| *** rcernin has quit IRC | 00:02 | |
| *** rcernin has joined #openstack-lbaas | 00:02 | |
| bar_ | cgoncalves, nice work! | 00:06 |
|---|---|---|
| bar_ | johnsom, Quota client: https://review.openstack.org/#/c/518767/ | 00:06 |
| johnsom | On the list! | 00:07 |
| *** rcernin has quit IRC | 00:07 | |
| johnsom | I would like to get that merged before Tuesday next week if we can. I need to cut a queens release of the client. | 00:07 |
| *** rcernin has joined #openstack-lbaas | 00:07 | |
| bar_ | I got another 1 or 2 of my own, you know. They're just sitting there, waiting for some core to come to their rescue. | 00:08 |
| johnsom | Or any other reviewer! grin | 00:09 |
| bar_ | one of them has already +1 | 00:09 |
| johnsom | Not 20? grin | 00:09 |
| bar_ | what can I say, tough crowd | 00:10 |
| johnsom | lol | 00:10 |
| openstackgerrit | Adam Harwell proposed openstack/octavia master: Optimize update_health process https://review.openstack.org/504875 | 00:13 |
| openstackgerrit | Adam Harwell proposed openstack/octavia master: Clean up test_update_db.py a little bit https://review.openstack.org/520863 | 00:13 |
| openstackgerrit | Adam Harwell proposed openstack/octavia master: WIP: Floating IP Network Driver (spans L3s) https://review.openstack.org/435612 | 00:13 |
| rm_work | i hope that's correct rebasing | 00:13 |
| *** armax has quit IRC | 00:54 | |
| openstackgerrit | Bar RH proposed openstack/python-octaviaclient master: Extend loadbalancer_create valid VIP parameters combinations https://review.openstack.org/519439 | 01:19 |
| openstackgerrit | Bar RH proposed openstack/python-octaviaclient master: Complement Octavia client with a set of features https://review.openstack.org/522666 | 01:19 |
| rm_work | putting the amphora-api in the client might be good, but it's an admin thing so it's less urgent | 01:20 |
| *** sticker has joined #openstack-lbaas | 01:25 | |
| bar_ | rm_work what do you suggest? is there a reason to defer it? | 01:26 |
| rm_work | no reason to defer | 01:27 |
| rm_work | just, if there's more important work | 01:27 |
| rm_work | but i'd really like to have it :P | 01:27 |
| rm_work | i assume like... | 01:27 |
| rm_work | openstack loadbalancer amphora list | 01:27 |
| rm_work | and | 01:28 |
| openstackgerrit | Michael Johnson proposed openstack/octavia master: Provider driver spec https://review.openstack.org/509957 | 01:28 |
| rm_work | openstack loadbalancer amphora list --loadbalancer <lb_id> | 01:28 |
| bar_ | it is working, even with filters (loadbalancer id, compute id, role, provisioning status) | 01:28 |
| bar_ | I use it for a few days now... | 01:28 |
| rm_work | err, wait, did you do it already? | 01:28 |
| bar_ | :-D | 01:29 |
| rm_work | i didn't actually see | 01:29 |
| rm_work | lol | 01:29 |
| rm_work | loool yes you did | 01:29 |
| rm_work | cool, i'm asking for something you already wrote | 01:29 |
| rm_work | lol | 01:29 |
| rm_work | yep you even did it the way i was thinking | 01:29 |
| rm_work | whelp, i'll review that when i have a chance, prolly tomorrow | 01:29 |
| bar_ | awesome, note that there dependencies (that await review too). | 01:30 |
| bar_ | *there're | 01:31 |
| bzhao | johnsom, ping. Hi sir, are you still online? | 01:49 |
| bar_ | rm_work, do you have any idea why the documentation of octavia api contains only a subset of the possible amphora statuses? | 01:54 |
| bzhao | rm_work, a ha, :). Sir, could you help me to try to understand the details of johnsom said in the UDP spec, https://review.openstack.org/#/c/503606/7/specs/version1.1/udp_support.rst in comments L70? I'm not sure I understood correct. :(. Sorry | 01:55 |
| *** bbbbzhao_ has joined #openstack-lbaas | 01:56 | |
| bzhao | Did johnsom suggest it should be still 1:1 UDP listener:vrrp_instance? And through check script for move the vrrp state of LB which is MASTER role already from MASTER to BACKUP if there is a UDP keepalived dead. | 01:58 |
| *** AlexeyAbashkin has joined #openstack-lbaas | 02:12 | |
| *** rcernin_ has joined #openstack-lbaas | 02:14 | |
| *** mestery_ has joined #openstack-lbaas | 02:16 | |
| *** AlexeyAbashkin has quit IRC | 02:16 | |
| rm_work | bar_: i'm not sure why it'd be partial -- unless it's assuming that EVERYTHING has some statuses, which should be added to that list implicitly | 02:16 |
| rm_work | bar_: which statuses does it list? can you link me the docs you're looking at? | 02:17 |
| *** ptoohill- has joined #openstack-lbaas | 02:17 | |
| *** ipsecguy_ has joined #openstack-lbaas | 02:19 | |
| bar_ | sec | 02:19 |
| *** isp has joined #openstack-lbaas | 02:20 | |
| *** annp has joined #openstack-lbaas | 02:20 | |
| bar_ | rm_work, https://github.com/openstack/octavia/blob/master/octavia/common/constants.py#L101-L102 | 02:20 |
| *** rcernin has quit IRC | 02:21 | |
| *** ipsecguy has quit IRC | 02:22 | |
| *** issp has quit IRC | 02:22 | |
| *** zigo has quit IRC | 02:22 | |
| *** mestery has quit IRC | 02:22 | |
| *** ptoohill1 has quit IRC | 02:22 | |
| *** mestery_ is now known as mestery | 02:22 | |
| bar_ | note that i couldn't find any reference to this constant in the code, but the content is identical to the API doc | 02:24 |
| *** zigo has joined #openstack-lbaas | 02:27 | |
| *** zigo is now known as Guest13268 | 02:29 | |
| *** fnaval has joined #openstack-lbaas | 02:34 | |
| openstackgerrit | Bar RH proposed openstack/python-octaviaclient master: Complement Octavia client with a set of features https://review.openstack.org/522666 | 02:35 |
| rm_work | those statuses are all legit | 02:38 |
| rm_work | and i have seen them all within the last hour :P | 02:38 |
| rm_work | bar_: ^^ | 02:38 |
| rm_work | which docs are you looking at? | 02:38 |
| rm_work | and what is the discrepancy exactly? | 02:38 |
| bar_ | there should be more | 02:39 |
| rm_work | err | 02:39 |
| rm_work | what would they be? | 02:39 |
| bar_ | e.g. pending_update is missing from this list | 02:39 |
| rm_work | AMPs dont go to PENDING_UPDATE | 02:39 |
| rm_work | they go from BOOTING->READY->ALLOCATED->(PENDING_DELETE->DELETED)/ERROR | 02:40 |
| rm_work | ah hmmmm actually I MIGHT have seen "PENDING_CREATE" | 02:40 |
| rm_work | but there is no PENDING_UPDATE state for them | 02:40 |
| bar_ | aha | 02:40 |
| bar_ | nor ACTIVE, right? | 02:41 |
| rm_work | right | 02:41 |
| rm_work | ACTIVE == ALLOCATED | 02:41 |
| rm_work | or READY, for spares-pool | 02:41 |
| rm_work | hmm it is interesting we don't have PENDING_CREATE there | 02:42 |
| rm_work | it's also interesting that "SUPPORTED_PROVISIONING_STATUSES" includes some amphora statuses | 02:42 |
| rm_work | that aren't possible for LBs | 02:42 |
| rm_work | amps don't have "provisioning/operating" statuses | 02:42 |
| rm_work | so the AMPHORA_* statuses shouldn't be listed in SUPPORTED_PROVISIONING_STATUSES | 02:43 |
| rm_work | just SUPPORTED_AMPHORA_STATUSES | 02:43 |
| rm_work | again, which docs page are you reading? | 02:43 |
| rm_work | I wonder if it is MORE correct | 02:43 |
| bar_ | parameters.yaml | 02:44 |
| rm_work | err | 02:44 |
| bar_ | need a link? | 02:45 |
| rm_work | yeah that'd be easier | 02:45 |
| bar_ | https://github.com/openstack/octavia/blob/master/api-ref/source/parameters.yaml#L146-L149 | 02:46 |
| rm_work | ok yeah, docs are correct mostly | 02:51 |
| rm_work | they are missing PENDING_CREATE | 02:51 |
| rm_work | if you wanted to add that, go for it :) | 02:51 |
| bar_ | no, I need to specify which statuses are valid for the client | 02:52 |
| openstackgerrit | Bar RH proposed openstack/python-octaviaclient master: Complement Octavia client with a set of features https://review.openstack.org/522666 | 02:52 |
| rm_work | yeah but i mean | 02:53 |
| rm_work | if you want to add that fix to one of your octavia patches, i'd +2 :P | 02:53 |
| bar_ | haha, ok. intriguing proposal indeed. | 02:54 |
| openstackgerrit | Bar RH proposed openstack/octavia master: Fix filtering in list API calls https://review.openstack.org/522689 | 03:01 |
| bar_ | rm_work, ^^ | 03:02 |
| openstackgerrit | Bar RH proposed openstack/python-octaviaclient master: Complement Octavia client with a set of features https://review.openstack.org/522666 | 03:03 |
| *** bar_ has quit IRC | 03:07 | |
| johnsom | bar_ what is all this???? | 03:22 |
| johnsom | Those two match.. | 03:23 |
| johnsom | Amps only have booting and ready for pre-allocated states | 03:23 |
| johnsom | rm_work where did you see PENDING_CREATE being used for an amphora? | 03:29 |
| johnsom | bzhao o/ | 03:30 |
| *** codenamelxl has joined #openstack-lbaas | 03:31 | |
| bzhao | johnsom, Hi sir. Could you please check the https://review.openstack.org/#/c/503606/7/specs/version1.1/udp_support.rst in L70 I replied if you are free? I'm not sure I understood correct. Sorry. :) | 03:31 |
| codenamelxl | Hi everyone, i'm having trouble with Floating IP for a LB. I can connect to all VM through Floating IP. Can Connect to LB VIP. However, the floating IP attached to the VIP just does not work. | 03:33 |
| johnsom | codenamelxl Are you using a DVR router? | 03:34 |
| codenamelxl | yes | 03:34 |
| johnsom | bzhao Looking | 03:34 |
| codenamelxl | it's work with legacy | 03:34 |
| bzhao | johnsom, Thank you, because I start up coding from backend layer to front layer, so I wish to make sure the backend part looks like. | 03:35 |
| johnsom | codenamelxl If your neutron is older than Pike DVR has a bug that does not work adding floating IPs to some types of ports. | 03:35 |
| codenamelxl | I'm using Pike DVR | 03:35 |
| johnsom | Hmm, they were supposed to have fixed it in Pike release. | 03:35 |
| johnsom | So the floating IP attaches, but traffic does not flow? | 03:36 |
| codenamelxl | Yes | 03:36 |
| codenamelxl | it attached | 03:36 |
| codenamelxl | traffic does not come through. Checked the Sec group. I can access the VIP too. | 03:37 |
| johnsom | Yeah, this was a long standing bug (like 4+ releases) in DVR. | 03:37 |
| johnsom | Floating IPs associated with an unbound port with DVR routers will not be distributed, but will be centralized and implemented in the SNAT namespace of the Network node or dvr_snat node. Floating IPs associated with allowed_address_pair port IP and are bound to multiple active VMs with DVR routers will be implemented in the SNAT namespace in the Network node or dvr_snat node. This will address VRRP use cases. | 03:37 |
| johnsom | More information about this is captured in bug 1583694. | 03:37 |
| openstack | bug 1583694 in neutron "[RFE] DVR support for Allowed_address_pair port that are bound to multiple ACTIVE VM ports" [Wishlist,Fix released] https://launchpad.net/bugs/1583694 - Assigned to Swaminathan Vasudevan (swaminathan-vasudevan) | 03:37 |
| johnsom | https://docs.openstack.org/releasenotes/neutron/pike.html | 03:37 |
| johnsom | That is the release note for Pike that addressed the issue. At least that is what I was told. | 03:38 |
| bzhao | https://bugs.launchpad.net/bugs/1733852 | 03:39 |
| openstack | Launchpad bug 1733852 in neutron "Incorrect ARP entries in new DVR routers for Octavia VRRP addresses" [Medium,In progress] - Assigned to Daniel Russell (danielr-2) | 03:39 |
| johnsom | That is all vague and complex, but I was told it fixed the bug we saw | 03:39 |
| johnsom | Oh no, there is another one???? | 03:39 |
| bzhao | Yeah. I found it just last week. | 03:39 |
| codenamelxl | Oops | 03:39 |
| johnsom | Well, sorry about that. You can poke folks in the neutron channel to see if they can backport that fix. | 03:42 |
| johnsom | The other alternative is to use a non-DVR router for your VIPs. | 03:42 |
| johnsom | bzhao Yeah, sorry I was not clear | 03:45 |
| johnsom | Can you chat now? | 03:45 |
| johnsom | On IRC? | 03:45 |
| bzhao | me ? | 03:45 |
| johnsom | yes | 03:45 |
| johnsom | Do you have a few minutes? I will try to explain better | 03:46 |
| bzhao | Yeah. | 03:46 |
| bzhao | Thank you | 03:46 |
| johnsom | Ok. So the short of it, you were correct, I was wrong. | 03:46 |
| johnsom | I figured out that yes, the UDP keepaliveds should have no VRRP. | 03:47 |
| bzhao | I just what to implement it like your suggest. So sorry to bother you. :) | 03:48 |
| johnsom | What I was proposing is to add a check script to the main VRRP keepalived. We already have a directory that it will run scripts in. | 03:48 |
| bzhao | OK, It's better to list the UDP keepalived process id and check whether all of them are alived in the single keepalived process | 03:49 |
| johnsom | This script will check the health of the UDP keepalived and it it is failed will fail the main VRRP keepalived checks to move the VIP | 03:49 |
| bzhao | Yeah. I think we are close. :) | 03:49 |
| *** rcernin has joined #openstack-lbaas | 03:50 | |
| bzhao | And the check script must be default for UDP cases. | 03:50 |
| *** rcernin_ has quit IRC | 03:51 | |
| johnsom | The script will not check the health of the UDP member servers, that is a different script | 03:52 |
| johnsom | It will just check the UDP keepalived health. | 03:52 |
| bzhao | Yeah, just check keepalived process. should we check the keepalived process id is alive? | 03:53 |
| *** links has joined #openstack-lbaas | 03:55 | |
| bzhao | for example, if a UDP listener startup, we let its process_id into a particular directory, then the check script just list the dir and check the process id. | 03:55 |
| bzhao | Once there is a failure, the main VRRP instance move the VIP to the backup one | 03:56 |
| johnsom | Yes, at a minimum. There is a signal you can send keepalived that dumps a status file to /tmp, it may have better info, but at a minimum check the process is running via kill -0 or systemctl status | 03:56 |
| johnsom | Yep | 03:56 |
| johnsom | This is how we check the additional haproxy processes for each listener | 03:57 |
| bzhao | OK, I know. Sorry. I think I understood now. | 03:59 |
| bzhao | Thank you | 03:59 |
| johnsom | Cool. No worries. This is why we do specifications. We can work out the details and make the best solution | 03:59 |
| bzhao | aha, Thanks to your guidance | 04:01 |
| bzhao | All of them are very useful. :) | 04:01 |
| johnsom | Ha, thanks! | 04:02 |
| johnsom | Let me know if there are more questions? | 04:02 |
| bzhao | Also, I make a contrast with haproxy and LVS concepts towards configuration options. I'm afraid that there are aome conflicts towards healthmonitor API. | 04:03 |
| johnsom | I commented on some that in the patch. | 04:04 |
| bzhao | Yeah. I just found the concepts are different. If we think that's OK. I can implement it like you mentioned . | 04:05 |
| rm_work | johnsom: i literally watched it happen | 04:05 |
| johnsom | bzhao Yeah, they may be slightly different | 04:05 |
| rm_work | johnsom: https://i.imgur.com/cmWeTYU.png | 04:05 |
| rm_work | it's pre-booting | 04:05 |
| rm_work | *absolute initial state* | 04:06 |
| rm_work | i happened to catch that while i was watching | 04:06 |
| johnsom | Hmmmm, what the heck is creating that.... I never intended that state to be there | 04:07 |
| bzhao | johnsom, OK, that's clear from my side. There is no any other question now, if there is any issue during its forward, I wish your kind help and advise. :) Thank you. | 04:07 |
| johnsom | bzhao Ok, no problem. Let's make sure we document those differences | 04:08 |
| bzhao | johnsom, aha, I will list the description from both side and add them into the doc work. | 04:09 |
| johnsom | rm_work Crumb, I found it. sigh | 04:10 |
| rm_work | johnsom: so you intended them to be CREATED in "BOOTING" state? | 04:12 |
| rm_work | the initial record? | 04:12 |
| johnsom | Yeah, pending create was intended to be booting | 04:13 |
| johnsom | But the DB record is created with PENDING_CREATE then moves to booting | 04:14 |
| bzhao | codenamelxl, I'm not sure whether this feature could solve u issue if your phycial env let u do this. DVR support centralized DNAT. Please check: https://review.openstack.org/#/c/485333 | 04:14 |
| rm_work | johnsom: i'm not sure if that's actually too bad | 04:14 |
| johnsom | Yeah | 04:15 |
| rm_work | because... initially they ... AREN'T booting | 04:15 |
| rm_work | technically | 04:15 |
| rm_work | since, you know... it's just a DB record | 04:15 |
| johnsom | Yeah | 04:15 |
| rm_work | and then when the nova create happens, it switches to BOOTING, right :P | 04:15 |
| johnsom | Yeah | 04:16 |
| rm_work | so yeah it'd just be a doc update thing | 04:16 |
| * rm_work shrugs | 04:16 | |
| johnsom | Yeah, it's docs and constants | 04:16 |
| rm_work | ah and maybe constants | 04:16 |
| rm_work | yeah | 04:16 |
| rm_work | k i'll be back in a bit, need to run to target | 04:17 |
| johnsom | Bar got it | 04:17 |
| johnsom | o/ | 04:17 |
| rm_work | Targé | 04:17 |
| rm_work | heh k | 04:17 |
| johnsom | Exactly | 04:17 |
| codenamelxl | @bzhao: https://bugs.launchpad.net/neutron/+bug/1733852 suggested that the bug just happen if i add the router after i created the LB. Just checked the router's ARP. It looks fine. Maybe, there is something else. | 04:22 |
| openstack | Launchpad bug 1733852 in neutron "Incorrect ARP entries in new DVR routers for Octavia VRRP addresses" [Medium,In progress] - Assigned to Daniel Russell (danielr-2) | 04:22 |
| *** threestrands_ has joined #openstack-lbaas | 04:24 | |
| *** threestrands_ has quit IRC | 04:24 | |
| *** threestrands_ has joined #openstack-lbaas | 04:24 | |
| *** threestrands has quit IRC | 04:26 | |
| codenamelxl | @bzhao: Checked the ARP on the amphora node. There is no ARP resolve for the router... Weird? | 04:33 |
| *** kbyrne has quit IRC | 04:46 | |
| codenamelxl | @bzhao Found it, the router can't reach the amphora. Not sure why. Any direction? Thanks | 04:48 |
| *** reedip has quit IRC | 05:03 | |
| *** rcernin_ has joined #openstack-lbaas | 05:09 | |
| *** rcernin has quit IRC | 05:09 | |
| bzhao | codenamelxl, the Compute Nodes routers ARP entries should override by the vrrp_port's mac, because vrrp_port use vip_port as its allowed_address_pair, but the l2 pop generate the ARP entry just from the subnet port(including unbound port), So I dont think we can solve them through API/configuration. If possible, you can modify the ARP entries VIP address:MAC, modify the MAC to the real LB mac which is a MASTER role LB, and try whether it works. | 05:14 |
| codenamelxl | Both of my IP point to the VRRP's MAC | 05:15 |
| bzhao | er, maybe there is any thing we don't cover. | 05:21 |
| rm_work | johnsom: errm, how would a *member* get a 409 on DELETE | 05:24 |
| codenamelxl | Yep, it might be just me screwing something up. Right now, i can ping the router from the Amphora, but the router can't ping the amphora. Any suggestion? | 05:24 |
| rm_work | if the LB itself is ACTIVE | 05:24 |
| rm_work | this is ... perplexing | 05:24 |
| johnsom | rm_work the LB is in a PENDING | 05:25 |
| rm_work | ok but like | 05:25 |
| rm_work | this test | 05:25 |
| rm_work | literally does "self.await_loadbalancer_active(self.lb_id)" | 05:26 |
| rm_work | and then tries to start deleting stuff (starting with members) | 05:26 |
| rm_work | and gets a 409 | 05:26 |
| rm_work | 2017-11-30 20:27:59.406 26528 INFO tempest.lib.common.rest_client [req-afb80d65-58ed-464e-ab9d-0898d365f6d0 ] Request (BasicOpsTest:test_octavia_failover): 200 GET https://octavia.phx-private.openstack.int.godaddy.com:443/v2.0/lbaas/loadbalancers/d31d4ff4-dd23-4d6c-bf76-cb38b802770a 0.411s | 05:26 |
| rm_work | ^^ that's it seeing the LB back in ACTIVE | 05:26 |
| rm_work | the next line is: | 05:26 |
| rm_work | 2017-11-30 20:28:07.359 26528 INFO tempest.lib.common.rest_client [req-5d448256-cbbc-4f07-a931-678aef2fdbd0 ] Request (BasicOpsTest:_run_cleanups): 409 DELETE https://octavia.phx-private.openstack.int.godaddy.com:443/v2.0/lbaas/pools/ed3a81f5-7905-4ae2-aeea-da0b08475da2/members/a8599923-8de3-47e6-831c-e561e568d7b1 0.149s | 05:26 |
| rm_work | so unless it somehow went immediately back to PENDING .... | 05:27 |
| johnsom | Right | 05:27 |
| rm_work | how would it go back to pending | 05:28 |
| johnsom | I would look at the timings and the o-cw logs | 05:28 |
| rm_work | yeah i guess so, ugh | 05:28 |
| rm_work | yeah API confirms somehow it's immutable | 05:31 |
| rm_work | but that makes no sense :/ | 05:31 |
| rm_work | ahh wut | 05:35 |
| rm_work | ok well, that's ... not what i expected. ok, different issue | 05:36 |
| rm_work | it actually failed while trying to do the self.await_loadbalancer_active | 05:36 |
| rm_work | T_T | 05:36 |
| *** sticker has quit IRC | 05:36 | |
| openstackgerrit | Adam Harwell proposed openstack/octavia-tempest-plugin master: WIP: Failover test https://review.openstack.org/501559 | 05:38 |
| *** fnaval has quit IRC | 05:39 | |
| johnsom | I beat the crud out of that API so pretty confident we don't have locking issues | 05:40 |
| rm_work | well | 05:41 |
| rm_work | I'm not sure what's happening here either | 05:41 |
| rm_work | or why it seems so darned consistently to fail right here | 05:42 |
| johnsom | It's not just this? http://logs.openstack.org/59/501559/13/check/octavia-v2-dsvm-scenario/97fe8b2/job-output.txt.gz#_2017-11-28_12_04_43_165394 | 05:46 |
| rm_work | no | 05:47 |
| rm_work | that's because devstack doesn't update the thing | 05:47 |
| rm_work | and i think that devstack is only in SINGLE mode | 05:47 |
| rm_work | this test only works in ACTIVE_STANDBY topo | 05:47 |
| rm_work | I need to figure out a way to do it for SINGLE and for upstream | 05:47 |
| rm_work | it wouldn't work on upstream ACTIVE_STANDBY either | 05:48 |
| rm_work | :( | 05:48 |
| johnsom | Yeah, before I think we didn't have enough ram on the gate hosts to boot act/stdby | 05:48 |
| rm_work | once we get flavors working | 05:48 |
| rm_work | we can have gate tests use that | 05:48 |
| rm_work | register a flavor for the two topos | 05:48 |
| rm_work | and run tests accordingly | 05:48 |
| *** gcheresh has joined #openstack-lbaas | 05:49 | |
| rm_work | oh damnit i did the wrong thing in my debug | 05:50 |
| *** eN_Guruprasad_Rn has joined #openstack-lbaas | 05:50 | |
| openstackgerrit | Adam Harwell proposed openstack/octavia-tempest-plugin master: WIP: Failover test https://review.openstack.org/501559 | 05:50 |
| *** armax has joined #openstack-lbaas | 06:01 | |
| *** pcaruana has joined #openstack-lbaas | 06:10 | |
| *** threestrands_ has quit IRC | 06:24 | |
| *** armax has quit IRC | 06:27 | |
| *** gcheresh has quit IRC | 06:35 | |
| openstackgerrit | Hengqing Hu proposed openstack/octavia-dashboard master: Update openstacksdk construction to be forward compatible https://review.openstack.org/524011 | 06:52 |
| *** reedip has joined #openstack-lbaas | 07:02 | |
| rm_work | ah somehow it tries to do tempest cleanup before the test is totally done or something O_o | 07:06 |
| rm_work | it's timing out hitting the API it looks like? wtf | 07:06 |
| * rm_work dies | 07:06 | |
| *** fnaval has joined #openstack-lbaas | 07:08 | |
| *** fnaval has quit IRC | 07:13 | |
| *** bbbbzhao_ has quit IRC | 07:26 | |
| *** rcernin_ has quit IRC | 07:29 | |
| *** fnaval has joined #openstack-lbaas | 07:43 | |
| *** fnaval has quit IRC | 07:47 | |
| *** slaweq has joined #openstack-lbaas | 08:03 | |
| *** AlexeyAbashkin has joined #openstack-lbaas | 08:15 | |
| openstackgerrit | ZhaoBo proposed openstack/octavia master: Support UDP load balance https://review.openstack.org/503606 | 08:27 |
| openstackgerrit | Bernard Cafarelli proposed openstack/octavia master: Rework amphora agent installation element https://review.openstack.org/522626 | 09:08 |
| bcafarel | rm_work: ^ merged amphora elements with a few enhancements :) | 09:14 |
| *** gcheresh has joined #openstack-lbaas | 09:15 | |
| *** isp is now known as issp | 09:21 | |
| *** ipsecguy_ is now known as ipsecguy | 09:45 | |
| *** codenamelxl has left #openstack-lbaas | 09:47 | |
| *** sri_ has joined #openstack-lbaas | 09:58 | |
| *** gcheresh has quit IRC | 10:28 | |
| *** salmankhan has joined #openstack-lbaas | 10:33 | |
| *** openstackgerrit has quit IRC | 10:33 | |
| *** fnaval has joined #openstack-lbaas | 10:43 | |
| *** salmankhan has quit IRC | 10:44 | |
| *** fnaval has quit IRC | 10:47 | |
| *** salmankhan has joined #openstack-lbaas | 11:14 | |
| *** fnaval has joined #openstack-lbaas | 11:43 | |
| *** fnaval has quit IRC | 11:48 | |
| *** salmankhan has quit IRC | 11:55 | |
| *** gcheresh has joined #openstack-lbaas | 12:31 | |
| *** yamamoto has quit IRC | 12:53 | |
| *** yamamoto has joined #openstack-lbaas | 12:54 | |
| *** yamamoto has quit IRC | 13:09 | |
| *** pckizer has quit IRC | 13:15 | |
| *** links has quit IRC | 13:17 | |
| *** openstackgerrit has joined #openstack-lbaas | 13:22 | |
| openstackgerrit | Bernard Cafarelli proposed openstack/octavia master: Rework amphora agent installation element https://review.openstack.org/522626 | 13:22 |
| *** pck has joined #openstack-lbaas | 13:23 | |
| *** yamamoto has joined #openstack-lbaas | 13:25 | |
| *** eN_Guruprasad_Rn has quit IRC | 13:27 | |
| *** pck has quit IRC | 13:30 | |
| *** pck has joined #openstack-lbaas | 13:30 | |
| *** yamamoto has quit IRC | 13:32 | |
| *** yamamoto has joined #openstack-lbaas | 13:33 | |
| -openstackstatus- NOTICE: gerrit has been restarted to get it back to its normal speed. | 13:51 | |
| *** links has joined #openstack-lbaas | 13:58 | |
| *** yamamoto has quit IRC | 14:01 | |
| *** AlexeyAbashkin has quit IRC | 14:02 | |
| *** salmankhan has joined #openstack-lbaas | 14:09 | |
| *** yamamoto has joined #openstack-lbaas | 14:17 | |
| *** links has quit IRC | 14:26 | |
| *** yamamoto has quit IRC | 14:33 | |
| *** slaweq has quit IRC | 14:41 | |
| *** slaweq has joined #openstack-lbaas | 14:41 | |
| *** Alex_Staf has joined #openstack-lbaas | 14:52 | |
| *** sdaniel has joined #openstack-lbaas | 14:53 | |
| sdaniel | Hi, I would like to ask, how can I identify why the Load balancer is not deployable ? | 14:55 |
| sdaniel | I'm running OS Ocata and tried to configure based on the official guidelines | 14:55 |
| *** yamamoto has joined #openstack-lbaas | 14:59 | |
| *** yamamoto has quit IRC | 15:00 | |
| *** armax has joined #openstack-lbaas | 15:32 | |
| *** yamamoto has joined #openstack-lbaas | 15:35 | |
| *** Alex_Staf has quit IRC | 15:39 | |
| *** fnaval has joined #openstack-lbaas | 15:50 | |
| *** Alex_Staf has joined #openstack-lbaas | 16:04 | |
| *** slaweq has quit IRC | 16:04 | |
| *** slaweq has joined #openstack-lbaas | 16:04 | |
| *** slaweq has quit IRC | 16:09 | |
| xgerman_ | sdaniel what means not deployable | 16:09 |
| *** Alex_Staf has quit IRC | 16:10 | |
| sdaniel | just a sec, I read it from the logs | 16:22 |
| *** armax has quit IRC | 16:27 | |
| sdaniel | Could not load neutron_lbaas.drivers.haproxy.namespace_driver.HaproxyNSDriver | 16:28 |
| sdaniel | I got this or the following: | 16:29 |
| xgerman_ | This is not Octavia but neutron lbaas | 16:29 |
| sdaniel | Loadbalancer 413c7406-a916-4278-bacd-dd4ffd2c8626 is not deployable. | 16:29 |
| xgerman_ | ok, so you want to check if the lbaas agent is running | 16:30 |
| johnsom | This is a known neutron warning that has no meaning: Could not load neutron_lbaas.drivers.haproxy.namespace_driver.HaproxyNSDriver | 16:30 |
| johnsom | Typically it does actually load the driver | 16:30 |
| sdaniel | Which one is the expected warning log from the two ? | 16:32 |
| sdaniel | Ok, so the first one | 16:33 |
| sdaniel | And what about with the following warning: "Stats socket not found for loadbalancer f533b6dd-495f-4393-817f-b21b3640c96e" ? | 16:35 |
| johnsom | Yeah, that happens with that old driver. A load balancer has been created but a listener has not yet been created. | 16:38 |
| *** slaweq has joined #openstack-lbaas | 16:42 | |
| *** slaweq_ has joined #openstack-lbaas | 16:42 | |
| sdaniel | And what about the other ? | 16:42 |
| xgerman_ | did you check if the agents are running? | 16:43 |
| sdaniel | The service is active, I did all the steps from here: https://docs.openstack.org/ocata/networking-guide/config-lbaas.html | 16:44 |
| sdaniel | But I can't ping | 16:44 |
| sdaniel | ( Dashboard is not installed at all, so I skipped that part) | 16:44 |
| *** slaweq has quit IRC | 16:46 | |
| sdaniel | neutron-lbaasv2-agent.service is running | 16:48 |
| xgerman_ | neutron agent list or so? | 16:48 |
| xgerman_ | ok | 16:48 |
| xgerman_ | you are not supposed to ping anyway | 16:48 |
| xgerman_ | this will run an haproxy inside a namesoace | 16:49 |
| sdaniel | (A have couple of agents: metadata, loadbalancerv2, linux bridge, dhcp) | 16:49 |
| johnsom | Yeah, since you are running the old driver, "neutron agent-list" should show a :-) next to it | 16:49 |
| sdaniel | Yes, I heard about that | 16:49 |
| sdaniel | ( I found out, the agent status result came from there ) | 16:50 |
| sdaniel | If i exec | 16:51 |
| sdaniel | ip netns command | 16:51 |
| sdaniel | I get only 1 result | 16:51 |
| sdaniel | a qdhcp-..... | 16:51 |
| xgerman_ | try to create a listener | 16:54 |
| *** yamamoto has quit IRC | 16:54 | |
| *** yamamoto has joined #openstack-lbaas | 16:56 | |
| *** yamamoto has quit IRC | 17:01 | |
| johnsom | Yes, after you have created a listener I would expect to see a q-lbaas namespace appear. At that point you know the driver is configured correctly and working | 17:03 |
| *** Alex_Staf has joined #openstack-lbaas | 17:11 | |
| *** armax has joined #openstack-lbaas | 17:11 | |
| *** AlexeyAbashkin has joined #openstack-lbaas | 17:13 | |
| *** Alex_Staf has quit IRC | 17:15 | |
| *** slaweq_ has quit IRC | 17:15 | |
| *** slaweq has joined #openstack-lbaas | 17:16 | |
| *** AlexeyAbashkin has quit IRC | 17:17 | |
| *** slaweq has quit IRC | 17:20 | |
| *** AlexeyAbashkin has joined #openstack-lbaas | 17:37 | |
| *** AlexeyAbashkin has quit IRC | 17:41 | |
| *** slaweq has joined #openstack-lbaas | 17:42 | |
| *** slaweq has quit IRC | 17:52 | |
| *** slaweq has joined #openstack-lbaas | 17:53 | |
| *** slaweq has quit IRC | 17:57 | |
| *** yamamoto has joined #openstack-lbaas | 17:58 | |
| *** yamamoto has quit IRC | 18:08 | |
| *** salmankhan has quit IRC | 18:12 | |
| *** salmankhan has joined #openstack-lbaas | 18:15 | |
| *** slaweq has joined #openstack-lbaas | 18:21 | |
| openstackgerrit | Jude Cross proposed openstack/neutron-lbaas master: [WIP] Remove unnecessary lazy-loaded queries https://review.openstack.org/477698 | 18:25 |
| *** slaweq has quit IRC | 18:26 | |
| *** sshank has joined #openstack-lbaas | 18:30 | |
| *** salmankhan has quit IRC | 18:32 | |
| *** sshank has quit IRC | 18:33 | |
| *** sshank has joined #openstack-lbaas | 18:41 | |
| *** salmankhan has joined #openstack-lbaas | 18:45 | |
| rm_work | johnsom / xgerman_ is Failover API syncronous? | 18:49 |
| rm_work | looks like it? | 18:49 |
| rm_work | not using a handler | 18:49 |
| xgerman_ | No, throws it on the queue | 18:49 |
| rm_work | hmmm | 18:49 |
| rm_work | really? | 18:49 |
| rm_work | so https://review.openstack.org/#/c/523242/4/octavia/controller/worker/controller_worker.py | 18:50 |
| xgerman_ | + failover hads the lowest priority if you run rate limiting… | 18:50 |
| rm_work | johnsom is marking the LB active again immediately after running self._perform_amphora_failover | 18:50 |
| rm_work | oh derp | 18:50 |
| rm_work | i thought i was in the API file | 18:51 |
| rm_work | nm | 18:51 |
| *** tongl has joined #openstack-lbaas | 18:51 | |
| xgerman_ | rm_work: https://review.openstack.org/#/c/524254/ | 19:04 |
| xgerman_ | backport of the VIP port fix | 19:04 |
| rm_work | +A | 19:05 |
| xgerman_ | thanks | 19:05 |
| xgerman_ | once this merges I need to play with my deployment scripts ;-) | 19:05 |
| rm_work | xgerman_: so how do you trgger an *amphora* failover with the API | 19:08 |
| rm_work | or does it only do LB failover right now | 19:08 |
| xgerman_ | LB failover | 19:08 |
| rm_work | k | 19:08 |
| xgerman_ | amphora would be on the octavia API | 19:09 |
| xgerman_ | wheread failing ove ran LB has maing for all kind of vendors | 19:09 |
| rm_work | yeah and this merged before the amp-api didn't it | 19:24 |
| rm_work | k, guess it needs to be added | 19:24 |
| openstackgerrit | Adam Harwell proposed openstack/octavia master: WIP: Floating IP Network Driver (spans L3s) https://review.openstack.org/435612 | 19:27 |
| *** fnaval has quit IRC | 19:28 | |
| *** sshank has quit IRC | 19:29 | |
| *** fnaval has joined #openstack-lbaas | 19:30 | |
| rm_work | would still love to see https://review.openstack.org/#/c/504875/ merged :) | 19:32 |
| rm_work | pretty plox | 19:32 |
| xgerman_ | =A | 19:33 |
| rm_work | ah looks like https://review.openstack.org/#/c/520863/3 needs +A again due to rebase too | 19:33 |
| rm_work | ^_^ | 19:33 |
| rm_work | thanks | 19:33 |
| *** tongl has quit IRC | 19:44 | |
| *** sdaniel has quit IRC | 19:50 | |
| johnsom | I am going to run to lunch, but I plan to keep working on the driver spec the rest of the day. | 19:52 |
| xgerman_ | +1 | 19:53 |
| openstackgerrit | Merged openstack/octavia master: Fix the failover API to not fail with immutable LB https://review.openstack.org/523242 | 20:04 |
| *** AlexeyAbashkin has joined #openstack-lbaas | 20:11 | |
| *** AlexeyAbashkin has quit IRC | 20:15 | |
| *** slaweq has joined #openstack-lbaas | 20:23 | |
| *** pcaruana has quit IRC | 20:25 | |
| *** slaweq has quit IRC | 20:27 | |
| openstackgerrit | Santhosh Fernandes proposed openstack/octavia master: [WIP] L3 ACTIVE-ACTIVE Data model impact https://review.openstack.org/524722 | 20:39 |
| openstackgerrit | Merged openstack/octavia master: Optimize update_health process https://review.openstack.org/504875 | 20:39 |
| openstackgerrit | Merged openstack/octavia master: Clean up test_update_db.py a little bit https://review.openstack.org/520863 | 20:39 |
| openstackgerrit | Santhosh Fernandes proposed openstack/octavia master: [WIP] L3 ACTIVE-ACTIVE Data model impact https://review.openstack.org/524722 | 20:40 |
| *** sanfern has quit IRC | 20:41 | |
| *** sanfern has joined #openstack-lbaas | 20:41 | |
| *** sanfern has quit IRC | 20:42 | |
| *** sanfern has joined #openstack-lbaas | 20:42 | |
| *** sanfern has quit IRC | 20:43 | |
| *** sanfern has joined #openstack-lbaas | 20:43 | |
| *** sanfern has quit IRC | 20:43 | |
| *** sanfern has joined #openstack-lbaas | 20:44 | |
| *** sanfern has quit IRC | 20:44 | |
| *** sanfern has joined #openstack-lbaas | 20:45 | |
| *** sanfern has quit IRC | 20:45 | |
| *** sanfern has joined #openstack-lbaas | 20:45 | |
| *** sanfern has quit IRC | 20:46 | |
| *** sanfern has joined #openstack-lbaas | 20:46 | |
| *** sanfern has quit IRC | 20:47 | |
| *** sanfern has joined #openstack-lbaas | 20:47 | |
| *** sanfern has quit IRC | 20:47 | |
| *** sanfern has joined #openstack-lbaas | 20:48 | |
| *** sanfern has quit IRC | 20:48 | |
| *** sanfern has joined #openstack-lbaas | 20:49 | |
| *** sanfern has quit IRC | 20:49 | |
| *** sanfern has joined #openstack-lbaas | 20:49 | |
| *** sanfern has quit IRC | 20:50 | |
| *** sanfern has joined #openstack-lbaas | 20:50 | |
| *** sanfern has quit IRC | 20:51 | |
| *** sanfern has joined #openstack-lbaas | 20:51 | |
| *** sanfern has quit IRC | 20:51 | |
| *** slaweq has joined #openstack-lbaas | 21:10 | |
| *** AlexeyAbashkin has joined #openstack-lbaas | 21:11 | |
| johnsom | rm_work Way to ping me on lunch... grin. I tried to jump in with some info in infra. Not sure if there was more needed or other discussions. | 21:13 |
| *** AlexeyAbashkin has quit IRC | 21:15 | |
| rm_work | yeah... | 21:18 |
| *** rcernin has joined #openstack-lbaas | 21:22 | |
| openstackgerrit | Adam Harwell proposed openstack/octavia master: WIP: Floating IP Network Driver (spans L3s) https://review.openstack.org/435612 | 21:28 |
| *** salmankhan has quit IRC | 21:30 | |
| *** sshank has joined #openstack-lbaas | 21:38 | |
| *** tongl has joined #openstack-lbaas | 21:42 | |
| *** ianychoi has quit IRC | 21:44 | |
| *** gcheresh has quit IRC | 21:49 | |
| openstackgerrit | Adam Harwell proposed openstack/octavia master: WIP: Floating IP Network Driver (spans L3s) https://review.openstack.org/435612 | 22:18 |
| *** salmankhan has joined #openstack-lbaas | 22:19 | |
| *** bar_ has joined #openstack-lbaas | 22:33 | |
| bar_ | rm_work, hey | 22:33 |
| bar_ | johnsom, were you looking for me yesterday? | 22:35 |
| johnsom | bar_ Seems like a year ago... ha | 22:35 |
| bar_ | ha, I have just read the log from yesterday | 22:36 |
| johnsom | bar_ Oh, nevermind. Adam clarified for me. | 22:37 |
| bar_ | johnsom, pending_create is here to stay, correct? | 22:37 |
| johnsom | Yeah, my bad on that one. | 22:37 |
| bar_ | np | 22:37 |
| *** sshank has quit IRC | 22:38 | |
| bar_ | johnsom, perhaps you would know why this patch (https://review.openstack.org/#/c/522666/) won't verify? | 22:41 |
| bar_ | I got dependency error for the former patch set, the current just isn't handed over to Zuul. so it seems. | 22:42 |
| johnsom | Umm That seems like a zuul bug/issue. There was an announcement about zuul issues this morning. I would try another recheck. | 22:44 |
| bar_ | done | 22:44 |
| johnsom | Hmm, no luck there either. I will ping in the infra channel | 22:45 |
| bar_ | thx! | 22:46 |
| openstackgerrit | Hengqing Hu proposed openstack/octavia master: Support UDP load balance https://review.openstack.org/503606 | 22:48 |
| johnsom | bar_ Yeah, ok, infra says you have a dependency loop, which you kind of do. | 22:50 |
| johnsom | bar_ Depend-On is only used for cross repo dependencies, if it's in the same repo you just use the parent/child relationship in git | 22:51 |
| bar_ | johnsom, ok, I did both to make sure | 22:51 |
| openstackgerrit | Michael Johnson proposed openstack/python-octaviaclient master: Complement Octavia client with a set of features https://review.openstack.org/522666 | 22:51 |
| johnsom | That should fix you up | 22:52 |
| bar_ | ok, great | 22:53 |
| *** slaweq has quit IRC | 22:54 | |
| *** dayou has quit IRC | 23:16 | |
| johnsom | rm_work xgerman Question for you folks. In the provider driver spec, should we be passing python objects over to the driver or JSON documents? I know if we do JSON documents the drivers can use jsonschema to do the validation of features they support or not, etc. | 23:18 |
| xgerman_ | usually I do python->python especially with a map it really doesn’t matter one way or the other | 23:24 |
| openstackgerrit | Adam Harwell proposed openstack/octavia master: WIP: Floating IP Network Driver (spans L3s) https://review.openstack.org/435612 | 23:44 |
| rm_work | if it doesn't have to go over the queue, i'm fine with python objects | 23:47 |
| rm_work | just has to be documented | 23:48 |
| *** fnaval has quit IRC | 23:48 | |
| rm_work | can't count the number of times i've passed slightly the wrong thing <_< | 23:48 |
| xgerman_ | :-) | 23:48 |
| johnsom | Yeah, no queue planned. Just trying to think of good validation methods. I found that voluptuous is in G-R so that is an option | 23:48 |
| rm_work | yeah we use that internally | 23:49 |
| rm_work | i'm used to it kinda | 23:49 |
| johnsom | Ok, I just didn't want to get too far without a few more brains on the topic. | 23:49 |
| johnsom | I think I can get through the LB section today. I may ping for a quick look over to make sure I'm not too far in out field before I tackle the rest of the doc next week. | 23:51 |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!
