Tuesday, 2020-06-23

*** TMM has quit IRC00:23
*** TMM has joined #openstack-lbaas00:23
sorrisonrm_work johnsom: going to get into some more octavia work, I'll take a look at that AZ scenario tempest test finally too!01:07
sorrisonjohnsom: I've been looking at running tempest tests against our octavia in production and hoping to make some changes to make this easier.01:08
sorrisoncurrently the tempest tests require you to set up admin creds and I'm hoping to change it so they don't need admin creds to run tests that don't need admin creds01:09
sorrisonThe other openstack projects allow this (some after patches from us to make this happen)01:09
rm_worksorrison: also need to finish the real AZ tests that create a true working new AZ/network and use that to do a traffic test01:32
rm_workI don't know enough on the networking OR the nova side to get that working in devstack, really :(01:32
sorrisonyip, hope to look at that soon too01:40
*** stingrayza has quit IRC01:51
*** stingrayza has joined #openstack-lbaas01:53
*** rcernin has quit IRC02:41
*** rcernin has joined #openstack-lbaas02:42
*** shtepanie has quit IRC02:53
sorrisonrm_work: Just wondering if there is a reason octavia doesn't use the standard member and admin roles in default policy and defines it own?03:03
sorrisonis there plans to move to the standard roles, which also defines reader which would line up with load-balancer_observer I think03:04
*** vishalmanchanda has joined #openstack-lbaas03:15
*** dmellado has quit IRC03:59
*** wuchunyang has joined #openstack-lbaas04:02
*** wuchunyang has quit IRC04:07
*** wuchunyang has joined #openstack-lbaas04:18
*** wuchunyang has quit IRC04:23
*** wuchunyang has joined #openstack-lbaas05:11
*** wuchunyang has quit IRC05:15
*** strigazi has quit IRC05:18
*** strigazi has joined #openstack-lbaas05:19
*** wuchunyang has joined #openstack-lbaas05:21
*** gcheresh has joined #openstack-lbaas05:21
*** wuchunyang has quit IRC05:26
*** wuchunyang has joined #openstack-lbaas05:41
*** wuchunyang has quit IRC05:48
johnsomsorrison Our policy role definition is based on the original specification that was more detailed. I think only nova and octavia implemented the "advanced RBAC" as it was originally intended. The new "spec" is a simplified version of that, which is fairly compatible as you mentioned. However, I think these simplified roles should be an option and not reduce the capability we have shipped for many releases. Like06:09
johnsomwe provide a legacy policy file for the old "owner or admin" I would expect we could provide a dumbed down policy file as an option that implements the "reader"/"admin" role.06:09
sorrisonjohnsom: it essentially just comes down to the name of the roles I think06:10
sorrisonit could work out of the box if it used the standard roles06:11
sorrisonand still have all the nice policy that octavia has06:11
sorrisonThere is one difference I think in that octavia defines a non global read only user06:14
sorrisonThis then feeds over to tempest and that is hard to use as it could instead use the standard primary, alt and admin users06:15
sorrisonI have been working on a patch on our environment to allow us to run tempest as not specify admin credentials https://github.com/NeCTAR-RC/octavia-tempest-plugin/commit/c1d88eb502f180c9edbf4c3b4f66b7ccfd57e15c06:17
sorrisonI'm wondering if this can go upstream, it would be backwards compatible if octavia used standard roles06:18
*** rpittau|afk is now known as rpittau06:34
*** born2bake has joined #openstack-lbaas06:35
*** maciejjozefczyk has joined #openstack-lbaas07:08
openstackgerritOmer Schwartz proposed openstack/octavia-tempest-plugin master: Tags are strings attached to an entity with the purpose of classification into groups. Tags support API tests were added.  https://review.opendev.org/73593907:42
*** aannuusshhkkaa has quit IRC07:51
openstackgerritMerged openstack/octavia master: Clarify health monitor behavior  https://review.opendev.org/73519108:06
*** rcernin has quit IRC08:18
openstackgerritCarlos Goncalves proposed openstack/octavia master: Add missing reload method in amphora noop driver  https://review.opendev.org/73746008:59
fricklerthis sdk test keeps failing with some lb error, would be great if someone could take a look https://zuul.opendev.org/t/openstack/build/110222aba61c450d8fac7c56ebc8d33009:05
openstackgerritGregory Thiemonge proposed openstack/octavia-tempest-plugin master: DNM check UDP pool fix  https://review.opendev.org/73728309:07
cgoncalvesfrickler, https://review.opendev.org/73746009:08
fricklercgoncalves: ah, great, thx09:10
*** gcheresh has quit IRC09:31
*** gcheresh has joined #openstack-lbaas09:32
*** ccamposr has quit IRC10:10
*** rpittau is now known as rpittau|bbl10:12
*** ccamposr has joined #openstack-lbaas10:16
*** tkajinam has quit IRC10:17
*** gcheresh has quit IRC10:55
*** wuchunyang has joined #openstack-lbaas10:56
*** wuchunyang has quit IRC10:59
*** ataraday_ has joined #openstack-lbaas11:16
*** gcheresh has joined #openstack-lbaas11:32
*** rpittau|bbl is now known as rpittau12:14
openstackgerritCarlos Goncalves proposed openstack/octavia master: Add amphora image tag capability to Octavia flavors  https://review.opendev.org/73752812:57
*** ccamposr has quit IRC13:12
rm_workhmm somehow i thought we already did that14:02
rm_workoh image TAG not image ID14:02
rm_workerr, or did we have either14:02
johnsomI know that I was going to do it, but never got time to.14:09
*** ccamposr has joined #openstack-lbaas14:13
cgoncalvesthought we'd be nice to our openstack multi-arch SIG14:17
*** riuzen has joined #openstack-lbaas14:24
*** TrevorV has joined #openstack-lbaas14:24
*** riuzen has quit IRC14:28
*** yamamoto has quit IRC14:45
*** yamamoto has joined #openstack-lbaas14:50
*** yamamoto has quit IRC15:18
*** maciejjozefczyk has quit IRC15:50
*** yamamoto has joined #openstack-lbaas15:57
*** rpittau is now known as rpittau|afk16:00
*** yamamoto has quit IRC16:07
openstackgerritMichael Johnson proposed openstack/octavia master: Add an upgrade check for YAML oslo policy file  https://review.opendev.org/73757116:07
openstackgerritMichael Johnson proposed openstack/octavia master: Add an upgrade check for YAML oslo policy file  https://review.opendev.org/73757116:11
*** dayou_ has quit IRC16:27
cgoncalvesCI is not having a good day... POST_FAILURE everywhere16:29
johnsomYep, zuul is broken16:30
johnsomWell, the storage backing zuul16:30
cgoncalvesreading #openstack-infra, seems a fix has merged. recheck should do16:30
*** gcheresh has quit IRC16:37
*** dayou_ has joined #openstack-lbaas16:41
*** dayou_ has quit IRC16:47
*** dayou_ has joined #openstack-lbaas17:00
*** spatel has joined #openstack-lbaas17:09
openstackgerritGregory Thiemonge proposed openstack/octavia-tempest-plugin master: DNM check UDP pool fix  https://review.opendev.org/73728317:11
*** ccamposr has quit IRC17:42
*** gcheresh has joined #openstack-lbaas17:42
rm_workFYI I'm back to working on the failover halt threshold thing17:50
rm_workgonna try to get that workable this week if possible17:50
rm_worki forget exactly what needed done... i feel like it maybe wasn't too far?17:51
johnsomI think there was a problem with it coming back, not sure, it's been too long17:51
*** servagem has quit IRC18:01
openstackgerritCarlos Goncalves proposed openstack/octavia master: Remove amp_image_id option  https://review.opendev.org/73759118:07
*** servagem has joined #openstack-lbaas18:10
*** gcheresh has quit IRC18:12
*** spatel has quit IRC18:29
*** aannuusshhkkaa has joined #openstack-lbaas19:08
cgoncalvesrm_work, could you please review https://review.opendev.org/#/c/737460/ (small patch)? it's a gate blocker for openstacksdk19:26
*** gcheresh has joined #openstack-lbaas19:47
*** yamamoto has joined #openstack-lbaas20:07
*** servagem has quit IRC20:09
*** yamamoto has quit IRC20:11
*** also_stingrayza has joined #openstack-lbaas20:35
*** stingrayza has quit IRC20:37
*** vishalmanchanda has quit IRC20:54
*** shtepanie has joined #openstack-lbaas21:11
shtepaniejohnsom are you available to discuss something right now?21:12
shtepaniewe (aannuusshhkkaa and rm_work) are looking at the second part of the interface refactor. we're looking at the update_health functions and we're second guessing if it's possible to do, and if so, what work could be done on it?21:14
shtepanieit may not need to be a driver?21:14
johnsomAre you talking about the "status" part of the health manager?21:15
rm_workit seems like most of what it's doing is ... very integral to the system working at all21:15
rm_workit does way more than just update the DB with statuses too -- it even deletes amps21:16
johnsomYeah, I don't think it needs to be a driver really. At least not now. The Octavia control plane/API are responsible for managing the status of objects.21:16
rm_workyeah, it's kinda core, not a third-party thing21:16
rm_workBUT, i guess possibly the driver exists because we rely on being able to no-op it?21:16
johnsomYeah, it does. In the future it will probably need to do notifications too, but we can ignore that for now.21:16
rm_workyeah I assume the notifications task can handle that21:17
johnsomUmm, hmm, I don't *think* we can no-op out just the status code.21:17
*** gcheresh has quit IRC21:17
johnsomWe are driver crazy, but maybe not *that* crazy21:18
rm_worki mean it actually already does?21:18
rm_workneed to see if we actually set it as no-op in gates21:18
rm_workbut there already is a "logging" driver for both21:18
johnsomYeah, I thought it was bundled with stats. So you swap out all or none for logging21:19
johnsom(I think logging might be the no-op there)21:19
johnsomSo, I guess what I am getting at is I think it would be fine to split the two. Make metrics driver based and make status processing static.21:20
rm_workhowever thinking about it, I believe that in a no-op testing situation, by virtue of no-oping the compute driver, there is never any health message, and therefore that code is unreachable anyway21:21
johnsomObviously status has to stay behind the thread/process pool, but doesn't rally need to be loaded/dynamic21:21
rm_workso maybe it doesn't NEED a noop21:21
rm_workk i think that should be an acceptable approach21:21
rm_workand lets us get rid of the other kinda broken interface bits21:21
rm_workthanks :D21:22
openstackgerritMerged openstack/octavia master: Add missing reload method in amphora noop driver  https://review.opendev.org/73746021:28
*** TrevorV has quit IRC21:35
*** TMM has quit IRC21:54
*** TMM has joined #openstack-lbaas21:54
*** born2bake has quit IRC22:41
*** born2bake has joined #openstack-lbaas22:50
*** tkajinam has joined #openstack-lbaas22:53
*** rcernin has joined #openstack-lbaas23:02
*** rcernin has quit IRC23:08
*** rcernin has joined #openstack-lbaas23:08
*** yamamoto has joined #openstack-lbaas23:10
*** born2bake has quit IRC23:13
openstackgerritMichael Johnson proposed openstack/octavia master: Fix devstack cleanup keypair delete  https://review.opendev.org/73765723:13
*** yamamoto has quit IRC23:42
*** armax has quit IRC23:59

Generated by irclog2html.py 2.17.2 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!