Tuesday, 2020-07-14

*** hoonetorg has quit IRC00:04
*** yamamoto has joined #openstack-neutron00:10
*** dsneddon has quit IRC00:12
*** dsneddon has joined #openstack-neutron00:14
*** hoonetorg has joined #openstack-neutron00:17
*** hoonetorg has quit IRC00:36
*** also_stingrayza has joined #openstack-neutron01:23
*** stingrayza has quit IRC01:25
*** spatel has joined #openstack-neutron01:55
*** spatel has quit IRC02:34
*** spatel has joined #openstack-neutron02:51
*** psahoo has joined #openstack-neutron02:53
*** dsneddon has quit IRC03:08
*** dsneddon has joined #openstack-neutron03:11
*** dsneddon has quit IRC03:17
*** dsneddon has joined #openstack-neutron03:22
*** dsneddon has quit IRC03:27
*** dsneddon has joined #openstack-neutron03:28
*** dsneddon has quit IRC03:34
*** dsneddon has joined #openstack-neutron03:44
*** sapd1 has quit IRC03:54
*** iurygregory has quit IRC03:54
*** TheJulia has quit IRC03:54
*** sapd1_x has joined #openstack-neutron03:54
*** mbuil has quit IRC03:54
*** mbuil has joined #openstack-neutron03:55
*** TheJulia has joined #openstack-neutron03:55
*** sapd__x has joined #openstack-neutron03:57
*** ratailor has joined #openstack-neutron04:20
*** dsneddon has quit IRC04:22
*** lajoskatona has joined #openstack-neutron04:40
openstackgerritTerry Wilson proposed openstack/neutron master: Clean up some of the OVN agent API methods  https://review.opendev.org/73710604:44
*** dsneddon has joined #openstack-neutron04:52
*** abdysn has joined #openstack-neutron04:54
*** ociuhandu has joined #openstack-neutron05:01
*** ociuhandu has quit IRC05:05
*** spatel has quit IRC05:09
*** waleedm has joined #openstack-neutron05:10
*** gcheresh has joined #openstack-neutron05:42
*** sapd__x has quit IRC05:59
*** dsneddon has quit IRC06:07
*** dsneddon has joined #openstack-neutron06:08
*** elod is now known as elod_off06:15
*** dsneddon has quit IRC06:18
*** hoonetorg has joined #openstack-neutron06:19
*** dsneddon has joined #openstack-neutron06:25
*** halali_ has quit IRC06:30
*** dsneddon has quit IRC06:37
*** boyvinall has joined #openstack-neutron06:46
*** halali_ has joined #openstack-neutron06:51
*** dsneddon has joined #openstack-neutron06:57
*** boyvinall has quit IRC06:59
*** rcernin has quit IRC07:05
*** waleedm_ has joined #openstack-neutron07:10
*** maciejjozefczyk has joined #openstack-neutron07:10
*** iurygregory_ has joined #openstack-neutron07:10
*** bengates has joined #openstack-neutron07:23
openstackgerritLajos Katona proposed openstack/networking-odl master: Fix pep8 job  https://review.opendev.org/73999707:26
*** bengates has quit IRC07:27
*** bengates has joined #openstack-neutron07:28
*** iurygregory_ is now known as iurygregory07:31
*** ralonsoh has joined #openstack-neutron07:31
*** halali_ has quit IRC07:34
*** yamamoto has quit IRC07:38
*** yamamoto has joined #openstack-neutron07:40
openstackgerritLajos Katona proposed openstack/networking-odl stable/ussuri: Fix pep8 job  https://review.opendev.org/74028407:41
*** sapd__x has joined #openstack-neutron07:55
*** dtantsur|afk is now known as dtantsur07:56
openstackgerritMaciej Józefczyk proposed openstack/neutron master: [DNM] Test OVN multinode with release branch  https://review.opendev.org/74049107:56
openstackgerritMaciej Józefczyk proposed openstack/neutron stable/ussuri: [OVN] Wait for WaitForDataPathBindingCreateEvent event in functional tests  https://review.opendev.org/73886507:58
openstackgerritMaciej Józefczyk proposed openstack/neutron stable/ussuri: [OVN] Unify OVN/OVS compilation  https://review.opendev.org/74025507:58
openstackgerritMaciej Józefczyk proposed openstack/neutron stable/ussuri: Stop installing Octavia in OVN jobs  https://review.opendev.org/74027907:58
openstackgerritMaciej Józefczyk proposed openstack/neutron stable/ussuri: [OVN] Stop using neutron_tempest_plugin in OVN singlenode job  https://review.opendev.org/74028107:59
*** Habib-A has joined #openstack-neutron08:01
Habib-AHello all please help :D  https://ask.openstack.org/en/question/128468/windows-server-2019-with-ovs-and-nova-compute/08:02
*** jlibosva has joined #openstack-neutron08:04
*** lucasagomes has joined #openstack-neutron08:08
*** rcernin has joined #openstack-neutron08:10
*** rcernin has quit IRC08:14
openstackgerritLi YaJie proposed openstack/neutron master: Fix delete-accepted-egress-direct-flow issue on deleting VM  https://review.opendev.org/73144608:14
*** rcernin has joined #openstack-neutron08:15
ralonsohbcafarel, hi!08:20
ralonsohdo you remember if the team meeting is canceled?08:20
fricklerralonsoh: according to slaweq's mail no, mlavalle is to take over08:20
ralonsohahhh ok, yes08:21
ralonsohthanks!08:21
openstackgerritLucas Alvares Gomes proposed openstack/neutron master: [OVN] Bump OVN version to v20.06.1  https://review.opendev.org/73580208:38
*** rcernin has quit IRC08:42
*** boyvinall has joined #openstack-neutron08:44
*** ociuhandu has joined #openstack-neutron08:50
*** donnyd has quit IRC08:53
*** donnyd has joined #openstack-neutron08:53
*** boyvinall has quit IRC08:54
*** also_stingrayza is now known as stingrayza08:58
openstackgerritRodolfo Alonso Hernandez proposed openstack/neutron master: String to byte conversion should provide the encoding type  https://review.opendev.org/74069309:01
openstackgerritvinay harsha mitta proposed openstack/neutron-lib master: Neutron agent-list API added with sort_key keyword  https://review.opendev.org/73847109:11
openstackgerritvinay harsha mitta proposed openstack/neutron-lib master: Neutron agent-list API added with sort_key keyword  https://review.opendev.org/73847109:15
*** gcheresh has quit IRC09:23
*** gcheresh has joined #openstack-neutron09:24
*** jangutter_ has joined #openstack-neutron09:28
openstackgerritDarragh O'Reilly proposed openstack/neutron master: Ensure drop flows on br-int at agent startup for DVR too  https://review.opendev.org/74072409:30
*** jangutter has quit IRC09:31
*** frickler is now known as frickler_pto09:44
*** frickler_pto is now known as frickler09:47
*** dtantsur is now known as dtantsur|bbl09:49
*** tkajinam has quit IRC10:12
*** Habib-A has quit IRC10:15
*** yamamoto has quit IRC10:16
*** dsneddon has quit IRC10:19
*** yamamoto has joined #openstack-neutron10:22
*** gouthamr has quit IRC10:22
*** gouthamr has joined #openstack-neutron10:23
*** dsneddon has joined #openstack-neutron10:23
*** Jeffrey4l has quit IRC10:23
*** Jeffrey4l has joined #openstack-neutron10:24
*** yamamoto has quit IRC10:26
*** jangutter has joined #openstack-neutron10:34
*** dsneddon has quit IRC10:35
*** dsneddon has joined #openstack-neutron10:36
*** jangutter_ has quit IRC10:37
*** Kaftantzis has joined #openstack-neutron10:46
openstackgerritMaciej Józefczyk proposed openstack/neutron master: [DNM] Test OVN multinode with release branch  https://review.opendev.org/74049110:51
*** dsneddon has quit IRC10:53
*** dsneddon has joined #openstack-neutron10:56
*** Kaftantzis has quit IRC11:01
*** yamamoto has joined #openstack-neutron11:02
*** dsneddon has quit IRC11:06
*** dsneddon has joined #openstack-neutron11:07
*** yamamoto has quit IRC11:08
*** rcernin has joined #openstack-neutron11:13
*** dsneddon has quit IRC11:13
*** dsneddon has joined #openstack-neutron11:33
*** Kaftantzis has joined #openstack-neutron11:39
*** yamamoto has joined #openstack-neutron11:40
*** dsneddon has quit IRC11:45
*** iurygregory has quit IRC11:46
*** dsneddon has joined #openstack-neutron11:46
*** sapd__x has quit IRC11:50
*** dsneddon has quit IRC11:53
*** dsneddon has joined #openstack-neutron11:54
*** spatel has joined #openstack-neutron11:58
*** iurygregory has joined #openstack-neutron12:00
*** raildo has joined #openstack-neutron12:02
*** spatel has quit IRC12:03
openstackgerritMerged openstack/ovn-octavia-provider master: Fix pep8 and functional jobs  https://review.opendev.org/72691712:06
*** yamamoto has quit IRC12:08
*** dsneddon has quit IRC12:11
*** dsneddon has joined #openstack-neutron12:11
*** yamamoto has joined #openstack-neutron12:13
*** yamamoto has quit IRC12:27
openstackgerritXiaoYu Zhu proposed openstack/neutron-specs master: L3 router support ecmp  https://review.opendev.org/72953212:28
openstackgerritMaciej Józefczyk proposed openstack/ovn-octavia-provider master: Update ROUND_ROBIN to SOURCE_IP_PORT in docs  https://review.opendev.org/74062012:30
openstackgerritJakub Libosvar proposed openstack/neutron master: ovn migration: Support stack name  https://review.opendev.org/74094912:37
*** yamamoto has joined #openstack-neutron12:41
*** dsneddon has quit IRC12:47
*** rcernin has quit IRC12:48
*** baojg has joined #openstack-neutron12:54
openstackgerritFlavio Fernandes proposed openstack/neutron master: [ovn]: port forwarding  https://review.opendev.org/72386312:54
*** Luzi has joined #openstack-neutron12:55
mnaserquestion -- is there any reason why neutron is currently using it's own retry code instead of the built in stuff inside oslo_db for retrying on deadlocks ? https://github.com/openstack/neutron-lib/blob/master/neutron_lib/db/api.py#L109-L12312:57
mnaseri'm seeing some issues that were fixed inside oslo.db -- so i think it would be at least good to rely on oslo_db for deadlocks/retry request/etc for it to handle, and letting neutron retry its own cases..12:58
*** spatel has joined #openstack-neutron12:58
*** dsneddon has joined #openstack-neutron12:58
openstackgerritBence Romsics proposed openstack/neutron master: metadata-ipv6: DHCP namespace  https://review.opendev.org/71548213:01
openstackgerritBence Romsics proposed openstack/neutron master: metadata-ipv6: Router namespace  https://review.opendev.org/71548313:01
openstackgerritBence Romsics proposed openstack/neutron master: metadata-ipv6: Accept link local address in X-Forwarded-For  https://review.opendev.org/71872913:02
ftarasenkolajoskatona: Hi! Can you check https://bugs.launchpad.net/networking-bagpipe/+bug/1883102 cause it fails after patch at another place. Logs are attached to the bug. Tnx13:02
openstackLaunchpad bug 1883102 in BaGPipe "bagpipe: bagpipe-bgp does not start with EVPN and OVS driver " [Undecided,In progress] - Assigned to Lajos Katona (lajos-katona)13:02
*** dsneddon has quit IRC13:08
*** dsneddon has joined #openstack-neutron13:10
*** yamamoto has quit IRC13:10
lajoskatonaftarasenko: sure13:11
lajoskatonaftarasenko: I read again your feedback in launchpad, and try to allocate time for it13:12
ftarasenkolajoskatona: thanks! I have lab with multiple servers available for tests at any time.13:13
*** baojg has quit IRC13:15
*** dsneddon has quit IRC13:16
*** baojg has joined #openstack-neutron13:16
lajoskatonaftarasenko: ok13:18
openstackgerritMerged openstack/ovn-octavia-provider master: Update ROUND_ROBIN to SOURCE_IP_PORT in docs  https://review.opendev.org/74062013:19
openstackgerritFlavio Fernandes proposed openstack/neutron master: [ovn]: port forwarding -- documentation  https://review.opendev.org/74095513:20
*** dsneddon has joined #openstack-neutron13:23
*** dsneddon has quit IRC13:28
*** Kaftantzis has quit IRC13:29
*** Luzi has quit IRC13:29
*** dsneddon has joined #openstack-neutron13:33
openstackgerritLajos Katona proposed openstack/networking-odl stable/ussuri: Fix pep8 job  https://review.opendev.org/74028413:35
*** dsneddon has quit IRC13:38
*** benj_ has quit IRC13:39
*** dsneddon has joined #openstack-neutron13:39
*** benj_ has joined #openstack-neutron13:41
*** Luzi has joined #openstack-neutron13:45
*** dtantsur|bbl is now known as dtantsur13:45
*** sean-k-mooney has joined #openstack-neutron13:46
*** waleedm has quit IRC13:47
*** waleedm_ has quit IRC13:47
fricklermnaser: I tend to agree, likely you should create a bug report that asks for that code to be improved13:48
mnaserfrickler: i did that, and i'm currently working on something right now to refactor to using oslo.db13:48
mnaserhttps://bugs.launchpad.net/neutron/+bug/188752313:48
openstackLaunchpad bug 1887523 in neutron "Deadlock detection code can be stale" [Undecided,New]13:48
openstackgerritMerged openstack/ovn-octavia-provider master: Add a hacking check for importing mock library  https://review.opendev.org/73467013:48
*** ralonsoh has quit IRC13:49
mnaserthe `_tag_retriables_as_unretriable` function is really confusing me though13:49
fricklermnaser: great, thx, added that to my watchlist. sorry I can't help with the details, I've never done much db work, also I'm off now ;)13:50
mnaserfrickler: same -- take care!! :)13:50
* mnaser will slowly figure it out13:50
*** frickler is now known as frickler_pto13:50
mnaserfound this https://github.com/openstack/neutron/commit/09c87425fa028dbe669e8c215e334297ccbf1c2a13:52
openstackgerritMaciej Józefczyk proposed openstack/neutron master: [OVN] Fix logic issue while deleting port with QoS  https://review.opendev.org/74096713:54
*** mlavalle has joined #openstack-neutron13:58
*** dsneddon has quit IRC14:00
*** ralonsoh has joined #openstack-neutron14:00
*** sapd__x has joined #openstack-neutron14:00
*** dave-mccowan has joined #openstack-neutron14:01
*** maciejjozefczyk has quit IRC14:01
*** dsneddon has joined #openstack-neutron14:02
openstackgerritLucas Alvares Gomes proposed openstack/neutron master: Change some tempest tests to explicitly use ML2/OVS  https://review.opendev.org/74096814:02
*** lucasagomes has quit IRC14:13
*** lucasagomes has joined #openstack-neutron14:14
*** armax has joined #openstack-neutron14:16
*** ratailor has quit IRC14:17
*** Luzi has quit IRC14:18
*** knikolla has joined #openstack-neutron14:21
*** yamamoto has joined #openstack-neutron14:28
*** jobewan has joined #openstack-neutron14:29
*** abdysn has quit IRC14:29
openstackgerritMohammed Naser proposed openstack/neutron master: quota: use native oslo.db retry  https://review.opendev.org/74097714:32
lajoskatonamnaser, frickler: Hi, regarding https://bugs.launchpad.net/neutron/+bug/1887523 , I added a question if it can be an RFE, and discuss on next drivers meeting ( or when slaweq will be back to have full team there)14:33
openstackLaunchpad bug 1887523 in neutron "Deadlock detection code can be stale" [Undecided,New]14:33
*** dave-mccowan has quit IRC14:33
mnaserlajoskatona: i just pushed up a patch of what this might look like -- https://review.opendev.org/740977 -- i'm going to try and test it in a pretty busy env14:34
lajoskatonamnaser: ok, but I am sure if you need some help or historycal background the team is ready to have discussion :-)14:37
*** dave-mccowan has joined #openstack-neutron14:39
openstackgerritMerged openstack/neutron master: migration: Use ansible-inventory to parse tripleo inventory  https://review.opendev.org/73821214:40
*** johnthetubaguy has quit IRC14:50
*** johnthetubaguy has joined #openstack-neutron14:52
zigoHi there! I'm trying to make the Ryan Tidwell's patch here: https://review.opendev.org/#/c/669395/ to work, though when creating the subnet with --service-type 'network:routed', Neutron's API reject me with a 400 error saying:14:54
zigo"All of the subnets on network '<NETWORK-ID>' must either all be associated with segments or all not associated with any segment."14:54
zigoIs it possible that, since a year ago, there was some new checks added to Neutron API so that it would reject the API call?14:54
zigoamotoki mlavalle do you remember, over the last year, reviewing such patch that could potentially break #66939514:54
zigo?14:55
*** dsneddon has quit IRC14:56
*** dsneddon has joined #openstack-neutron14:57
openstackgerritFlavio Fernandes proposed openstack/neutron master: [ovn]: port forwarding -- documentation  https://review.opendev.org/74095514:59
*** yamamoto has quit IRC15:03
ralonsohlajoskatona, do you need something? I saw you in the meeting channel15:08
lajoskatonaralonsoh: no I just went there to see what happens around qos :-)15:09
ralonsohlajoskatona, ah ok15:09
lajoskatonaralonsoh: and say that I progress (really) slowly with https://bugs.launchpad.net/neutron/+bug/188280415:10
openstackLaunchpad bug 1882804 in neutron "RFE: allow replacing the QoS policy of bound port" [Wishlist,Confirmed] - Assigned to Lajos Katona (lajos-katona)15:10
ralonsohlajoskatona, do you need some help?15:11
lajoskatonaralonsoh: I think now, I see now that I can cover it, thanks15:12
*** dsneddon has quit IRC15:14
*** dsneddon has joined #openstack-neutron15:15
*** armax has quit IRC15:18
*** hamalq has joined #openstack-neutron15:23
openstackgerritMatt Vinall proposed openstack/neutron stable/ussuri: fix bulk reload exception and race hazard  https://review.opendev.org/74098815:26
openstackgerritMerged openstack/neutron master: Remove tempest-integrated-networking job from neutron queue  https://review.opendev.org/73618615:27
*** hamalq has quit IRC15:29
*** dsneddon has quit IRC15:31
openstackgerritFlavio Fernandes proposed openstack/neutron master: [ovn]: port forwarding -- extend support for OVN usage (cont.)  https://review.opendev.org/74095515:32
*** hamalq has joined #openstack-neutron15:35
*** baojg has quit IRC15:37
openstackgerritFlavio Fernandes proposed openstack/neutron master: [ovn]: port forwarding -- documentation  https://review.opendev.org/74095515:38
*** baojg has joined #openstack-neutron15:38
openstackgerritFlavio Fernandes proposed openstack/neutron master: [ovn]: port forwarding -- extend support for OVN usage (cont.)  https://review.opendev.org/74099315:40
*** clarkb has joined #openstack-neutron15:41
clarkbHi, we've noticed that logstash indexing keeps falling behind lately and digging into our logs for indexing it seems that q-svc logs in neutron jobs have grown quite large and are likely contribnuting to the problem15:41
clarkbhttps://storage.gra.cloud.ovh.net/v1/AUTH_dcaab5e32b234d56b626f72581e3644c/zuul_opendev_logs_a60/733250/19/check/neutron-tempest-dvr-ha-multinode-full/a6025d9/controller/logs/screen-q-svc.txt15:41
clarkbhttps://storage.gra.cloud.ovh.net/v1/AUTH_dcaab5e32b234d56b626f72581e3644c/zuul_opendev_logs_464/735125/10/check/neutron-ovn-rally-task/4648c27/controller/logs/screen-q-svc.txt15:42
clarkbthose are a couple of the examples I've found. Are ya'll able to take a look and see if that is expected or intentional?15:42
clarkbIts likely to be an operator concern eventually if we start filling disks in production too15:42
clarkbwc -l screen-q-svc.txt -> 792240 ; grep Time-cost screen-q-svc.txt| wc -l -> 61402515:48
clarkbseems like instrumentation of function calls has been added to logging output and its exploding the size of that output15:48
clarkbmaybe that should be off by default and only enabled when you're doing performance checking? and even then probably redirected to a separate file so that we don't try and index it all15:49
*** dsneddon has joined #openstack-neutron15:50
*** bengates_ has joined #openstack-neutron15:53
*** spatel has quit IRC15:54
*** dsneddon has quit IRC15:55
*** bengates has quit IRC15:57
*** dsneddon has joined #openstack-neutron15:58
*** lucasagomes has quit IRC15:59
*** dsneddon has quit IRC16:05
*** bengates_ has quit IRC16:13
*** bengates has joined #openstack-neutron16:13
*** dsneddon has joined #openstack-neutron16:17
*** bengates has quit IRC16:17
*** aedc_ has joined #openstack-neutron16:24
*** aedc_ has quit IRC16:24
openstackgerritAndrew Bogott proposed openstack/neutron-tempest-plugin master: Add basic tests for quota access policy  https://review.opendev.org/74099916:27
*** aedc has quit IRC16:27
openstackgerritFlavio Fernandes proposed openstack/neutron master: [ovn]: port forwarding -- IDL changes  https://review.opendev.org/74100016:29
openstackgerritFlavio Fernandes proposed openstack/neutron master: [ovn]: port forwarding -- IDL changes  https://review.opendev.org/74100016:35
*** Reepicheep has joined #openstack-neutron16:44
*** jlibosva has quit IRC16:50
*** Reepicheep has quit IRC16:51
*** Reepicheep has joined #openstack-neutron16:55
*** ociuhandu_ has joined #openstack-neutron16:58
*** njohnston is now known as njohnston|lunch16:58
*** yamamoto has joined #openstack-neutron17:01
*** ociuhandu has quit IRC17:01
*** ociuhandu_ has quit IRC17:02
*** lajoskatona has quit IRC17:27
*** dtantsur is now known as dtantsur|afk17:31
*** psahoo has quit IRC17:38
openstackgerritRodolfo Alonso Hernandez proposed openstack/neutron master: Remove "six" library  https://review.opendev.org/74051017:38
openstackgerritAndrew Bogott proposed openstack/neutron-tempest-plugin master: Add basic tests for quota access policy  https://review.opendev.org/74099917:40
openstackgerritRodolfo Alonso Hernandez proposed openstack/neutron stable/stein: [stable only] Configure logging in keepalived_state_change  https://review.opendev.org/73926717:40
openstackgerritRodolfo Alonso Hernandez proposed openstack/neutron master: Remove "six" library  https://review.opendev.org/74051017:44
*** yamamoto has quit IRC17:53
openstackgerritRodolfo Alonso Hernandez proposed openstack/neutron master: [DNM] Test oslo.privsep patch  https://review.opendev.org/74101717:54
openstackgerritMohammed Naser proposed openstack/neutron master: quota: use native oslo.db retry  https://review.opendev.org/74097717:55
*** ralonsoh has quit IRC17:57
*** dsneddon has quit IRC17:58
openstackgerritLukas Steiner proposed openstack/neutron master: Flush ebtables arp protect chains before deleting them  https://review.opendev.org/74058817:59
clarkbbnemec: the logging I've called out above is from oslo utils, any idea what may be causing that?18:06
bnemecclarkb: I don't think the log messages are coming from oslo. It looks like the source is timecost: https://opendev.org/openstack/neutron-lib/commit/492fd109ed92f15e421ebbdbd4fba5a9129eef1718:11
bnemec(which is using oslo.utils for timing)18:12
clarkbgot it18:12
clarkbnjohnston|lunch: ^ when lunch is over can you maybe reevaluate that?18:12
clarkbI'm not sure that having it on all the time is useful, particularly since it is likely to fill up disks in production18:13
*** sapd__x has quit IRC18:18
openstackgerritMohammed Naser proposed openstack/neutron master: quota: use native oslo.db retry  https://review.opendev.org/74097718:27
*** njohnston|lunch is now known as njohnston18:41
njohnstonclarkb: ack, but that is a debug level log and rarely do prod environments run with debug logs enabled18:43
clarkbnjohnston: I've been told the opposite is true18:44
clarkb(everyone has to run at debug level because its the only way to undersatnd why things break)18:44
njohnstonok.  I'll push a change to disable the logging in the one place I added it, which is here: https://opendev.org/openstack/neutron/src/branch/master/neutron/objects/ports.py#L40918:47
clarkbone approach could be to make it toggelable so that you can enable it when you know you want the profiling data (this is what we do in zuul for example). Another option is to have it log to another location though python logging levels make that a little clunky (you'd have to namespace it maybe then set that namespace to log above debug level by default?)18:49
njohnstonYeah, I was hoping to avoid Yet Another Config Option18:50
njohnstonbut c’est la vie18:51
clarkbin zuul's case it uses signals so not a config option18:53
clarkbsignal turns on the behavior then repeating the signal turns it off18:53
clarkbthat way you can toggle it on a running system in a simple and reliable way18:53
*** slaweq has joined #openstack-neutron18:54
*** slaweq has quit IRC18:56
*** ramishra has quit IRC18:58
*** ociuhandu has joined #openstack-neutron19:14
*** ociuhandu has quit IRC19:23
*** ramishra has joined #openstack-neutron19:25
*** yamamoto has joined #openstack-neutron19:51
*** also_stingrayza has joined #openstack-neutron20:03
*** stingrayza has quit IRC20:07
*** yamamoto has quit IRC20:51
*** gcheresh has quit IRC21:07
*** dsneddon has joined #openstack-neutron21:11
*** markvoelker has joined #openstack-neutron21:23
*** markvoelker has quit IRC21:27
*** yamamoto has joined #openstack-neutron21:35
openstackgerritAndrew Bogott proposed openstack/neutron-tempest-plugin master: Add basic tests for quota access policy  https://review.opendev.org/74099921:35
*** spatel has joined #openstack-neutron21:45
*** spatel has quit IRC22:04
*** spatel has joined #openstack-neutron22:23
*** spatel has quit IRC22:23
*** rcernin has joined #openstack-neutron22:44
*** armax has joined #openstack-neutron22:47
*** rcernin has quit IRC22:51
*** rcernin has joined #openstack-neutron22:51
*** tkajinam has joined #openstack-neutron22:58
*** hamalq has quit IRC22:58
*** mlavalle has quit IRC23:01
*** yamamoto has quit IRC23:02
*** armax has quit IRC23:12
mnaserwould anyone at all know why we have an update here -- https://github.com/openstack/neutron/blob/0efff0534689b2b9256852aeff536f51dc7266e1/neutron/objects/quota.py#L145-L147 ?23:22
mnaserthe comment "as lock mode was just for protecting dirty bits an update on dirty will prevent the race." doesn't explain much23:22
mnaserthere's quite literally no reason for that update lock to be there23:25
mnaserhttps://review.opendev.org/#/c/442181/23:26
mnaserthis is the change23:27
mnaserok, so pretty much, the only use is for counting usage, and this means that for the whole time we do the quota recalculation, we have a hold on the row23:29
mnaseri.e. https://github.com/openstack/neutron/blob/55be92e253a33c9a51ac9db3e444942920372904/neutron/quota/resource.py#L250-L29423:29
*** yamamoto has joined #openstack-neutron23:33
mnaserhttps://github.com/openstack/neutron/blob/d28654a0c740844ab726d557480ebb3fd3ac66c7/neutron/db/quota/driver.py#L225-L23423:33
mnaserand then we do a lot of counting here23:33
mnaserso if we're creating many ports at once, we have many locks on the port/project_id row waiting one after the other23:34
clarkbmnaser: this could be totally wrong, but I think the .first() is rlated to that. Because you may end up getting more results after first() it holds the lock open while you do things in the old code23:42
clarkbwith the added update() it does similar but less explicitly as you might update more rows. However, it appears we only ever expect a single row so maybe you can drop the update and use .one() instead of .first()?23:42
clarkbunless of course that lock needs to be help for thints in count23:42
mnaserclarkb: well, after parsing the code some more, it seems that this update() actually is there to do a select for update to lock the row for resource counting23:43
mnaserbut the problem here is when you're getting 250+ port creations all at once, you start to end up with a bunch of deadlocks, as some threads are waiting for the thread before them to 'finish counting23:43
mnaserfollowing the stack further up, that function is called in here -- https://github.com/openstack/neutron/blob/55be92e253a33c9a51ac9db3e444942920372904/neutron/quota/resource.py#L258-L26023:44
mnaserwhich is also then called here (indirectly via count): https://github.com/openstack/neutron/blob/d28654a0c740844ab726d557480ebb3fd3ac66c7/neutron/db/quota/driver.py#L225-L23423:45

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