| bzhao__ | johnsom: Thanks vey much for your help. Have a nice sleep . ;-) | 03:48 |
| abaindur | johnsom: Hi, i'm hitting some cryptic errors while trying just the first, basic loadbalancer create operation. We are hitting this both with the CLI and a POST curl api call. are you free to help? | 03:54 |
| abaindur | All we see in logs is: | 03:55 |
| abaindur | 2018-08-09 03:43:17.666 17329 DEBUG wsme.api [req-d41e4193-be0a-42cb-9bcd-2230a7c920c9 - fce11e73824a4f1a8f6ffaed87b3a16a - default default] Client-side error: Invalid input for field/attribute load_balancer. Value: ''. unable to convert to LoadBalancerRootPOST. Error: __init__() takes exactly 1 argument (2 given) format_exception /opt/pf9/octavia/lib/python2.7/site-packages/wsme/api.py:222 | 03:55 |
| abaindur | the curl returns an HTTP 400 | 03:56 |
| abaindur | with the same error string: {"debuginfo": null, "faultcode": "Client", "faultstring": "Invalid input for field/attribute load_balancer. Value: ''. unable to convert to LoadBalancerRootPOST. Error: __init__() takes exactly 1 argument (2 given)"} | 03:56 |
| johnsom | abaindur: it is late for me, but what is the URL you are posting to and what is the json content of the POST? | 03:58 |
| abaindur | http://paste.openstack.org/show/727690/ | 03:58 |
| abaindur | I only did "openstack loadbalancer create --name lb1 --vip-subnet-id b6a8eb74-9197-4f7d-8280-68b8da64deaa" | 03:59 |
| abaindur | The curl i tried, I got manually from the --debug of the CLI command. The json looks to be correct, looking at the API doc | 03:59 |
| abaindur | np, if you're sleepy i'll ping you tomorrow | 04:00 |
| bzhao__ | abaindur: I think you maybe add the lb name | 04:02 |
| abaindur | the CLI command is what i got straight from the lbaas doc | 04:02 |
| bzhao__ | abaindur: Oh, sorry . | 04:02 |
| johnsom | I am out of office for a few days, so let me look | 04:03 |
| bzhao__ | let me recheck the log | 04:03 |
| abaindur | '{"loadbalancer": {"vip_subnet_id": "b6a8eb74-9197-4f7d-8280-68b8da64deaa", "name": "lb1", "admin_state_up": true}}' | 04:03 |
| abaindur | is the -d payload in POST | 04:04 |
| johnsom | Yeah, that looks fine | 04:04 |
| johnsom | I don't understand where you are getting load_balancer from | 04:04 |
| johnsom | Which version of Octavia are you using? | 04:04 |
| abaindur | Queens | 04:04 |
| abaindur | we have the endpoint service type as "load-balancer" in Keystone | 04:05 |
| abaindur | openstack endpoint list | 04:05 |
| abaindur | | d3ffec8b5af6492cbc8e2a0ae1e82fc4 | RegionOne | octavia | load-balancer | True | internal | https://test-du-arjun-oct-roo-966.platform9.horse/octavia | | 04:05 |
| abaindur | | d5ed0279efb841069ee6f7e7647b4e12 | RegionOne | octavia | load-balancer | True | admin | https://test-du-arjun-oct-roo-966.platform9.horse/octavia | 04:05 |
| abaindur | is that correct? | 04:06 |
| johnsom | Yes, that is correct | 04:07 |
| abaindur | hmm, and our clients are installed w/ queens upper constraints versions | 04:08 |
| johnsom | HTTP/1.1 400 Bad Request | 04:11 |
| johnsom | Server: nginx/1.14.0 | 04:11 |
| johnsom | Why does it say the API endpoint is nginx? | 04:11 |
| johnsom | We do NOT use nginx | 04:11 |
| johnsom | Is it going to some kind of proxy server? | 04:12 |
| abaindur | We have an nginx reverse proxy that redirects /octavia to 127.0.0.1:9876 | 04:12 |
| abaindur | We have this for all services | 04:12 |
| johnsom | Yeah, the client doesn't matter as you are going straight to the API with curl | 04:13 |
| johnsom | Did you install from pip/git? | 04:13 |
| abaindur | Octavia? | 04:14 |
| johnsom | I am pretty sure the API process is looking for an load_balancer root tag on the json, but we never released that. it has always been loadbalancer | 04:14 |
| johnsom | Yeah | 04:14 |
| abaindur | No, we installed from source | 04:14 |
| abaindur | in our own venv | 04:15 |
| johnsom | Can you do a git log in the directory so we can see the commit sha | 04:15 |
| johnsom | ? | 04:15 |
| abaindur | well rather, packaged it in an rpm, then install rpm | 04:15 |
| johnsom | There may have been some commits during ocata or pike that had that, but we never had an actual release with that as the root tag. Certainly not queens | 04:17 |
| abaindur | commit 99319c24080bdfa51f31cb29e40381670b05e697 | 04:17 |
| abaindur | Merge "Allow using spaces for primary common name in SSL certificates" into stable/queens | 04:17 |
| abaindur | hey wait a sec | 04:18 |
| johnsom | That SHA looks ok | 04:19 |
| abaindur | when i curled directly to https://127.0.0.1:9876 I got an curl: (35) SSL received a record that exceeded the maximum permissible length. | 04:19 |
| abaindur | so i made it just http | 04:19 |
| abaindur | And it went thru | 04:19 |
| johnsom | Ok, so I think something is wrong with your nginx config and it's going somewhere is should not , or ???? Actually it is a pretty strange error | 04:20 |
| abaindur | it didnt work with https | 04:21 |
| abaindur | i'm reading in the docs, "As such, there is presently no authentication required to use the Octavia API" | 04:22 |
| johnsom | That is the Octavia v1 API | 04:22 |
| johnsom | v2 has full RBAC | 04:22 |
| abaindur | https://serverfault.com/questions/379772/how-do-i-resolve-the-error-ssl-received-a-record-that-exceeded-the-maximum-perm | 04:23 |
| johnsom | You would have had to configure the Octavia API to use TLS/SSL for HTTPS. As a wsgi application we usually rely on the web server to deal with TLS/SSL. If you are running in non-wsgi mode you would have had to added custom config to add TLS/SSL to the simple_server | 04:25 |
| abaindur | our nginx proxy actually redirects to proxy_pass http://127.0.0.1:9876/; | 04:25 |
| abaindur | same as what works for our other services | 04:25 |
| johnsom | Check the API log file. What is the path the request is coming into Octavia as? | 04:26 |
| abaindur | thats the thing, we dont see anything in the nginx log files | 04:27 |
| abaindur | except: | 04:27 |
| abaindur | 2018-08-09 03:57:56.911 17329 DEBUG wsme.api [req-10739a7d-fc26-4efa-9e71-bf3f44dd073a - fce11e73824a4f1a8f6ffaed87b3a16a - default default] Client-side error: Invalid input for field/attribute load_balancer. Value: ''. unable to convert to LoadBalancerRootPOST. Error: __init__() takes exactly 1 argument (2 given) format_exception /opt/pf9/octavia/lib/python2.7/site-packages/wsme/api.py:222 | 04:27 |
| johnsom | No not nginx, the octavia API process, the process on port 9876 | 04:27 |
| johnsom | that nginx is forwarding to | 04:27 |
| abaindur | sorry i meant octavia | 04:28 |
| abaindur | Thats all in octavia | 04:28 |
| abaindur | nginx is showing: 127.0.0.1 - - [09/Aug/2018:03:57:48 +0000] "POST /octavia/v2.0/lbaas/loadbalancers HTTP/1.1" 400 400 - 216 "-" "curl/7.29.0" "-" to: 127.0.0.1:9876 0.009 0.010 8080 43564 - | 04:28 |
| johnsom | Yeah, well, if it is passing /octavia through to the backend it will fail. Without the actual octavia API log I'm not sure. I don't know how your nginx is configured | 04:29 |
| johnsom | We don't have a /octavia path | 04:30 |
| bzhao__ | This may raise by octavia api layer. It had reach the octavia I think. They regitered the endpoint with a /octavia path. | 04:32 |
| bbbbzhao_ | https://www.irccloud.com/pastebin/5t1y3dZA/ | 04:32 |
| johnsom | Which is ok if their nginx layer strips that when it passes it to Octavia | 04:32 |
| abaindur | http://paste.openstack.org/show/727694/ | 04:32 |
| abaindur | for comparison, our nginx locations to /designate and /neutron endpoints, which work 100% | 04:33 |
| abaindur | But ok... i suppose its a problem on our end somewehre | 04:34 |
| abaindur | thanks for the help | 04:34 |
| johnsom | Let me go read up on nginx config. | 04:34 |
| abaindur | reason being we have that, we dont want to expose that port on public ip | 04:36 |
| johnsom | Yes, we don't use ports either | 04:37 |
| johnsom | On the host with your octavia API process, can you do a "netstat -anp" and look for the process with the 9876 port? | 04:38 |
| johnsom | We need to find the actual octavia API process and it's log | 04:38 |
| abaindur | yea, i did a GET | 04:38 |
| abaindur | it does not return an error, but it returns an empty | 04:38 |
| abaindur | [root@test-du-arjun-oct-roo-966.platform9.horse arjun(admin)]# openstack loadbalancer list | 04:38 |
| abaindur | and nothing else | 04:38 |
| abaindur | I dont see it being logged in octavi-api.log | 04:39 |
| johnsom | It looks like you are just proxying with nginx, so something else is hosting the wsgi API | 04:39 |
| abaindur | tcp 0 0 127.0.0.1:9876 0.0.0.0:* LISTEN 17329/python | 04:39 |
| abaindur | 17329 ? S 0:05 /opt/pf9/octavia/bin/python /opt/pf9/octavia/bin/octavia-api --config-file /etc/octavia/octavia.conf | 04:39 |
| abaindur | [root@test-du-arjun-oct-roo-966.platform9.horse arjun(admin)]# ps -ef | grep octavia | 04:40 |
| abaindur | root 3797 17580 0 04:40 pts/0 00:00:00 grep --color=auto octavia | 04:40 |
| abaindur | octavia 17329 1 0 03:39 ? 00:00:05 /opt/pf9/octavia/bin/python /opt/pf9/octavia/bin/octavia-api --config-file /etc/octavia/octavia.conf | 04:40 |
| abaindur | It is indeed the correct proc | 04:40 |
| johnsom | Ok, so it is running the "simple_sever". Is it started with systemd? | 04:40 |
| abaindur | yes | 04:40 |
| johnsom | There is going to be a journalctl for the system service that starts it. That will likely contain the actual log given your setup | 04:41 |
| abaindur | Are we missing some parameter for it to be logging? | 04:41 |
| johnsom | No, you would have had to add some config to disable it | 04:42 |
| abaindur | we see some logs, like API server, 2018-08-09 03:39:32.384 17329 INFO octavia.cmd.api [-] Starting API server on 127.0.0.1:9876 | 04:42 |
| abaindur | log_dir = is set correctly in the .conf file as wel | 04:42 |
| johnsom | There you go, that should be the place | 04:42 |
| abaindur | I meant I see that in the existing octavi-api.log | 04:42 |
| abaindur | not journalctl | 04:42 |
| johnsom | You don't see your requests in there? Even when you curl direct to the port? | 04:42 |
| abaindur | after that, I only see the error i pasted before | 04:43 |
| *** irenab has joined #openstack-lbaas | 04:44 | |
| abaindur | No | 04:46 |
| bbbbzhao_ | https://www.irccloud.com/pastebin/V3fJF1tz/ | 04:46 |
| abaindur | I am tailing it via journalctl -f | 04:46 |
| johnsom | Ok, I am pretty sure it is a nginx config issue. I think you need a rewrite rule inside the location block that strips off the /octavia from the URL path. | 04:48 |
| johnsom | I don't know where your request logs are going, but I'm pretty sure it's passing /octavia to the API simple server and that will not work and will be rejected. | 04:49 |
| abaindur | well, even when it worked, when i made the req directly to http://127.0.0.1:9876/, i did not see any API req being logged | 04:51 |
| abaindur | I only saw: | 04:51 |
| abaindur | 2018-08-09 03:59:51.027 17329 DEBUG oslo_db.sqlalchemy.engines [req-ff74941b-8cc7-4075-859d-c47639e8a6a7 - fce11e73824a4f1a8f6ffaed87b3a16a - default default] MySQL server mode set to STRICT_TRANS_TABLES,STRICT_ALL_TABLES,NO_ZERO_IN_DATE,NO_ZERO_DATE,ERROR_FOR_DIVISION_BY_ZERO,TRADITIONAL,NO_AUTO_CREATE_USER,NO_ENGINE_SUBSTITUTION _check_effective_sql_mode /opt/pf9/octavia/lib/python2.7/site-packages/ | 04:51 |
| abaindur | oslo_db/sqlalchemy/engines.py:290 | 04:51 |
| abaindur | 2018-08-09 04:14:05.831 17329 DEBUG octavia.network.drivers.noop_driver.driver [req-73570b10-4ce6-4bd1-8655-d4a4ee86f912 - fce11e73824a4f1a8f6ffaed87b3a16a - default default] Subnet NoopManager no-op, get_subnet subnet_id b6a8eb74-9197-4f7d-8280-68b8da64deaa get_subnet /opt/pf9/octavia/lib/python2.7/site-packages/octavia/network/drivers/noop_driver/driver.py:136 | 04:51 |
| abaindur | 2018-08-09 04:14:05.832 17329 DEBUG octavia.db.repositories [req-73570b10-4ce6-4bd1-8655-d4a4ee86f912 - fce11e73824a4f1a8f6ffaed87b3a16a - default default] Checking quota for project: fce11e73824a4f1a8f6ffaed87b3a16a object: <class 'octavia.common.data_models.LoadBalancer'> check_quota_met /opt/pf9/octavia/lib/python2.7/site-packages/octavia/db/repositories.py:302 | 04:51 |
| abaindur | 2018-08-09 04:14:05.950 17329 DEBUG octavia.network.drivers.noop_driver.driver [req-73570b10-4ce6-4bd1-8655-d4a4ee86f912 - fce11e73824a4f1a8f6ffaed87b3a16a - default default] Network NoopManager no-op, allocate_vip loadbalancer <octavia.common.data_models.LoadBalancer object at 0x7facbe87c8d0> allocate_vip /opt/pf9/octavia/lib/python2.7/site-packages/octavia/network/drivers/noop_driver/driver.py:32 | 04:51 |
| abaindur | 2018-08-09 04:14:05.956 17329 INFO octavia.api.v2.controllers.load_balancer [req-73570b10-4ce6-4bd1-8655-d4a4ee86f912 - fce11e73824a4f1a8f6ffaed87b3a16a - default default] Sending created Load Balancer 7c9f31a7-1b23-4c96-a2ac-018e58a1721f to the handler | 04:51 |
| johnsom | It did work correctly though right? it did create the BL? | 04:51 |
| johnsom | Ok, those are ok logs | 04:52 |
| johnsom | Though you are running the noop driver, which won't actually do anything | 04:52 |
| abaindur | Dont see any LBs | 04:53 |
| johnsom | So, maybe when not running under wsgi that "simple_server" will only log the request in debug | 04:53 |
| abaindur | dont see any logs in worker, besides the 2018-08-07 23:45:02.044 11763 INFO octavia.controller.queue.consumer [-] Starting consumer... | 04:53 |
| johnsom | Yeah, with no-op drivers configured you won't get any load balnacer. It is all just ghosts in the DB. We use the noop drivers for testing | 04:54 |
| abaindur | oh, is the drivers configured for api service? | 04:54 |
| abaindur | ok, we have that on the config file where our worker/healthmanager/housekeeper resides | 04:54 |
| johnsom | Yes, some of those drivers matter for the API | 04:54 |
| bbbbzhao_ | api_handler = queue_producer | 04:55 |
| abaindur | what do you mean by simple_server? | 04:55 |
| johnsom | Well, you are running the API as a standalone application, which uses the python "simple_server", it's not very scalable or production quality. All(most) OpenStack services are intended to be run as wgsi applications per the community goal in Pike: https://governance.openstack.org/tc/goals/pike/deploy-api-in-wsgi.html | 04:56 |
| johnsom | Yeah, the queue_producer is one, but networking is required as well to verify the subnets, etc. the user passes in | 04:59 |
| johnsom | It's mostly for validation calls | 05:00 |
| johnsom | Though I think the VIP port may be created in API as well | 05:01 |
| abaindur | Our [api] section is left unfilled (at defaults) | 05:01 |
| ramishra | It seems octavia has broken heat gate, http://logs.openstack.org/44/584544/2/check/heat-functional-convg-mysql-lbaasv2-py35/52f40b7/logs/screen-o-cw.txt.gz#_Aug_08_19_59_12_483217 I assume it's the udp support merged yesterday | 05:01 |
| johnsom | This is the code for the standalone API process: https://github.com/openstack/octavia/blob/master/octavia/cmd/api.py#L16 | 05:01 |
| bbbbzhao_ | not sure, they are same with previous release https://www.irccloud.com/pastebin/lhFQzm3h/ | 05:02 |
| ramishra | johnsom: Hi!, any idea what's gone wrong | 05:02 |
| johnsom | ramishra This is an issue with the patch merge order. It should be resolved tomorrow when everything merges and updates | 05:02 |
| ramishra | Hmm.. how do we merge changes till tomorrow, it's RC1 week. do you suggest disabling the tests is the only option? | 05:03 |
| ramishra | johnsom: ^^ | 05:03 |
| johnsom | ramishra Are you using the tarballs.openstack.org testing image? | 05:03 |
| ramishra | johnsom: yes | 05:03 |
| johnsom | ramishra Yeah, that is the issue. Let me look at when that periodic job runs | 05:04 |
| johnsom | ramishra Basically that pulled the code mid-merge of a chain of patches and has bad code. I think it will update and build a new image in about an hour and a half | 05:05 |
| bbbbzhao_ | {'error': '400 Bad Request: Invalid protocol information for Listener', 'http_code': 400} | 05:06 |
| johnsom | ramishra We saw the same with some of our tests. That image is bad | 05:06 |
| ramishra | johnsom: OK, once I disable the tests, bringing it back would require plenty of convincing to do, wich I want to avoid | 05:06 |
| bbbbzhao_ | seem using the old amp image. not the full merge one | 05:06 |
| johnsom | Yeah, watch this folder: http://tarballs.openstack.org/octavia/test-images/ the images should update in about an hour and a half, then recheck | 05:07 |
| johnsom | bbbbzhao_ Patch 2 merged, the image built, then patch followup merged. so that image is missing followup | 05:08 |
| ramishra | johnsom: Thanks, will check in an hour or so.. I assume we would try and ensure that we avoid this kind of issue in the future just before release:) | 05:08 |
| bbbbzhao_ | johnsom: Yeah, the image miss the followup, as we remove the protocol check in that patch. | 05:09 |
| johnsom | ramishra Yeah it was unfortunate. | 05:09 |
| bbbbzhao_ | ramishra: Sorry for that. | 05:10 |
| ramishra | johnsom, bbbbzhao_ np:), thanks! | 05:10 |
| *** pcaruana has joined #openstack-lbaas | 06:38 | |
| openstackgerrit | OpenStack Proposal Bot proposed openstack/octavia master: Imported Translations from Zanata https://review.openstack.org/590128 | 06:46 |
| *** luksky has joined #openstack-lbaas | 07:03 | |
| *** ispp has joined #openstack-lbaas | 07:06 | |
| *** velizarx has joined #openstack-lbaas | 07:09 | |
| *** yamamoto has joined #openstack-lbaas | 07:11 | |
| *** rcernin has quit IRC | 07:16 | |
| *** yamamoto has quit IRC | 07:22 | |
| abaindur | anyone know what permissions the certs have to be? | 07:26 |
| abaindur | on worker, I see following error when it tries to open the ca_certificate file | 07:27 |
| abaindur | CertificateGenerationException: Could not sign the certificate request: Failed to load CA Certificate /root/octcerts/ca_01.pem. | 07:27 |
| abaindur | I am able to open the file myself in python: | 07:27 |
| abaindur | >>> ca_cert = open('/root/octcerts/ca_01.pem', 'rb').read() | 07:27 |
| *** nmagnezi_ has joined #openstack-lbaas | 07:39 | |
| openstackgerrit | OpenStack Proposal Bot proposed openstack/neutron-lbaas master: Imported Translations from Zanata https://review.openstack.org/590173 | 07:44 |
| *** kobis1 has joined #openstack-lbaas | 07:46 | |
| *** velizarx has quit IRC | 07:46 | |
| *** velizarx has joined #openstack-lbaas | 07:53 | |
| *** nmagnezi_ has quit IRC | 07:57 | |
| *** nmagnezi has joined #openstack-lbaas | 07:57 | |
| *** dulek has joined #openstack-lbaas | 08:21 | |
| *** aojea has joined #openstack-lbaas | 08:22 | |
| *** KeithMnemonic has quit IRC | 08:23 | |
| *** ktibi has joined #openstack-lbaas | 08:37 | |
| *** celebdor has joined #openstack-lbaas | 08:38 | |
| *** abaindur has quit IRC | 08:45 | |
| *** aojea has quit IRC | 08:54 | |
| *** ktibi has quit IRC | 09:00 | |
| *** ktibi has joined #openstack-lbaas | 09:01 | |
| *** ispp has quit IRC | 09:10 | |
| *** salmankhan has joined #openstack-lbaas | 09:14 | |
| *** luksky has quit IRC | 09:56 | |
| *** kobis1 has quit IRC | 10:09 | |
| openstackgerrit | OpenStack Release Bot proposed openstack/neutron-lbaas master: Update reno for stable/rocky https://review.openstack.org/590223 | 10:27 |
| openstackgerrit | OpenStack Release Bot proposed openstack/octavia-dashboard master: Update reno for stable/rocky https://review.openstack.org/590226 | 10:27 |
| openstackgerrit | OpenStack Release Bot proposed openstack/octavia master: Update reno for stable/rocky https://review.openstack.org/590232 | 10:27 |
| *** kobis1 has joined #openstack-lbaas | 10:34 | |
| *** luksky has joined #openstack-lbaas | 10:35 | |
| *** frickler has left #openstack-lbaas | 11:08 | |
| *** aojea has joined #openstack-lbaas | 11:20 | |
| *** aojea has quit IRC | 11:33 | |
| *** longkb has quit IRC | 11:40 | |
| *** aojea has joined #openstack-lbaas | 11:45 | |
| *** yamamoto has joined #openstack-lbaas | 11:57 | |
| *** aojea has quit IRC | 11:58 | |
| *** amuller has joined #openstack-lbaas | 12:00 | |
| *** velizarx has quit IRC | 12:13 | |
| *** velizarx has joined #openstack-lbaas | 12:15 | |
| *** oanson has joined #openstack-lbaas | 12:42 | |
| *** ianychoi has quit IRC | 12:49 | |
| *** ramishra has quit IRC | 12:59 | |
| *** velizarx has quit IRC | 13:02 | |
| *** velizarx has joined #openstack-lbaas | 13:03 | |
| *** ramishra has joined #openstack-lbaas | 13:08 | |
| *** yamamoto has quit IRC | 13:31 | |
| openstackgerrit | German Eichberger proposed openstack/neutron-lbaas master: Fix proxy extension for neutron RBAC https://review.openstack.org/554004 | 13:35 |
| openstackgerrit | German Eichberger proposed openstack/neutron-lbaas master: Gate API test for the lbaasv2-proxy plugin https://review.openstack.org/539350 | 13:35 |
| xgerman_ | ^^^ this needs to land in Rocky. Working on that for more than a year… that’s for review velocity in this place | 13:36 |
| openstackgerrit | German Eichberger proposed openstack/neutron-lbaas master: Gate API test for the lbaasv2-proxy plugin https://review.openstack.org/539350 | 13:38 |
| *** amotoki_ is now known as amotoki | 14:01 | |
| johnsom | That patch is blocked by a dependency | 14:04 |
| nmagnezi | johnsom, o/ ; maybe you'll want this in Rocky https://review.openstack.org/#/c/589408/ | 14:09 |
| *** KeithMnemonic has joined #openstack-lbaas | 14:19 | |
| openstackgerrit | Merged openstack/neutron-lbaas master: Update reno for stable/rocky https://review.openstack.org/590223 | 14:24 |
| openstackgerrit | Merged openstack/octavia master: Update reno for stable/rocky https://review.openstack.org/590232 | 14:26 |
| openstackgerrit | Merged openstack/octavia-dashboard master: Update reno for stable/rocky https://review.openstack.org/590226 | 14:33 |
| cgoncalves | xgerman_, could you help me refreshing my mind please :) can one recover a LB that was once operational from an ERROR after failover failed? | 14:39 |
| xgerman_ | if you have the amp API maybe; usually ERROR is final and all you can do is delete | 14:40 |
| xgerman_ | also I have a ton of LB in ERROR which still work… look at operational status if it’s ONLINE it’s still working | 14:40 |
| *** hongbin has joined #openstack-lbaas | 14:41 | |
| xgerman_ | but we should have someway to self heal those LB | 14:41 |
| cgoncalves | xgerman_, what do you mean by having the amp API? | 14:44 |
| xgerman_ | the amp API allows you to fix amps by failing them over. | 14:44 |
| xgerman_ | I don’t recall if you run an lb failover if it will go back into ACTIVE or just not run at all | 14:45 |
| cgoncalves | scenario is that the cluster was rebooted and the HM started to failover amps because computes were still not up. failover then failed due to black of compute capacity | 14:45 |
| cgoncalves | ah, that amp API, right | 14:46 |
| cgoncalves | s/black/lack/ | 14:46 |
| xgerman_ | yeah, I have fun adventures when nodes are being rebooted, too | 14:52 |
| xgerman_ | I also added a self healing topic for Denver. My ops people are also not thrilled with our frequent going into ERROR when something goes wrong | 14:53 |
| xgerman_ | to be fair it’s all nova and neutron throwing errors and we are stopping to not make things worse | 14:54 |
| xgerman_ | but neutron and nova throw too many errors - so something’s gotta give ;-) | 14:54 |
| *** aojea has joined #openstack-lbaas | 14:55 | |
| cgoncalves | octavia could hint users to the compute or network services :) | 14:56 |
| cgoncalves | it would be so handy to operators. instead of start looking into octavia logs, they could with great confidence immediately start looking at nova or neutron | 14:57 |
| xgerman_ | yep, we should have some description on the LB | 14:57 |
| xgerman_ | but then we also should have some “repair” command once nova/neutron are happy again | 14:58 |
| cgoncalves | why not invoking failover again? | 15:01 |
| xgerman_ | that might do it | 15:02 |
| *** Swami has joined #openstack-lbaas | 15:03 | |
| *** ivve has quit IRC | 15:03 | |
| *** kobis1 has quit IRC | 15:05 | |
| *** aojea has quit IRC | 15:06 | |
| *** aojea has joined #openstack-lbaas | 15:07 | |
| *** celebdor has quit IRC | 15:08 | |
| *** fnaval has joined #openstack-lbaas | 15:16 | |
| *** celebdor has joined #openstack-lbaas | 15:32 | |
| *** phuoc has quit IRC | 15:32 | |
| *** SumitNaiksatam has quit IRC | 15:59 | |
| *** velizarx has quit IRC | 16:16 | |
| *** hongbin has quit IRC | 16:37 | |
| *** ktibi has quit IRC | 16:37 | |
| *** Swami has quit IRC | 16:37 | |
| *** aojea has quit IRC | 16:51 | |
| *** salmankhan has quit IRC | 16:58 | |
| *** phuoc has joined #openstack-lbaas | 17:01 | |
| *** ramishra has quit IRC | 17:03 | |
| *** luksky has quit IRC | 17:27 | |
| *** SumitNaiksatam has joined #openstack-lbaas | 18:00 | |
| openstackgerrit | German Eichberger proposed openstack/neutron-lbaas master: Gate API test for the lbaasv2-proxy plugin https://review.openstack.org/539350 | 18:11 |
| *** celebdor has quit IRC | 18:14 | |
| jiteka | Hello I got few question for you :) | 18:19 |
| jiteka | 1. Is ubuntu bionic supported for amphora glance image in Octavia rocky ? | 18:19 |
| jiteka | Last time I tried in queens diskimage-builder script failed on fetching files from cloud-images.ubuntu.com | 18:19 |
| *** colby_ has joined #openstack-lbaas | 18:29 | |
| xgerman_ | I have seen a bionic gate fly by. So yes | 18:30 |
| xgerman_ | also the latest diskimage builder will support it. | 18:30 |
| jiteka | thanks xgerman_ :) | 18:39 |
| jiteka | and second question | 18:42 |
| jiteka | 2. I saw that Flavors support is targeted for Rocky but is it plan to backport it to Queens ? | 18:42 |
| jiteka | If not, with active/standby topology how easy/hard it is to go from single-flavor to multi-flavor | 18:42 |
| jiteka | when offering multiple class of service for LB going from : amphora VM running among all other users VMs with CPU and RAM overcomitting to dedicated host-aggregate (or cells) with faster hardware with better NICs | 18:49 |
| jiteka | and what is the difference between "flavors" targeted for Octavia Version 3.0 | 18:54 |
| jiteka | and Flavor support targeted for Considerations for Octavia 3.0+ ? | 18:54 |
| jiteka | Is it the same feature but just also in "beyond rocky" if not ready for rocky or something else ? | 18:54 |
| xgerman_ | 2) Flavors will be Stein - no plans to backport | 18:57 |
| xgerman_ | ^^ jiteka | 18:58 |
| xgerman_ | there are no firm plans for Octavia 3.0 — not sure if this is even a thing | 18:58 |
| *** luksky has joined #openstack-lbaas | 18:58 | |
| jiteka | thanks xgerman_ | 19:02 |
| jiteka | will look for queens to rocky upgrade procedure then | 19:02 |
| xgerman_ | well, we still have to code flavors ;-) | 19:02 |
| jiteka | and for bionic I confirm that it works | 19:02 |
| jiteka | just got an issue with squashfs on centos : http://paste.openstack.org/show/727762/ | 19:04 |
| jiteka | last image I created with diskimage-builder didn't complained about squashfs-tools not being installed | 19:04 |
| jiteka | ohh | 19:05 |
| jiteka | I though that you were using stein | 19:05 |
| jiteka | to say rocky | 19:05 |
| jiteka | [...] Beer stein (/ˈstaɪn/ STYNE), or simply stein, is an English neologism for either traditional beer mugs made out of stoneware, or specifically ornamental beer mugs that are usually sold as souvenirs or collectibles. In German, the word stein means stone and is not used to refer to a beverage container. [...] | 19:06 |
| jiteka | I tought it was a joke rocky/rock/stein | 19:06 |
| xgerman_ | Stein will start in a few week ;- | 19:06 |
| xgerman_ | lol, no, it’s the name for the next release - | 19:06 |
| jiteka | yes I wasn't aware of the you release codename | 19:07 |
| jiteka | that would help me to understand | 19:07 |
| jiteka | s/you/new | 19:07 |
| *** SumitNaiksatam has left #openstack-lbaas | 19:25 | |
| *** pcaruana has quit IRC | 19:33 | |
| *** amuller has quit IRC | 19:40 | |
| *** aojea has joined #openstack-lbaas | 19:47 | |
| *** abaindur has joined #openstack-lbaas | 19:54 | |
| *** abaindur has quit IRC | 19:54 | |
| *** abaindur has joined #openstack-lbaas | 19:55 | |
| openstackgerrit | German Eichberger proposed openstack/neutron-lbaas master: Gate API test for the lbaasv2-proxy plugin https://review.openstack.org/539350 | 19:56 |
| *** abaindur has quit IRC | 20:31 | |
| *** abaindur has joined #openstack-lbaas | 20:31 | |
| abaindur | Am I right in assuming that by default octavia will only wait 100 seconds for an amphora VM to become active? | 21:02 |
| abaindur | # amp_active_retries = 10 | 21:02 |
| abaindur | # amp_active_wait_sec = 10 | 21:02 |
| abaindur | Is there any downside increasing this to something mcuh larger, like 5 minutes? | 21:03 |
| abaindur | and might it be better to increase the retries, or the wait duration? | 21:05 |
| *** dosaboy has joined #openstack-lbaas | 21:14 | |
| abaindur | Also... we would like the amphora VMs to come up in the same tenant as the user who creates the LB | 21:20 |
| abaindur | we noticed they are instead coming up in services tenant. Is there any way to do this? | 21:21 |
| *** yamamoto has joined #openstack-lbaas | 22:04 | |
| *** luksky has quit IRC | 22:07 | |
| *** rcernin has joined #openstack-lbaas | 22:09 | |
| *** yamamoto has quit IRC | 22:21 | |
| *** aojea has quit IRC | 22:46 | |
| *** fnaval has quit IRC | 22:54 | |
| *** abaindur has quit IRC | 23:50 | |
| *** abaindur has joined #openstack-lbaas | 23:51 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!