*** edmondsw has joined #openstack-placement | 00:07 | |
*** edmondsw has quit IRC | 00:12 | |
*** mriedem has quit IRC | 00:29 | |
*** edmondsw has joined #openstack-placement | 01:29 | |
openstackgerrit | Tetsuro Nakamura proposed openstack/nova master: Use common functions in granular fixture https://review.openstack.org/588113 | 01:39 |
---|---|---|
*** edmondsw has quit IRC | 01:50 | |
*** edmondsw has joined #openstack-placement | 01:51 | |
*** edmondsw has quit IRC | 01:55 | |
openstackgerrit | Tetsuro Nakamura proposed openstack/nova master: Use common functions in NonSharedStorageFixture https://review.openstack.org/588114 | 02:01 |
*** e0ne has joined #openstack-placement | 02:23 | |
*** e0ne has quit IRC | 02:23 | |
openstackgerrit | Merged openstack/nova master: Make ResourceTracker.stats node-specific https://review.openstack.org/587636 | 02:30 |
openstackgerrit | Vishakha Agarwal proposed openstack/nova master: 'Updated_at' is NULL when show aggregate info https://review.openstack.org/580271 | 03:13 |
*** edmondsw has joined #openstack-placement | 03:19 | |
*** edmondsw has quit IRC | 03:24 | |
openstackgerrit | Zhenyu Zheng proposed openstack/nova master: Fix a typo in comment in resource_provider.py https://review.openstack.org/588145 | 06:26 |
openstackgerrit | huanhongda proposed openstack/nova master: Destroy evacuated instance while unset nova-compute forced_down https://review.openstack.org/587807 | 06:58 |
openstackgerrit | Tetsuro Nakamura proposed openstack/nova master: Increase max_unit in placement test fixture https://review.openstack.org/588158 | 07:14 |
openstackgerrit | Tetsuro Nakamura proposed openstack/nova master: Refactor AllocationFixture in placement test https://review.openstack.org/588159 | 07:14 |
*** avolkov has joined #openstack-placement | 08:18 | |
*** tssurya has joined #openstack-placement | 08:26 | |
*** cdent has joined #openstack-placement | 08:49 | |
*** tetsuro__ has joined #openstack-placement | 08:59 | |
*** tetsuro__ has quit IRC | 08:59 | |
*** tetsuro_ has joined #openstack-placement | 08:59 | |
*** tetsuro_ has quit IRC | 09:02 | |
*** tetsuro_ has joined #openstack-placement | 09:04 | |
*** tetsuro_ has quit IRC | 09:04 | |
cdent | morning gibi , thanks for getting all that stuff merged | 09:14 |
gibi | cdent: those were good stuff, worth to merge :) | 09:14 |
*** fanzhang has joined #openstack-placement | 09:23 | |
*** edmondsw has joined #openstack-placement | 09:48 | |
openstackgerrit | Merged openstack/nova master: Add note about reschedules and num_attempts in filter_properties https://review.openstack.org/582412 | 10:36 |
openstackgerrit | Stephen Finucane proposed openstack/nova master: tox: Ensure reused envdirs share the same deps https://review.openstack.org/588207 | 10:37 |
openstackgerrit | Chen proposed openstack/nova master: Make nova-manage db archive_delete_rows take --all-cells https://review.openstack.org/587858 | 10:48 |
openstackgerrit | huanhongda proposed openstack/nova master: Destroy evacuated instance while unset nova-compute forced_down https://review.openstack.org/587807 | 10:52 |
*** cdent has quit IRC | 11:59 | |
openstackgerrit | Merged openstack/nova master: Hook resource_tracker to remove stale node information https://review.openstack.org/587922 | 12:34 |
*** s10 has joined #openstack-placement | 12:44 | |
*** tssurya has quit IRC | 12:51 | |
openstackgerrit | Eric Fried proposed openstack/nova master: doc: fix resize user guide link https://review.openstack.org/588097 | 12:52 |
openstackgerrit | Stephen Finucane proposed openstack/nova master: Remove unused stubbing function from test https://review.openstack.org/586968 | 12:53 |
*** tssurya has joined #openstack-placement | 12:54 | |
*** mriedem has joined #openstack-placement | 12:57 | |
*** cdent has joined #openstack-placement | 13:02 | |
*** jaypipes has joined #openstack-placement | 13:02 | |
openstackgerrit | Balazs Gibizer proposed openstack/nova master: Improve NeutronFixture and remove unncessary stubbing https://review.openstack.org/587412 | 13:15 |
*** tetsuro_ has joined #openstack-placement | 13:20 | |
*** tetsuro_ has quit IRC | 13:24 | |
*** cdent has quit IRC | 13:27 | |
openstackgerrit | Merged openstack/nova stable/queens: Call generate_image_url only for legacy notification https://review.openstack.org/584969 | 14:00 |
*** cdent has joined #openstack-placement | 14:02 | |
efried | cdent jaypipes: I'd like to draw your attention to https://review.openstack.org/#/c/588091/ and https://review.openstack.org/#/c/588094/ which are two more approaches to eliminating the duplicate placement calls on every periodic. I made no attempt to fix unit & functional tests; but as you can see the tempest/devstack tests are green (except for one, not sure if related, rechecking to find out). | 14:02 |
efried | cdent: If you have a way of slotting these into your instrumentation or whatever to see if they do in fact dedup the placement calls, that would be handy. | 14:04 |
jaypipes | efried: ack. will get to those shortly (just wrapping up reviews on the patches below there) | 14:04 |
cdent | efried: my instrumentation is: run devstack, look at the logs for placement | 14:04 |
efried | cool. I'll wait to rebase since you're still reviewing. Sounds like you're going to want to solve @safe_connect first. | 14:04 |
jaypipes | efried: yeah | 14:05 |
efried | jaypipes: IMO that should be done entirely separately. I had a patch that was working on that. I'll go back to it... | 14:06 |
jaypipes | efried: I like your approach of fixing this one method at a time. the problem with "fixing" safe_connect in a single patch is that the patch ends up being YUGE and you will end up running into a boatload of failures in the functional tests that are being hidden by safe_connect()... | 14:07 |
jaypipes | (I know because I've tried this multiple times already) | 14:07 |
efried | jaypipes: Yes indeed. But now I'm not sure what you're asking for anymore. | 14:07 |
efried | jaypipes: FYI, this was my latest attempt: https://review.openstack.org/#/c/584593/ | 14:08 |
jaypipes | efried: I like your approach, I just don't like the strict_get() thing (I feel it's just being used to gather all the safe_connect() failures up into a single PlacementConnectFailure exception. | 14:08 |
efried | Exactly true ^ | 14:08 |
efried | jaypipes: It was a way to fix things one call stack at a time, and eventually strict_get would be eliminated and the PlacementConnectFailure conversion would be folded into @safe_connect itself. | 14:09 |
jaypipes | efried: but why not change safe_connect's exceptions to inherit from a single exception class (PlacementAPIConnectFailure or whatevs) and then simply add patches that remove @safe_connect decorator from one method at a time, with the call site catching PlacementAPIConnectFailure and logging the error (or re-raising) as appropriate? | 14:09 |
jaypipes | that way there's no need for a wrapper like strict_get() | 14:10 |
efried | jaypipes: You mean making a separate, transitional @safe_connect decorator? | 14:10 |
jaypipes | no, I mean just removing it entirely and allowing the PlacementAPIConnectFailure to raise. | 14:10 |
efried | There is no PlacementAPIConnectFailure. | 14:11 |
jaypipes | removing it entirely *from one method of report client at a time*, that is.. | 14:11 |
jaypipes | efried: no, I'm saying create that exception, have all the existing exceptions that the @safe_connect decorator currently catches inherit from that | 14:11 |
efried | You mean wrapping .request() with a try/except that converts all those exceptions to PlacementAPIConnectFaliure... | 14:11 |
jaypipes | no.. | 14:11 |
efried | Well, we can't make ksa exceptions inherit from PlacementAPIConnectFailure... | 14:11 |
efried | we need something that converts those. | 14:12 |
jaypipes | sure, that's fine. | 14:12 |
efried | ...which would either be a different/fixed @safe_connect or a catch-all in our .request() in the report client. | 14:13 |
jaypipes | no... | 14:13 |
efried | the former (a fixed @safe_connect) is what I'm trying with https://review.openstack.org/#/c/584593/ | 14:13 |
jaypipes | one sec, lemme throw a gist up | 14:13 |
efried | ight | 14:13 |
efried | gibi: I don't see a _stop_compute(). Is there one, or do I have to make it up? | 14:17 |
efried | perhaps it's as simple as compute.stop(); compute.start() | 14:19 |
efried | yeah, that seems to work just fine. | 14:22 |
openstackgerrit | Jay Pipes proposed openstack/nova master: DNM - example https://review.openstack.org/588295 | 14:31 |
jaypipes | efried: ^^ no tests, not verified, just example of what I've been thinking... | 14:31 |
efried | ack | 14:32 |
openstackgerrit | Konstantinos Samaras-Tsakiris proposed openstack/nova master: Hide hypervisor id on windows guests https://review.openstack.org/579897 | 14:32 |
efried | jaypipes: Okay, so it would actually need to be | 14:34 |
efried | except (the five-ish ksa exceptions that @safe_connect currently handles): | 14:34 |
efried | ...which we could define in a const tuple somewhere. | 14:34 |
openstackgerrit | Eric Fried proposed openstack/nova master: WIP/PoC: safe_connect shouldn't hide failures https://review.openstack.org/584593 | 14:34 |
efried | jaypipes: FYI ^ attempts to take a big hammer to the test failures... | 14:35 |
gibi | efried: sorry, I'm on a spotty connection. if start() stop() works for you thats cool. i have an enhancement for it somewhere on gerrit... | 14:35 |
efried | gibi: TestCase has restart_compute_service, which ought to suffice. The big part of the work will be refactoring, because currently that shared private method incorporates running periodics explicitly. | 14:37 |
gibi | efried: Here it is, but it does more than what you need https://review.openstack.org/#/c/512552/ | 14:38 |
gibi | efried: if you feel this work is more than the valie it generates then I can accept that and you can ignore my comment | 14:38 |
gibi | s/valie/value/ | 14:39 |
efried | oo, your patch may solve a problem I didn't think about having, which is whether the instances will continue to survive after the compute restart. | 14:39 |
efried | If they don't, that might be okay, as long as their allocations don't get cleaned up... | 14:39 |
efried | gibi: It's a good idea; but yeah, it would be a nontrivial amount of work. Not sure if it's worth it or not. | 14:40 |
gibi | kvm instances survive nova-compute restart | 14:40 |
gibi | efried: then a todo is enough. we can do it later. | 14:40 |
efried | ight | 14:40 |
gibi | a libvirt instance definition (not the running domain) survives the whole compute host restart | 14:41 |
efried | gibi: Then what's with this comment? https://review.openstack.org/#/c/512552/10/nova/test.py@474 | 14:42 |
jaypipes | efried: all the ksa exceptions inherit from heystoneauth1.exceptions.ClientException, though, right? | 14:43 |
efried | jaypipes: Oh, do they? That would be convenient, looking... | 14:43 |
jaypipes | efried: I suppose it would be very useful to have a decorator that just did retries versus safe_connect which does retries as well as hide any exceptions that might occur.. | 14:43 |
efried | jaypipes: safe_connect doesn't retry | 14:44 |
efried | you're saying it would be useful to have a decorator that does retries | 14:44 |
efried | like the `retries` decorator in report.py? | 14:45 |
* efried confused again | 14:45 | |
gibi | efried: that comment is about the case when the definition of a server survives a compute host restart | 14:45 |
jaypipes | efried: yeah, sorry, you're right. | 14:45 |
efried | though @retries sucks in that it returns False when it runs out of retries. | 14:45 |
efried | jaypipes: Anyway, we shouldn't try to combine safe_connect and retries. Their relative usefulnesses overlap, but don't coincide. | 14:46 |
efried | jaypipes: Confirmed that ClientException is the common base, cool. | 14:47 |
*** efried is now known as efried_afk | 14:55 | |
openstackgerrit | Merged openstack/nova master: tox: Ensure reused envdirs share the same deps https://review.openstack.org/588207 | 14:59 |
openstackgerrit | Merged openstack/nova master: Fix a typo in comment in resource_provider.py https://review.openstack.org/588145 | 15:18 |
openstackgerrit | Stephen Finucane proposed openstack/nova master: libvirt: guest: introduce blockStats instead of domain.blockStats https://review.openstack.org/526833 | 15:35 |
*** tssurya has quit IRC | 16:16 | |
*** s10 has quit IRC | 16:33 | |
openstackgerrit | Balazs Gibizer proposed openstack/nova master: Refactor NeutronFixture https://review.openstack.org/588338 | 16:54 |
*** efried_afk is now known as efried | 17:14 | |
openstackgerrit | Matt Riedemann proposed openstack/nova master: Remove old check_attach version check in API https://review.openstack.org/588348 | 17:36 |
openstackgerrit | Eric Fried proposed openstack/nova master: [placement] Debug log per granular request group https://review.openstack.org/588350 | 17:38 |
cdent | efried: I wish you hadn't done that | 17:40 |
efried | warum nicht? | 17:40 |
efried | This one is cheap and easy, dude. | 17:40 |
cdent | because you've violated the stack | 17:41 |
efried | eh? | 17:41 |
cdent | you're stack oriented, not queue oriented | 17:41 |
cdent | and that's bad for getting stuff done, well | 17:41 |
cdent | there's no doubt that this is a a nice add | 17:41 |
* efried waits for 'but' | 17:42 | |
cdent | but it's enough to say "we did something" but not really enough to say "we analysed the problem and came up with a good solution" | 17:42 |
cdent | also | 17:42 |
cdent | since nested doesn't work yet from anywhere but placement itself, request groups are rare | 17:42 |
cdent | so for the common case, there is no new information here | 17:42 |
efried | Completely true. | 17:42 |
efried | wishing I hadn't done it might be a bit strong, though. | 17:43 |
cdent | so if you have the strenght of mind to remember that this is halfway and we'll come back to it for th ereal thing | 17:43 |
cdent | then that's cool | 17:43 |
cdent | but given our history, I'm not sure that as a group we have that strenght of mind | 17:43 |
efried | Oh, sheee-yit, of course I understand this isn't the whole solution. | 17:43 |
efried | maybe I should have made that clearer in my email. | 17:44 |
efried | I'll follow up. | 17:44 |
cdent | yes, "wish you hadn't done that" is a bit strong. I don't want it to go away. | 17:44 |
cdent | oh, I didn't know about any email, I just saw the link above | 17:44 |
openstackgerrit | Merged openstack/nova master: Remove unused stubbing function from test https://review.openstack.org/586968 | 17:44 |
efried | not much to the email | 17:44 |
openstackgerrit | Merged openstack/nova master: Improve NeutronFixture and remove unncessary stubbing https://review.openstack.org/587412 | 17:45 |
cdent | a followup would be awesome to explain why it's not (yet) very useful | 17:45 |
cdent | my "wish" is more related to "ugh, I wish we could linearize more", c.f. our conversation about "getting big arcs done" | 17:46 |
efried | cdent: fup'd | 17:47 |
cdent | thanks, perfect | 17:48 |
efried | cdent: Yes, linearize I get, but this was too easy to pass up. It took me like five minutes, 4 1/2 of which was getting the output into paste.o.o and composing the email. | 17:48 |
* cdent nods | 17:49 | |
cdent | yeah, sorry for being assy about it. I'm frustrated by the number of dangling questions and issues in my message on that thread. They've been accumulating for so long and I'm worn by choosing or needing to accept it. I don't want to anymore. | 17:50 |
efried | Agree that the bigger questions need to be answered, and will probably lead to big feature work one way or another. But IMO anything we can do cheap and easy like this, we should do, because it's clearly useful and a near-zero drain on perf and on dev time. | 17:52 |
cdent | I don't dispute that. | 17:55 |
openstackgerrit | Eric Fried proposed openstack/nova master: WIP/PoC: safe_connect shouldn't hide failures https://review.openstack.org/584593 | 17:57 |
cdent | I reckon about some rather large %age of my dev time is blown on depression from languishing bigger questions | 17:59 |
cdent | maybe I should switch to stack-based time mgt | 17:59 |
cdent | thanks for the fup fup efried | 18:23 |
efried | fo sho | 18:24 |
openstackgerrit | Eric Fried proposed openstack/nova master: [placement] Add /reshaper handler for POST https://review.openstack.org/576927 | 18:53 |
openstackgerrit | Eric Fried proposed openstack/nova master: reshaper: Look up provider if not in inventories https://review.openstack.org/585033 | 18:53 |
openstackgerrit | Eric Fried proposed openstack/nova master: Make get_allocations_for_resource_provider sane https://review.openstack.org/584598 | 18:53 |
openstackgerrit | Eric Fried proposed openstack/nova master: Report client: Real get_allocs_for_consumer https://review.openstack.org/584599 | 18:53 |
openstackgerrit | Eric Fried proposed openstack/nova master: Report client: get_allocations_for_provider_tree https://review.openstack.org/584648 | 18:53 |
openstackgerrit | Eric Fried proposed openstack/nova master: Report client: _reshape helper, placement min bump https://review.openstack.org/585034 | 18:53 |
openstackgerrit | Eric Fried proposed openstack/nova master: Report client: update_from_provider_tree w/reshape https://review.openstack.org/585049 | 18:53 |
openstackgerrit | Eric Fried proposed openstack/nova master: Compute: Handle reshaped provider trees https://review.openstack.org/576236 | 18:53 |
openstackgerrit | Eric Fried proposed openstack/nova master: WIP: only _init_compute_node on startup https://review.openstack.org/588094 | 18:53 |
efried | rebase only | 18:54 |
openstackgerrit | Eric Fried proposed openstack/nova master: WIP: Remove redundant _update()s https://review.openstack.org/588091 | 18:54 |
efried | I get *why* zuul refuses to run when the patch is in merge conflict, but it's inconvenient when I want to use zuul to identify test failures. | 18:55 |
efried | ...without having to rebase the world first | 18:55 |
edleafe | that's what local tests are for | 18:56 |
efried | edleafe: Sure, except toxing the world takes 20 minutes and raises the ambient temperature by two degrees. | 18:59 |
cdent | efried: I forget how it works, but is it the case that you've just eaten 10 zuul nodes? | 19:02 |
efried | could be, could be. | 19:02 |
efried | I would have only eaten one if zuul would just run on the merge-conflicting patch. | 19:02 |
openstackgerrit | Merged openstack/nova master: Wait for vif plugging during live migration job https://review.openstack.org/578551 | 19:03 |
efried | it's also annoying that submitting a series changes all the topics to whatever branch you were running on. | 19:03 |
openstackgerrit | Merged openstack/nova master: Stop setting glance_api_version in cinder.conf in nova-live-migration https://review.openstack.org/579871 | 19:03 |
efried | since I'm griping about infra that's probably not worth changing. | 19:03 |
openstackgerrit | Merged openstack/nova master: doc: fix resize user guide link https://review.openstack.org/588097 | 19:03 |
cdent | efried: do you know about -t -T on git review? If you're aware in advance, that might be some kind of help | 19:06 |
openstackgerrit | Merged openstack/nova master: Hyper-V + OVS: plug vifs before starting VMs https://review.openstack.org/585661 | 19:06 |
efried | oo, no, looking... | 19:06 |
openstackgerrit | Merged openstack/nova master: Complete the api-ref of security group rule https://review.openstack.org/580109 | 19:06 |
efried | cdent: I don't see -T | 19:07 |
efried | cdent: But -t won't help me, cause my gripe is that 7 out of those 9 patches *should* be on the same topic - the one corresponding to the branch I was on - but the other two's topics should be left tf alone. | 19:07 |
cdent | Oh, I think gerrit's thinking on that is that if you are in a stack, they must be the same topic, otherwise why are they in the same stack? | 19:08 |
cdent | -T is in 1.26, dunno what it actually does based on the man page | 19:08 |
efried | one is a bug fix that enables something else in the stack; the other is an unrelated thing that has to go on top of that stack because it uses a thing that was introduced in that stack. | 19:09 |
efried | What I don't get is why "it" (not sure if "it" is git-review or gerrit) feels the need to change the topic of a patch that already has one. | 19:09 |
cdent | dunno | 19:09 |
cdent | I gotta run, I hope you figure out something useful | 19:10 |
cdent | bbl | 19:10 |
efried | cdent: I was at 1.23. Upgrading... | 19:11 |
efried | -T might do it - or it might clear the existing topic... | 19:11 |
openstackgerrit | Eric Fried proposed openstack/nova master: Grease test_try_deallocate_network_retry_direct https://review.openstack.org/588364 | 19:31 |
*** cdent has quit IRC | 19:35 | |
*** avolkov has quit IRC | 19:35 | |
openstackgerrit | Merged openstack/nova master: Fix all invalid obj_make_compatible test case https://review.openstack.org/574240 | 19:55 |
openstackgerrit | Matt Riedemann proposed openstack/nova master: Scrub hw:cpu_model from API samples https://review.openstack.org/588371 | 20:02 |
openstackgerrit | Eric Fried proposed openstack/nova master: WIP/PoC: safe_connect shouldn't hide failures https://review.openstack.org/584593 | 20:07 |
*** takashin has joined #openstack-placement | 20:48 | |
openstackgerrit | Eric Fried proposed openstack/nova master: Grease some more tests hitting RetryDecorator https://review.openstack.org/588391 | 20:53 |
openstackgerrit | Eric Fried proposed openstack/nova master: Grease some more tests hitting RetryDecorator https://review.openstack.org/588391 | 20:55 |
*** mriedem is now known as mriedem_afk | 21:28 | |
openstackgerrit | Takashi NATSUME proposed openstack/nova master: Reno for notification-transformation-rocky https://review.openstack.org/588403 | 22:20 |
openstackgerrit | Eric Fried proposed openstack/nova master: [placement] Add /reshaper handler for POST https://review.openstack.org/576927 | 22:45 |
openstackgerrit | Eric Fried proposed openstack/nova master: reshaper: Look up provider if not in inventories https://review.openstack.org/585033 | 22:45 |
openstackgerrit | Eric Fried proposed openstack/nova master: Make get_allocations_for_resource_provider sane https://review.openstack.org/584598 | 22:45 |
openstackgerrit | Eric Fried proposed openstack/nova master: Report client: Real get_allocs_for_consumer https://review.openstack.org/584599 | 22:45 |
openstackgerrit | Eric Fried proposed openstack/nova master: Report client: get_allocations_for_provider_tree https://review.openstack.org/584648 | 22:45 |
openstackgerrit | Eric Fried proposed openstack/nova master: Report client: _reshape helper, placement min bump https://review.openstack.org/585034 | 22:45 |
openstackgerrit | Eric Fried proposed openstack/nova master: Report client: update_from_provider_tree w/reshape https://review.openstack.org/585049 | 22:45 |
openstackgerrit | Eric Fried proposed openstack/nova master: Compute: Handle reshaped provider trees https://review.openstack.org/576236 | 22:45 |
openstackgerrit | Eric Fried proposed openstack/nova master: WIP: Remove redundant _update()s https://review.openstack.org/588091 | 22:45 |
efried | cdent: git review -T did the trick ^ -- thanks for that! | 22:46 |
*** takashin has left #openstack-placement | 23:02 | |
*** edmondsw has quit IRC | 23:08 | |
*** jaypipes has quit IRC | 23:14 | |
*** jaypipes has joined #openstack-placement | 23:16 |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!