Monday, 2019-01-28

« Sunday, 2019-01-27 Index Tuesday, 2019-01-29 »
*** ileixe has joined #openstack-lbaas00:54
*** hongbin has joined #openstack-lbaas01:03
openstackgerritYang JianFeng proposed openstack/octavia master: Add listener and pool protocol validation  https://review.openstack.org/59404001:06
*** Dinesh_Bhor has joined #openstack-lbaas01:36
*** hongbin has quit IRC01:37
*** hongbin has joined #openstack-lbaas01:56
*** sapd1 has joined #openstack-lbaas02:22
*** hongbin has quit IRC02:25
*** ileixe has left #openstack-lbaas02:55
*** yamamoto has joined #openstack-lbaas02:56
*** psachin has joined #openstack-lbaas03:02
*** sapd1 has quit IRC03:39
*** hongbin has joined #openstack-lbaas03:46
*** sapd1 has joined #openstack-lbaas03:55
*** ramishra has joined #openstack-lbaas04:02
*** Dinesh_Bhor has quit IRC04:15
*** Dinesh_Bhor has joined #openstack-lbaas04:16
*** hongbin has quit IRC04:55
*** sapd1 has quit IRC06:32
*** ramishra has quit IRC07:08
*** mkuf_ has joined #openstack-lbaas07:11
*** mkuf has quit IRC07:12
Dinesh_BhorAnyone has all in one node Octavia ansible playbooks?07:26
*** rpittau has joined #openstack-lbaas07:32
*** dayou has quit IRC07:39
*** ramishra has joined #openstack-lbaas07:41
*** dayou has joined #openstack-lbaas07:42
*** gcheresh has joined #openstack-lbaas07:53
*** pcaruana has joined #openstack-lbaas08:01
*** yamamoto has quit IRC09:07
*** ccamposr has joined #openstack-lbaas09:08
*** yamamoto has joined #openstack-lbaas09:11
*** yamamoto has quit IRC09:11
*** yamamoto has joined #openstack-lbaas09:12
*** yamamoto has quit IRC09:16
openstackgerritCarlos Goncalves proposed openstack/octavia stable/queens: Fix prefix for vip_ipv6  https://review.openstack.org/63318309:24
*** yamamoto has joined #openstack-lbaas10:00
*** yamamoto has quit IRC10:06
*** sapd1 has joined #openstack-lbaas10:10
*** psachin has quit IRC10:45
*** salmankhan has joined #openstack-lbaas10:49
*** salmankhan has quit IRC10:57
*** gcheresh has quit IRC11:20
*** Dinesh_Bhor has quit IRC11:32
*** psachin has joined #openstack-lbaas11:42
*** yamamoto has joined #openstack-lbaas12:00
*** sapd1 has quit IRC12:04
*** salmankhan has joined #openstack-lbaas12:15
*** ramishra has quit IRC12:20
*** pcaruana has quit IRC12:36
*** pcaruana has joined #openstack-lbaas12:37
*** yamamoto has quit IRC12:37
*** salmankhan has quit IRC12:46
*** salmankhan has joined #openstack-lbaas12:47
*** salmankhan has quit IRC12:56
*** yamamoto has joined #openstack-lbaas13:05
*** ramishra has joined #openstack-lbaas13:09
openstackgerritMerged openstack/octavia-dashboard master: Imported Translations from Zanata  https://review.openstack.org/63339713:16
*** yamamoto has quit IRC13:19
*** pcaruana has quit IRC13:32
*** gcheresh has joined #openstack-lbaas13:37
*** yamamoto has joined #openstack-lbaas13:40
*** yamamoto has quit IRC13:40
*** yamamoto has joined #openstack-lbaas13:40
*** psachin has quit IRC13:41
*** yamamoto has quit IRC13:45
iceyis octavia expected to work with IPV6 addresses for the public IP? I just got a nice traceback when trying to create a loadbalancer on a V6 subnet13:49
iceybasically: `pymysql.err.DataError: (1406, "Data too long for column 'ip_address' at row 1")`13:49
*** pcaruana has joined #openstack-lbaas13:50
*** trown|outtypewww is now known as trown14:03
*** fnaval has joined #openstack-lbaas14:56
xgermanicey: we have tests for that - so Iam surprised there is a traceback14:57
iceyxgerman: I don't see that issue when using an ipv4 address14:57
*** kaiokmo has quit IRC15:03
johnsomicey: what version are you using? Queens patches haven’t merged yet. Rocky and Stein are heavily tested.15:31
iceyjohnsom: rocky, from the Ubuntu Cloud Archive (actually, proposed)15:31
iceyjohnsom: I'm trying it out in my lab15:31
johnsomHmm, I wonder if that doesn’t have the patches yet. When I get in the office I will check that we have cut the Rocky release with the patches.15:33
johnsomIf not, I can probably get it on pypi in a day or two.15:35
*** sapd1 has joined #openstack-lbaas15:37
*** openstackgerrit has quit IRC15:51
iceyjohnsom: I _seem_ to have all of the migrations that are on github16:02
iceyjohnsom: https://pastebin.ubuntu.com/p/pG735Mxb2k/16:08
johnsomicey Yeah, ok, there is still an outstanding ipv6 bug for the VIP. I see it. For members we have a 64 byte ip address column, but for the VIP it is only 36, which is too short for a fully expanded IPv6 address. I am guessing the reserved IP address space we are using in the gate tests have shortened blocks (zeros removed).16:19
iceyjohnsom: \o/16:21
johnsomI will get on fixing that today.16:21
johnsomI don't think we can backport that fix though. It would need to be a local change.16:22
johnsomWould you like the SQL to fix the DB?16:22
*** sapd1 has quit IRC16:25
johnsomOk, so this is why I am super confused: https://github.com/openstack/octavia/blob/master/octavia/db/migration/alembic_migrations/versions/82b9402e71fd_update_vip_address_size.py16:39
johnsomSomething must have stepped on this16:39
johnsomOh, great, it was just written wrong16:42
*** pcaruana has quit IRC16:47
*** KeithMnemonic has joined #openstack-lbaas17:00
*** gcheresh has quit IRC17:00
*** reedip has joined #openstack-lbaas17:05
reedipjohnsom : hi, can you please share your opinions on https://review.openstack.org/621493 and https://review.openstack.org/598476 ? They are octavia-tempest-plugin patches, and might be useful for the provider driver tests17:06
johnsomOk17:07
reedipthanks a lot johnsom :)17:07
reedipjohnsom: also, can I ask one question?17:07
johnsomOf course17:07
*** dims has quit IRC17:08
reedipjohnsom: currently the octavia has a common/utils section, which uses the configuration file to fetch some information. I wanted to know if there is a chance to mograte the utils section to octavia-lib?17:08
reedipIMO they can be , but wanted to know your thoughts17:10
reediphttps://github.com/openstack/octavia/blob/master/octavia/common/utils.py for reference17:11
johnsomI think everything except "get_network_driver" probably can be moved over to octavia-lib if it is useful17:11
reedipjohnsom : how can we use get_network_driver from Octavia?17:11
johnsomAs it is now, but that is not valid for octavia-lib17:12
reedipoctavia isnt there in the global requirements17:12
*** rpittau has quit IRC17:12
johnsomYes, and it should not be17:12
reedipjohnsom : yeah, I understood that part, since its tightly coupled with the config section17:12
johnsomIt is not a library, but a service17:12
johnsomAnd the pbr setup entry points17:12
reedipjohnsom : agreed.... so for the OVN driver, should we use the PBR entry points and read the configuration ?17:13
johnsomI think if your driver needs to talk directly to neutron, it will need it's own way of interacting with neutron.  Or we figure out a octavia-lib callback that does it on your behalf.17:14
johnsomWhat do you need to go directly to neutron for?17:15
reedipjohnsom : currently OVN is using the octavia's get_network_driver function from the utils . among some other util functions as well. Since the OVN driver is closely dependent on Octavia, I think a callback might be a great option17:16
reedipjohnsom: since the driver belongs to Octavia, should it talk directly to neutron?  I think Octavia can abstract it17:16
johnsomWe can, but I don't know that we necessarily need to restrict drivers from making their own OpenStack calls if they need to. They just need to handle that on their own.17:18
johnsomWhat network driver calls are you making?17:18
*** openstackgerrit has joined #openstack-lbaas17:19
openstackgerritCarlos Goncalves proposed openstack/python-octaviaclient master: Add --provider to load balancer create  https://review.openstack.org/63356217:19
reedipjohnsom: getting the network driver information from octavia using the get_network_driver function and then using it to get network, subnet information17:19
cgoncalvesreedip, ^ this will "fix" the LB create command in https://docs.openstack.org/networking-ovn/latest/admin/loadbalancer.html where you have --provider although it doesn't exist  yet ;)17:19
reedipcgoncalves: I AM REVIEWING THE PATCH AND I LIKE IT :D17:20
cgoncalvesthat was... BOLD :)17:20
johnsomreedip I think I need to look at your driver and understand better what you are using our network driver for. Then I can have a better idea of which approach is best.17:22
johnsomreedip I will note that calling that violates the provider driver documentation......  https://docs.openstack.org/octavia/latest/contributor/guides/providers.html#driver-support-library See the big RED shiny boxes.....17:23
johnsomhttps://docs.openstack.org/octavia/latest/contributor/guides/providers.html#stable-provider-driver-interface17:23
reedipjohnsom : interesting :)17:24
reedipjohnsom : okay, I got your point, that the implementation needs correction here17:25
johnsomJust saying.....  But we can certainly work through coming up with a better solution.....17:25
reediphttps://github.com/openstack/networking-ovn/blob/master/networking_ovn/octavia/ovn_driver.py#L320 is a sample of where we are directly importing Octavia17:25
reediphowever, I think octavia-lib is out of context and can be safely imported here :)17:25
johnsomYeah, but that patch hasn't merged yet. Those are the original guidelines....17:26
johnsomSo if you just need the VIP subnet, network, and port IDs, we give you those in the LB information already.....17:29
johnsomOr your driver is responsible for creating the port from the ID info they pass in....17:30
reedipjohnsom : yeah, port creation and deletion for VIP is needed17:32
reedipso we need the Octavia Network Driver , which gives the callback to Neutron17:32
johnsomI need to read all of your code. I'm not sure you need to do that.17:33
reedipjohnsom : sure,the code is there to review....17:33
reedipjohnsom: whenever you get time, you can let me know your opinion17:34
*** trown is now known as trown|lunch17:42
*** ccamposr has quit IRC17:51
openstackgerritMichael Johnson proposed openstack/octavia master: Fix VIP address DB storage size to support IPv6  https://review.openstack.org/63357218:06
johnsomSigh, a mistake three years ago caught up with us....18:06
openstackgerritMichael Johnson proposed openstack/octavia master: Fix VIP address DB storage size to support IPv6  https://review.openstack.org/63357218:10
*** yamamoto has joined #openstack-lbaas18:14
iceyjohnsom: no worries on not backporting, completely understand; I was logging back on (EU time here) to ask if you had a bug ref so I could take a poke at fixing it in my AM :-)18:21
johnsomicey Already posted: https://review.openstack.org/63357218:23
johnsomIt is a bummer that the fix for this three years ago was wrong and didn't get caught. However, I have added more tests now to make sure this doesn't happen again.18:23
iceythanks johnsom  :)18:23
iceyjohnsom: sometimes it takes somebody trying to use it in anger to find those edges :-) (which is actually why I'm trying it out in my lab)18:24
johnsomYeah, the last year or so has been a big push by the team to improve our testing coverage.  Still some work to do....18:24
iceyjohnsom: indeed! I'm just glad to see it coming along!18:25
iceyjohnsom: I don't see that down_revision on github?18:30
icey(I'm trying to apply your patch directly in my lab before I +1 it)18:31
johnsomicey It is off this patch: https://review.openstack.org/#/c/621323/31/octavia/db/migration/alembic_migrations/versions/211982b05afc_add_flavor_id_to_lb.py18:31
iceythanks johnsom18:31
johnsomThere were two DB revisions already in flight and close to merge, so I branched it there.18:31
johnsomBasically I didn't want to change and rebase 20 some patches, so based it on a reasonable spot in that chain18:33
*** trown|lunch is now known as trown18:36
iceysounds reasonable :)18:36
iceylgtm johnsom :)18:43
*** ramishra has quit IRC18:53
*** reedip has quit IRC18:58
*** yamamoto has quit IRC19:06
*** yamamoto has joined #openstack-lbaas19:08
*** yamamoto has quit IRC19:09
*** yamamoto has joined #openstack-lbaas19:09
*** yamamoto has quit IRC19:09
*** yamamoto has joined #openstack-lbaas19:10
*** yamamoto has quit IRC19:14
*** dims has joined #openstack-lbaas19:15
*** aojea has joined #openstack-lbaas19:59
*** yboaron has quit IRC20:09
*** yboaron has joined #openstack-lbaas20:09
*** aojea has quit IRC20:12
*** yboaron has quit IRC20:15
*** aojea has joined #openstack-lbaas20:20
*** aojea has quit IRC20:20
*** aojea has joined #openstack-lbaas20:20
*** fnaval has quit IRC20:37
openstackgerritAdam Harwell proposed openstack/octavia master: Experimental multi-az support  https://review.openstack.org/55896221:00
openstackgerritAdam Harwell proposed openstack/octavia master: WIP: Floating IP Network Driver (spans L3s)  https://review.openstack.org/43561221:11
colin-is there anything more human readable on the topic of quota management for octavia? https://developer.openstack.org/api-ref/load-balancer/v2/index.html?expanded=show-project-quota-detail,update-a-quota-detail,list-quota-detail#list-quota21:19
colin-my google-fu fails me21:19
colin-based on that i was hoping to provide an project arg to the api when i listed LBs but alas it does not like that21:20
johnsomNot really, the API reference is the bulk of the quota docs right now.  So you want to list the quotas for a specific project?21:21
johnsomThe curl example shows that: curl -X GET -H "X-Auth-Token: <token>" http://198.51.100.10:9876/v2/lbaas/quotas?project_id=e3cd678b11784734bc366148aa37580e21:21
johnsomhttps://developer.openstack.org/api-ref/load-balancer/v2/index.html?expanded=list-quota-detail#id16521:21
colin-yeah, will proceed with the model that describes, that will work with a valid token i'm sure21:22
johnsomYeah, if you aren't an admin, you only get your quota details21:22
colin-looking at a sane default implementation of quota and just sizing it up, listing was just an effort to confirm resources appear how i imagine them21:22
colin-i'll proceed with that for now, thx21:23
johnsomcolin- You might be looking for this: https://review.openstack.org/55754821:24
colin-oh cool, this looks promising yeah21:26
colin-will dig in21:26
rm_workyeah i kinda stopped caring about that after moving companies21:29
rm_workbut i could clean it up and try to get it merged if someone else wants it21:29
rm_worki lost a lot of motivation after apparently *no one else cares* about that kind of info, lol21:29
rm_workevery time i look at this flavor chain, my eyes start to blur21:39
johnsomMe too, it's too long.... 21 patches21:39
johnsomWe need to start merging!  grin21:40
rm_work... yeah >_>21:43
*** aojea has quit IRC22:03
*** rcernin has joined #openstack-lbaas22:13
colin-do you run it anywhere atm rm_work? just curious22:13
rm_workwe did, yes22:13
rm_workbut i haven't looked at it in a few months (because yeah, switched jobs)22:14
colin-understandable22:14
*** rcernin has quit IRC22:15
*** rcernin has joined #openstack-lbaas22:15
*** aojea has joined #openstack-lbaas22:20
rm_workeiddccidnvgegkkcvjlrebudgvnetdcjlveilntfujdj22:22
rm_workgraaahh i am starting tor really dislike the way this yubikey works, lol22:22
johnsomlol22:23
johnsomNow we just need to guess which key it is....22:23
rm_workisn't it just kinda like... randomness or something? <_< I don't know how they work22:23
johnsomIt depends on what mode it's setup for22:24
*** aojea has quit IRC22:24
*** aojea has joined #openstack-lbaas22:25
colin-will the amphora jinja template make any determinations wrt core availability? will the haproxy cfg leverage more than one core by default?22:25
johnsomcolin- I have been sitting on that for a number of reasons:22:26
johnsom1. The multi-core code in haproxy was broken and limited.  This is fixed in 1.8 (really want 1.9) with the new threads support.22:26
johnsom2. One core can push 14Gbps, and it hasn't been clear that people really need to push it hard enough to look at the multi-threaded stuff, so it dropped in priority.22:27
johnsom3. We needed flavors to make it make sense22:27
johnsom4. CPU pinning is also reasonably important in a multi-vcpu amp, which gets a bit complicated. I.e. pin the nics to cores, haproxy to others, etc.22:28
johnsomSo, basically my next step towards that was doing the flavors work. The first part is this: https://review.openstack.org/#/c/631906/22:29
*** aojea has quit IRC22:29
colin-understood, appreciate the detail22:30
johnsomDefinitely on the roadmap, has a plan, just needs some resources, etc.22:30
johnsomI think the really drivers are conncurrent connection counts and then the TLS offload load.22:30
johnsomcolin- It would be good feedback for prioritization if you do have applications/needs that would use a multi-vcpu amp.22:35
colin-ok, not yet but if that changes i'll communicate it22:36
colin-really encouraged by the nova flavor change :)22:36
johnsomUgh, git is not being my friend. Ok, rebases22:39
openstackgerritMichael Johnson proposed openstack/octavia-tempest-plugin master: Add the flavor profile service client.  https://review.openstack.org/63040522:39
openstackgerritMichael Johnson proposed openstack/octavia-tempest-plugin master: Add the flavor service client.  https://review.openstack.org/63040722:40
openstackgerritMichael Johnson proposed openstack/octavia-tempest-plugin master: Add the provider service client.  https://review.openstack.org/63040822:40
openstackgerritMichael Johnson proposed openstack/octavia-tempest-plugin master: Adds flavor profile API tests  https://review.openstack.org/63041122:40
openstackgerritMichael Johnson proposed openstack/octavia-tempest-plugin master: Adds flavor API tests  https://review.openstack.org/63080422:40
openstackgerritMichael Johnson proposed openstack/octavia-tempest-plugin master: Adds provider API tests  https://review.openstack.org/63110522:40
openstackgerritMichael Johnson proposed openstack/octavia-tempest-plugin master: Add the provider flavor capabilities service client.  https://review.openstack.org/63110722:40
openstackgerritMichael Johnson proposed openstack/octavia-tempest-plugin master: Adds provider flavor capabilities API tests  https://review.openstack.org/63111322:40
openstackgerritMichael Johnson proposed openstack/octavia-tempest-plugin master: Add a flavor to the load balancer CRUD scenarios  https://review.openstack.org/63135322:40
openstackgerritMichael Johnson proposed openstack/octavia-tempest-plugin master: Add amphora update service client and API test  https://review.openstack.org/63329522:41
johnsomHmmm, I expected one of those to merge conflict given the git issues I see local. Well, time to rebase the local patch and see what happens.22:41
openstackgerritMichael Johnson proposed openstack/octavia-tempest-plugin master: Add amphora failover API test  https://review.openstack.org/63361422:43
johnsomThere we go... Now did it push everything it should have....22:43
johnsomYep, groovy22:44
*** jiteka has quit IRC22:44
johnsomOne more onto the chain...22:45
*** jiteka has joined #openstack-lbaas22:46
*** yamamoto has joined #openstack-lbaas23:49
rm_workjohnsom: why is this so shitty? https://review.openstack.org/#/c/618322/18/api-ref/source/v2/examples/flavorprofile-show-response.json23:52
rm_workdo we like pain?23:52
johnsomBecause it's an embedded json document?23:52
rm_workwhat's the point of embedding it?23:53
rm_workwhy not just let them put json in?23:53
rm_workand let the code take care of storing it appropriately?23:53
johnsomThen you are flip flopping it in the controller23:53
rm_workso we force people to manually create weird escaped json strings? lol23:54
johnsomNo, the CLI handles it for you23:54
*** yamamoto has quit IRC23:54
rm_worki guess that's fair, probably uncommon to be hitting the API with curl23:54
rm_workit just happens to be our examples23:54
rm_workin which ... wat23:54
johnsomhttp://logs.openstack.org/42/632842/6/check/openstack-tox-docs/bfeea42/html/admin/flavors.html#flavor-profiles23:55
openstackgerritGerman Eichberger proposed openstack/octavia master: Refactor the unplugging of the VIP  https://review.openstack.org/61368523:58
openstackgerritMichael Johnson proposed openstack/python-octaviaclient master: Adds loadbalancer amphora configure to OSC  https://review.openstack.org/63362623:59
« Sunday, 2019-01-27 Index Tuesday, 2019-01-29 »

Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!