Tuesday, 2019-09-10

*** goldyfruit_ has joined #openstack-containers00:20
*** sapd1_x has quit IRC00:35
*** hongbin has joined #openstack-containers01:35
*** spsurya has joined #openstack-containers02:18
*** ykarel|away has joined #openstack-containers02:25
*** dave-mccowan has quit IRC03:35
*** hongbin has quit IRC03:46
*** ykarel|away has quit IRC03:53
*** ramishra has joined #openstack-containers03:54
*** udesale has joined #openstack-containers04:07
*** ykarel|away has joined #openstack-containers04:08
*** ykarel|away is now known as ykarel04:09
*** ricolin has joined #openstack-containers05:00
*** sapd1_x has joined #openstack-containers06:25
*** rcernin has quit IRC07:09
*** threestrands has quit IRC07:20
*** ykarel is now known as ykarel|lunch07:39
*** trident has quit IRC07:50
*** lpetrut has joined #openstack-containers07:55
*** trident has joined #openstack-containers08:01
*** sapd1_x has quit IRC08:18
*** flwang1 has joined #openstack-containers08:57
flwang1strigazi: ping08:57
*** ykarel|lunch is now known as ykarel|09:01
*** ykarel| is now known as ykarel09:01
*** rcernin has joined #openstack-containers09:39
strigaziflwang1: hello09:52
flwang1did you see my email? are we going to have the meeting on Wed 9:00AM UTC or Tue?09:53
strigaziboth work for me09:53
*** udesale has quit IRC10:09
*** udesale has joined #openstack-containers10:10
brtknrstrigazi: any idea how to diagnose the calico-node CrashLoopBackoff issue?10:54
brtknrIve realised it is not to do with your change... I am getting the same problem on master branch10:54
brtknrDespite the cluster reporting its health as HEALTHY10:54
*** udesale has quit IRC11:14
strigazibrtknr: http://paste.openstack.org/raw/774676/ works for me11:15
strigazimaster plus my pacth11:15
brtknrstrigazi: on devstack?11:16
strigaziy11:16
brtknrwhat base OS?11:16
brtknrubuntu or centos?11:16
brtknrcan you share your local.conf?11:17
*** dave-mccowan has joined #openstack-containers11:17
strigazifor dsvm?11:18
strigazifor the OS of devstack ubuntu bionic11:18
strigazibrtknr: my local.conf http://paste.openstack.org/raw/774677/11:19
*** mgoerens has joined #openstack-containers11:22
brtknrstrigazi: is your devstack host baremetal or vm?11:37
*** ykarel is now known as ykarel|afk11:50
strigazibrtknr: baremetal11:58
*** goldyfruit_ has quit IRC12:18
strigazibrtknr: flwang1 2019-09-10 12:20:49.163 [FATAL][632] int_dataplane.go 824: Kernel's RPF check is set to 'loose'.  This would allow endpoints to spoof their IP address.  Calico requires net.ipv4.conf.all.rp_filter to be set to 0 or 1. If you require loose RPF and you are not concerned about spoofing, this check can be disabled by setting the IgnoreLooseRPF configuration parameter to 'true'.12:27
strigazibrtknr: flwang1 fix: http://paste.openstack.org/raw/774682/12:28
brtknrstrigazi: is that from the log I shared?12:31
strigazino12:39
strigazilog calico-<hash> calico-node -f12:39
openstackgerritSpyros Trigazis proposed openstack/magnum master: k8s_fedora: Set rp_filter=1 for calico  https://review.opendev.org/68124412:45
strigazibrtknr: with patch above http://paste.openstack.org/raw/774683/12:46
strigaziflwang1: ^^12:46
brtknrstrigazi: nice, I will lookout for [fatal] log messages in the future12:52
openstackgerritSpyros Trigazis proposed openstack/magnum master: k8s_fedora: Set rp_filter=1 for calico  https://review.opendev.org/68124412:55
strigaziready to test, conformance passes. Please try it too. (min two worker nodes)12:55
*** udesale has joined #openstack-containers12:57
brtknrah, it worked for me after setting that on the hose13:08
brtknrah, it worked for me after setting that on the host13:08
brtknrstrange13:08
brtknrhow do you do conformance test? sonobuoy?13:09
brtknrstrigazi: ^13:09
*** ykarel|afk is now known as ykarel13:10
brtknrstrigazi: thanks13:18
*** goldyfruit_ has joined #openstack-containers13:23
*** rcernin has quit IRC13:45
*** ramishra has quit IRC13:51
*** ramishra has joined #openstack-containers13:51
strigazibrtknr: yes sonobuoy? just 'sonobuoy run'14:09
strigazisonobuoy run --e2e-focus "should provide DNS for service" is the simplest test to check if the network works14:10
strigazibrtknr: ^^14:10
*** ianychoi_ is now known as ianychoi14:37
*** ykarel is now known as ykarel|afk14:42
*** udesale has quit IRC15:00
*** udesale has joined #openstack-containers15:00
*** spsurya has quit IRC15:05
squarebracketi have a couple blades both with their own separate openstack install. on one, when i spin up a cluster, i have to go to the instances and type some stuff on my keyboard to get some randomness for generating an ssh key, on the other blade i don't. anyone know why that might be the case?15:39
*** ykarel|afk is now known as ykarel|away15:42
*** ykarel|away has quit IRC15:47
*** ykarel|away has joined #openstack-containers16:06
*** lpetrut has quit IRC16:12
*** colby_ has joined #openstack-containers16:37
colby_Hey everyone. What is the highest kubernetes version I can use on Rocky?16:38
colby_We are running 7.1.016:47
*** ykarel|away has quit IRC16:54
*** ykarel|away has joined #openstack-containers17:03
*** udesale has quit IRC17:12
squarebracketoh, i see. the one where it doesn't have enough randomness is only sandybridge, so it doesn't have the hardware rng :\17:24
squarebracketin that case, does someone here know where in the magnum code the nova VMs are created? i'll see if adding a virtio rng device makes the entropy larger.17:28
*** trident has quit IRC17:46
ykarel|awaysquarebracket, cluster vms are created using heat, defined in heat template for the driver17:55
ykarel|awayex, for k8s fedora-atomic, https://github.com/openstack/magnum/blob/master/magnum/drivers/k8s_fedora_atomic_v1/templates/kubemaster.yaml#L690-L70317:55
*** trident has joined #openstack-containers17:59
*** ykarel|away has quit IRC18:39
*** goldyfruit___ has joined #openstack-containers18:49
*** goldyfruit_ has quit IRC18:52
*** ricolin has quit IRC18:58
*** flwang1 has quit IRC19:01
*** pcaruana has quit IRC19:20
squarebracketyeah thanks, found that. unfortunately, it seems OS::Nova::Server doesn't have a way to add other devices, so I guess I'm SOL on that fround :(19:59
brtknrsquarebracket: can you add metadata to the image/flavor to invoke virtio rng?21:16
brtknrwe did this recently to support FA29 in devstack21:17
squarebracketbrtknr: oooh, good question!21:18
brtknrandrein: take a look at this: https://review.opendev.org/#/c/678294/5/devstack/lib/magnum21:21
brtknrimage property needs to be  --property hw_rng_model=virtio"21:21
brtknrflavor property --property hw_rng:allowed=True --property hw_rng:rate_bytes=1024 --property21:21
brtknrhw_rng:rate_period=121:21
brtknrsquarebracket: someone asked the same question the other day, this was my answer21:21
*** henriqueof1 has joined #openstack-containers21:30
*** henriqueof has quit IRC21:31
*** flwang has joined #openstack-containers21:49
squarebracketbrtknr: nice, that worked! thanks. i guess those are properties that openstack understands? or are those passed directly to qemu?22:17
*** rcernin has joined #openstack-containers22:59
*** henriqueof1 has quit IRC23:02
*** goldyfruit___ has quit IRC23:17

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