| johnsom | rm_work I am also seeing a bunch of grenade failures. They seem to all be nova or cinder failures outside our stuff. | 00:15 |
|---|---|---|
| johnsom | Very annoying | 00:15 |
| johnsom | I have seen: | 00:16 |
| johnsom | [ERROR] /opt/stack/new/grenade/projects/70_cinder/resources.sh:200 SSH to the client did not work, something very wrong | 00:16 |
| johnsom | setUpClass (tempest.api.compute.servers.test_server_addresses.ServerAddressesTestJSON) [0.000000s] ... FAILED | 00:16 |
| johnsom | keystoneauth1.exceptions.connection.ConnectFailure: Unable to establish connection to https://192.168.48.135:9696/v2.0/security-group-rules?security_group_id=a28312f4-886f-47cb-9032-60d7c9fc4a0d: ('Connection aborted.', OSError("(104, 'ECONNRESET')",)) | 00:17 |
| johnsom | tempest.api.compute.servers.test_attach_interfaces.AttachInterfacesUnderV243Test.test_add_remove_fixed_ip [69.753469s] ... FAILED | 00:17 |
| johnsom | All on different jobs | 00:18 |
| johnsom | None of which are related to the patches under test | 00:18 |
| johnsom | Just rechecked another grenade: {0} tempest.api.compute.servers.test_attach_interfaces.AttachInterfacesUnderV243Test.test_add_remove_fixed_ip [71.884997s] ... FAILED | 00:24 |
| johnsom | Here is a new one: {0} tempest.api.network.test_routers.RoutersIpV6Test.test_add_multiple_router_interfaces [98.141138s] ... FAILED | 00:31 |
| johnsom | urllib3.exceptions.ReadTimeoutError: HTTPConnectionPool(host='104.130.117.200', port=9696): Read timed out. (read timeout=60) | 00:32 |
| *** yamamoto has joined #openstack-lbaas | 00:57 | |
| *** goldyfruit has joined #openstack-lbaas | 00:58 | |
| *** yamamoto has quit IRC | 01:02 | |
| johnsom | Looks like neutron is just *slow* creating networks: | 01:06 |
| johnsom | INFO neutron.wsgi [None req-90e1fd30-b3df-4ab0-914e-39988864a1b7 tempest-RoutersIpV6Test-381827999 tempest-RoutersIpV6Test-381827999] 104.130.117.200 "POST /v2.0/networks HTTP/1.1" status: 201 len: 769 time: 77.2738550 | 01:06 |
| johnsom | It took 77 seconds to respond, tempest timeout was 60 | 01:07 |
| johnsom | ubuntu-bionic-rax-ord-0012796682 was the host, so known slow host | 01:09 |
| rm_work | ALL of the ones i have seen were the nova no hosts issue | 01:10 |
| rm_work | which can fail on whatever test happens to come first that requires a VM I think | 01:10 |
| johnsom | Well, I at least feel better that I RCA'd one of my grenade failures | 01:11 |
| *** ricolin has joined #openstack-lbaas | 01:25 | |
| *** ltomasbo has quit IRC | 01:50 | |
| *** yamamoto has joined #openstack-lbaas | 02:47 | |
| *** yamamoto has quit IRC | 02:49 | |
| *** yamamoto has joined #openstack-lbaas | 02:49 | |
| *** abaindur has quit IRC | 03:04 | |
| *** abaindur has joined #openstack-lbaas | 03:04 | |
| *** abaindur has quit IRC | 03:04 | |
| *** ramishra has joined #openstack-lbaas | 03:08 | |
| *** yamamoto has quit IRC | 03:11 | |
| *** yamamoto has joined #openstack-lbaas | 03:12 | |
| *** psachin has joined #openstack-lbaas | 03:38 | |
| *** yamamoto has quit IRC | 04:45 | |
| *** goldyfruit has quit IRC | 04:54 | |
| *** spatel has joined #openstack-lbaas | 05:15 | |
| *** yamamoto has joined #openstack-lbaas | 05:19 | |
| *** spatel has quit IRC | 05:20 | |
| *** yamamoto has quit IRC | 05:24 | |
| *** yamamoto has joined #openstack-lbaas | 05:27 | |
| *** ricolin has quit IRC | 06:00 | |
| *** gcheresh_ has joined #openstack-lbaas | 06:11 | |
| *** rcernin has quit IRC | 06:53 | |
| *** AlexStaf has quit IRC | 07:01 | |
| *** yamamoto has quit IRC | 07:04 | |
| *** ccamposr__ has joined #openstack-lbaas | 07:32 | |
| *** ccamposr has quit IRC | 07:35 | |
| *** yamamoto has joined #openstack-lbaas | 07:43 | |
| *** yamamoto has quit IRC | 07:48 | |
| *** maciejjozefczyk has joined #openstack-lbaas | 08:12 | |
| *** tesseract has joined #openstack-lbaas | 08:17 | |
| *** ivve has joined #openstack-lbaas | 08:20 | |
| *** maciejjozefczyk has quit IRC | 08:21 | |
| *** dmellado has quit IRC | 08:21 | |
| *** dmellado has joined #openstack-lbaas | 08:23 | |
| *** dmellado has quit IRC | 08:25 | |
| *** dmellado has joined #openstack-lbaas | 08:27 | |
| *** tkajinam has quit IRC | 08:29 | |
| *** AlexStaf has joined #openstack-lbaas | 08:30 | |
| *** rpittau|afk is now known as rpittau | 08:57 | |
| *** chungpht has quit IRC | 09:11 | |
| *** abaindur has joined #openstack-lbaas | 09:15 | |
| *** trident has quit IRC | 09:16 | |
| *** dmellado has quit IRC | 09:20 | |
| *** dmellado has joined #openstack-lbaas | 09:24 | |
| *** trident has joined #openstack-lbaas | 09:26 | |
| *** yamamoto has joined #openstack-lbaas | 09:40 | |
| *** spatel has joined #openstack-lbaas | 09:50 | |
| *** pcaruana has joined #openstack-lbaas | 09:54 | |
| *** spatel has quit IRC | 09:55 | |
| *** yamamoto has quit IRC | 10:06 | |
| *** yamamoto has joined #openstack-lbaas | 10:10 | |
| *** yamamoto has quit IRC | 10:15 | |
| *** gcheresh_ has quit IRC | 10:39 | |
| *** gcheresh_ has joined #openstack-lbaas | 10:43 | |
| *** henriqueof has joined #openstack-lbaas | 11:55 | |
| *** maciejjozefczyk has joined #openstack-lbaas | 11:57 | |
| *** maciejjozefczyk has quit IRC | 12:02 | |
| *** trident has quit IRC | 12:09 | |
| *** trident has joined #openstack-lbaas | 12:18 | |
| *** tesseract has quit IRC | 12:28 | |
| *** tesseract has joined #openstack-lbaas | 12:29 | |
| *** ccamposr__ has quit IRC | 12:51 | |
| *** psachin has quit IRC | 12:51 | |
| *** vishalmanchanda has quit IRC | 12:52 | |
| *** dulek has quit IRC | 12:52 | |
| *** amotoki has quit IRC | 12:52 | |
| *** coreycb has quit IRC | 12:52 | |
| *** openstackstatus has quit IRC | 12:53 | |
| *** ccamposr__ has joined #openstack-lbaas | 12:54 | |
| *** psachin has joined #openstack-lbaas | 12:54 | |
| *** vishalmanchanda has joined #openstack-lbaas | 12:54 | |
| *** dulek has joined #openstack-lbaas | 12:54 | |
| *** amotoki has joined #openstack-lbaas | 12:54 | |
| *** coreycb has joined #openstack-lbaas | 12:54 | |
| *** yamamoto has joined #openstack-lbaas | 13:16 | |
| *** yamamoto has quit IRC | 13:20 | |
| *** spatel has joined #openstack-lbaas | 13:30 | |
| *** maciejjozefczyk has joined #openstack-lbaas | 13:32 | |
| *** yamamoto has joined #openstack-lbaas | 13:33 | |
| *** spatel has quit IRC | 13:34 | |
| *** maciejjozefczyk has quit IRC | 13:39 | |
| *** henriqueof has quit IRC | 13:40 | |
| *** yamamoto has quit IRC | 13:58 | |
| *** psachin has quit IRC | 14:14 | |
| *** henriqueof has joined #openstack-lbaas | 14:20 | |
| *** goldyfruit has joined #openstack-lbaas | 14:30 | |
| *** yamamoto has joined #openstack-lbaas | 14:32 | |
| *** yamamoto has quit IRC | 14:38 | |
| *** gcheresh_ has quit IRC | 14:46 | |
| *** ricolin has joined #openstack-lbaas | 14:52 | |
| *** TrevorV has joined #openstack-lbaas | 14:56 | |
| *** gcheresh_ has joined #openstack-lbaas | 15:14 | |
| *** ccamposr has joined #openstack-lbaas | 15:26 | |
| *** maciejjozefczyk has joined #openstack-lbaas | 15:28 | |
| *** ccamposr__ has quit IRC | 15:29 | |
| *** goldyfruit_ has joined #openstack-lbaas | 15:43 | |
| *** goldyfruit has quit IRC | 15:45 | |
| *** ivve has quit IRC | 16:02 | |
| *** Trevor_V has joined #openstack-lbaas | 16:12 | |
| *** TrevorV has quit IRC | 16:15 | |
| *** ricolin has quit IRC | 16:20 | |
| *** AlexStaf has quit IRC | 16:28 | |
| *** KeithMnemonic has joined #openstack-lbaas | 17:00 | |
| *** yamamoto has joined #openstack-lbaas | 17:02 | |
| *** rpittau is now known as rpittau|afk | 17:04 | |
| *** yamamoto has quit IRC | 17:06 | |
| *** tesseract has quit IRC | 17:15 | |
| guilhermesp | hello all | 17:18 |
| guilhermesp | do we expect octavia to create neutron resources or just consume them? | 17:18 |
| guilhermesp | for example, users wants to attach a FIP into their LB, but you need previously to allocate a FIP to the project | 17:18 |
| guilhermesp | users that wants to create a LB with a fixed ip of a vxlan network, but we need to previously create a port for it | 17:19 |
| johnsom | Octavia does create an manage resources in neutron, but floating IPs in not one of them. If the user wants a floating IP they will have to attach it after the load balancer is created. | 17:19 |
| johnsom | As for fixed IP vxlan, they can either create the load balancer directly specifying that, or they can pass in a pre-created neutron port. | 17:20 |
| guilhermesp | hum I see... I do see a LB with Error status and has a fixed ip defined, but it doesn't exists in the subnet | 17:20 |
| guilhermesp | yep, pre-created port | 17:20 |
| guilhermesp | if an user wants a fixed ip, it needs to create beforehand | 17:20 |
| guilhermesp | that makes sense then | 17:21 |
| johnsom | Provisioning_status error means the cloud failed in some action Octavia was requesting of it. I.e. neutron is broken, etc. | 17:21 |
| johnsom | You can specify a fixed IP at LB create time, no need to pre-create it | 17:21 |
| guilhermesp | also, I've been seeing in all my funcional LBs the Operating Status as offline | 17:21 |
| johnsom | That likely means either the load balancer is not currently passing traffic or your Octavia deployment is not working. | 17:22 |
| johnsom | If the load balancer is working, check your lb-mgmt-net. It is likely that the amphorae cannot connect back to the health manager processes. | 17:22 |
| guilhermesp | but that's the thing... my listener responds to members, like it is working as expected, but status is offline | 17:23 |
| guilhermesp | hum | 17:23 |
| guilhermesp | gotcha | 17:23 |
| johnsom | Yeah, ok, so your lb-mgmt-net is not working | 17:23 |
| johnsom | amphroae send heartbeats back to the controllers on UDP 5555. You should see messages in the health manager log if you enable debug logging on it. | 17:24 |
| johnsom | DEBUG octavia.amphorae.drivers.health.heartbeat_udp [-] Received packet from ('192.168.0.149', 46000) {{(pid=14074) dorecv /opt/stack/octavia/octavia/amphorae/drivers/health/heartbeat_udp.py:189}} | 17:24 |
| *** ivve has joined #openstack-lbaas | 17:24 | |
| johnsom | and | 17:25 |
| johnsom | DEBUG octavia.controller.healthmanager.health_drivers.update_db [-] Health Update finished in: 0.00808617501752451 seconds {{(pid=16283) update_health /opt/stack/octavia/octavia/controller/healthmanager/health_drivers/update_db.py:72}} | 17:25 |
| guilhermesp | gotcha. I will do some debugging later on | 17:25 |
| johnsom | +1 | 17:25 |
| colin- | if i have 10 spares and i tell octavia to loadbalancer amphora failover each of them, do the first five just eat the last five? | 18:01 |
| johnsom | No, it will delete the amphora you list and the housekeeping process will boot a replacement fresh | 18:05 |
| *** yamamoto has joined #openstack-lbaas | 18:13 | |
| *** maciejjozefczyk has quit IRC | 18:15 | |
| *** yamamoto has quit IRC | 18:18 | |
| *** baffle has quit IRC | 18:22 | |
| *** gcheresh_ has quit IRC | 18:34 | |
| *** yamamoto has joined #openstack-lbaas | 18:53 | |
| *** yamamoto has quit IRC | 18:58 | |
| *** henriqueof has quit IRC | 19:05 | |
| *** abaindur has quit IRC | 19:37 | |
| *** abaindur has joined #openstack-lbaas | 20:13 | |
| *** henriqueof has joined #openstack-lbaas | 20:26 | |
| *** abaindur has quit IRC | 20:28 | |
| *** abaindur has joined #openstack-lbaas | 20:28 | |
| openstackgerrit | Merged openstack/octavia master: Fix listeners with SNI certificates https://review.opendev.org/690444 | 20:43 |
| *** gcheresh_ has joined #openstack-lbaas | 20:47 | |
| *** henriqueof has quit IRC | 21:10 | |
| *** henriqueof has joined #openstack-lbaas | 21:19 | |
| *** goldyfruit_ has quit IRC | 21:23 | |
| *** rcernin has joined #openstack-lbaas | 21:28 | |
| *** henriqueof has quit IRC | 21:33 | |
| *** henriqueof has joined #openstack-lbaas | 21:34 | |
| *** servagem has quit IRC | 21:37 | |
| rm_work | sorrison: you had a chance to look at rebasing your patch on top of mine yet? | 21:42 |
| sorrison | Just looking at it now | 21:42 |
| sorrison | About to -1 you for your patch as found a few bugs :-) | 21:42 |
| rm_work | Yeah please do: P | 21:42 |
| rm_work | Only have minimal unit testing yet, was going to start on tempest today | 21:43 |
| rm_work | And it was basically just a giant copy/paste/find/replace job lol | 21:43 |
| sorrison | yeah I picked up on that, looks good though | 21:46 |
| *** gcheresh_ has quit IRC | 21:47 | |
| *** yamamoto has joined #openstack-lbaas | 21:48 | |
| *** goldyfruit_ has joined #openstack-lbaas | 21:55 | |
| *** henriqueof has quit IRC | 21:58 | |
| *** henriqueof has joined #openstack-lbaas | 22:00 | |
| *** henriqueof has quit IRC | 22:00 | |
| *** Trevor_V has quit IRC | 22:03 | |
| sorrison | rm_work: question when returning a LB from the API should it return availability_zone.name as availability_zone or availability_zone_id and then up to the client to go and resolve that? | 22:11 |
| sorrison | I'm leaning towards returning the name | 22:12 |
| rm_work | Err, probably ID the way it's designed now? Though name would be more in line with other AZ implementations | 22:14 |
| rm_work | But different from flavors | 22:14 |
| rm_work | Name IS unique for both, but | 22:22 |
| rm_work | We allow deleting a zone and making a new zone with the same name, so | 22:23 |
| rm_work | That could POSSIBLY lead to confusion? But I don't know if that would happen often enough for it to be a design concern | 22:23 |
| rm_work | Just rambling some thoughts | 22:23 |
| rm_work | If we're using ID as the key in the AZ objects, then we need to return ID | 22:24 |
| rm_work | I think that's going to be my assertion for the moment | 22:25 |
| sorrison | I could use the name as the key in az objects then? | 22:47 |
| sorrison | I just thinking from a client side the ID of an AZ is pretty useless | 22:47 |
| sorrison | An AZ is basically just a name in all other projects too | 22:48 |
| *** yamamoto has quit IRC | 22:49 | |
| rm_work | I mean, if we do that, we should commit fully | 22:53 |
| *** tkajinam has joined #openstack-lbaas | 22:54 | |
| rm_work | Just remove the id column and use name as PK | 22:54 |
| rm_work | Thoughts johnsom ? | 22:54 |
| rm_work | It's a divergence from the flavor model | 22:55 |
| rm_work | Which I was trying to avoid | 22:55 |
| johnsom | Well, everything we have with names is returned as the ID in our API. | 22:55 |
| rm_work | But otherwise it's a divergence from the Openstack AZ model | 22:55 |
| johnsom | Typically for uniqueness reasons | 22:55 |
| rm_work | So... Which one is more important | 22:55 |
| rm_work | I'm actually kinda ok with removing id and using name as the pk | 22:56 |
| rm_work | In this specific case because it would be consistent with other AZ usage? | 22:56 |
| sorrison | my vote would be to use id as the link internally but from the API perspective hide availability_zone_id and show availability_zone which would be the name | 22:56 |
| johnsom | Give me a minute to look around before I vote. | 22:56 |
| rm_work | At least nova does it that way, need to look at neutron and cinder | 22:57 |
| sorrison | cinder is same as nova | 22:57 |
| rm_work | Hmm yeah that's possible too though kinda pointless maybe | 22:57 |
| rm_work | Dunno | 22:57 |
| rm_work | Well, as long as name is unique I guess it's fine | 22:57 |
| rm_work | Just change all the lookups to be name based | 22:57 |
| johnsom | Yeah, nova isn't exactly the poster child for how to do things right either.... | 22:58 |
| sorrison | Yeah I guess because name is unique then we can use name as the foreign key and use that everywhere | 22:59 |
| rm_work | Right, but the question is, do we stay consistent to the existing model, even if it's a little wonky? | 22:59 |
| rm_work | Yeah then name also becomes immutable IMO | 22:59 |
| sorrison | I think that would be fine | 22:59 |
| sorrison | also loadbalancer.operating_status is linked by the name | 23:00 |
| sorrison | and topology | 23:00 |
| rm_work | Yeah I don't think linking by name is inherently bad | 23:00 |
| rm_work | Just inconsistent with the flavor model | 23:00 |
| johnsom | Yeah, I think name as ID is fine. It will always have to be unique, so you can't have duplicate AZs across regions, etc. | 23:08 |
| johnsom | https://usercontent.irccloud-cdn.com/file/jiaa8Zv3/image.png | 23:09 |
| johnsom | But I guess in OpenStack region is a complete duplicate control plane. | 23:09 |
| sorrison | yeah another region would mean another octavia | 23:10 |
| johnsom | Personally I always default to using abstract IDs, that way I don't paint myself in a corner, but it's true the other projects just pass names around | 23:11 |
| *** ivve has quit IRC | 23:14 | |
| *** KeithMnemonic has quit IRC | 23:15 | |
| rm_work | k, i can make changes today | 23:37 |
| rm_work | you think keep ID in our DB and just change stuff to *return* names? or ... actually remove ID | 23:37 |
| johnsom | Less headache to just use the name as the primary key IMO. If you aren't going to expose the ID. | 23:59 |
| johnsom | Otherwise you will be adding/dropping the ID column all of the time | 23:59 |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!
