| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-nova - 2017-03-27 | |||
| 10:23:01 | bauzas | voila | |
| 10:23:16 | bauzas | ygl: you only have one compute, right? | |
| 10:24:25 | ygl | bauzas: yes | |
| 10:24:49 | moshele | sfinucan: hi | |
| 10:25:02 | sfinucan | moshele: howdy | |
| 10:25:06 | openstackgerrit | Stephen Finucane proposed openstack/nova master: tox: Stop calling config/policy generators twice https://review.openstack.org/450153 | |
| 10:25:18 | bauzas | ygl: that just tells you something went bad in the compute node for the first time and when it retried the scheduler, it said "sorry, but since you tried this compute which was wrong, I don't have more computes to give you" | |
| 10:25:49 | bauzas | ygl: so the problem is tied within the compute node log | |
| 10:25:51 | moshele | sfinucan: can you please review my HW offload support for openvswitch patches, https://review.openstack.org/#/c/398277/ and https://review.openstack.org/#/c/398265/ | |
| 10:26:51 | ygl | bauzas: i see this error in compute | |
| 10:26:53 | ygl | bauzas: NotFound: Basic.publish: (404) NOT_FOUND - no exchange 'scheduler_fanout' in vhost '/' | |
| 10:27:19 | bauzas | ygl: is the compute running ? | |
| 10:27:49 | bauzas | ygl: because the above is a rabbit error telling you that the compute can't fanout to the scheduler MQ | |
| 10:28:22 | bauzas | ygl: the compute does that for telling to the scheduler(s) how many aggregates and instances it knows | |
| 10:28:48 | ygl | bauzas: ok | |
| 10:28:58 | ygl | bauzas: i have restarted the rabbitmq service in controller | |
| 10:29:05 | ygl | bauzas: launching again to check | |
| 10:29:25 | bauzas | ygl: you should rather look at why the compute rescheduled the first time the instance | |
| 10:30:03 | ygl | bauzas: i didn't get u | |
| 10:30:30 | bauzas | ygl: so, the boot went that way : | |
| 10:30:41 | bauzas | api > conductor > scheduler > conductor > compute | |
| 10:30:51 | bauzas | then the compute returned an exception | |
| 10:31:00 | bauzas | so the compute rescheduled the instance, by saying | |
| 10:31:11 | bauzas | compute > conductor > scheduler > conductor | |
| 10:31:27 | bauzas | that's where you saw the NoValidHosts error due to the RetryFilter | |
| 10:31:37 | ygl | bauzas: it again went into error state | |
| 10:31:38 | bauzas | because it was rescheduling an instance request | |
| 10:31:50 | ygl | bauzas: how to proceed now ? | |
| 10:31:58 | bauzas | and given you only have one node, it told you that there were no other nodes | |
| 10:32:22 | bauzas | ygl: like I said, you should look at *why* the compute is unable to spawn your instance before rescheduling it | |
| 10:32:24 | ygl | bauzas: shall I restart the compute node and check ? | |
| 10:32:30 | bauzas | stop restarting things :) | |
| 10:32:36 | bauzas | and rather look at logs :) | |
| 10:33:08 | bauzas | somewhere in the nova-compute log, you have an ERROR that triggers a reschedule | |
| 10:33:44 | bauzas | you could look at the request-id again | |
| 10:35:55 | ygl | bauzas: i cant find any error messages in nova-compute logs | |
| 10:36:20 | bauzas | you're grepping ERROR right? | |
| 10:36:25 | ygl | yes | |
| 10:36:32 | bauzas | okay, instead of doing that | |
| 10:36:43 | bauzas | take the request-id | |
| 10:37:00 | bauzas | find that request-id in the conductor, and see when it calls the compute node | |
| 10:37:07 | ygl | bauzas: u mean the instance-id ? | |
| 10:37:15 | bauzas | no, the context req-id | |
| 10:37:26 | bauzas | oh, you don't know what's this ? | |
| 10:37:55 | ygl | bauzas: sorry, i dont know . is it the message id of rabbitmq u mean ? | |
| 10:37:57 | bauzas | it's the request UUID that's generated for every request coming from the keystone middleware | |
| 10:38:20 | bauzas | it's useful for tracking RPCs | |
| 10:38:31 | ygl | bauzas:where can i capture that ? | |
| 10:38:37 | bauzas | so, in the log, you have something named "req-XXXXX" | |
| 10:38:49 | bauzas | *that* is the request-id | |
| 10:38:54 | ygl | bauzas: inthe nova-compute logs u mean ? | |
| 10:39:01 | bauzas | in every log :) | |
| 10:39:03 | openstackgerrit | Ghanshyam Mann proposed openstack/nova master: Fix api-ref for adminPass behavior https://review.openstack.org/449450 | |
| 10:39:18 | gmann | alex_xu: thanks, can you check my reply | |
| 10:40:09 | gmann | alex_xu: i think config option still useful. no ? | |
| 10:43:05 | gmann | alex_xu: even exposing that based on config is not good. Not sure what is actual need of that or we could have remove that in 2.14 like in evacuate. | |
| 10:44:17 | bauzas | ygl: https://developer.openstack.org/api-guide/compute/faults.html | |
| 10:44:27 | openstackgerrit | Ghanshyam Mann proposed openstack/nova master: Fix api-ref for adminPass behavior https://review.openstack.org/449450 | |
| 10:44:49 | openstackgerrit | John Garbutt proposed openstack/nova master: Add description to policies in console_output.py https://review.openstack.org/448684 | |
| 10:46:14 | ygl | bauzas: i am clueless now | |
| 10:46:17 | openstackgerrit | Stephen Finucane proposed openstack/nova master: libvirt: Remove dead code https://review.openstack.org/449708 | |
| 10:49:13 | openstackgerrit | John Garbutt proposed openstack/nova master: Add description to policies in certificates.py https://review.openstack.org/448267 | |
| 11:04:01 | openstackgerrit | John Garbutt proposed openstack/nova master: Add description to policies in admin_actions https://review.openstack.org/440829 | |
| 11:04:16 | openstackgerrit | John Garbutt proposed openstack/nova master: Remove unused policy rule in admin_actions.py https://review.openstack.org/447611 | |
| 11:04:46 | sean-k-mooney | moshele: hi | |
| 11:04:54 | ygl | bauzas: hi | |
| 11:06:40 | moshele | sean-k-mooney: hi, can you review HW offload support for openvswitch https://review.openstack.org/#/c/398277/ and https://review.openstack.org/#/c/398265/ | |
| 11:07:03 | sean-k-mooney | moshele: yes ill take a look. | |
| 11:07:21 | sean-k-mooney | https://review.openstack.org/#/c/447269/ is failing tempest test | |
| 11:07:26 | sean-k-mooney | im not sure why | |
| 11:07:41 | openstackgerrit | John Garbutt proposed openstack/nova master: Add description to policies in keypairs.py https://review.openstack.org/448461 | |
| 11:07:46 | moshele | sean-k-mooney: I will take a look | |
| 11:07:48 | sfinucan | moshele: Done | |
| 11:08:14 | sean-k-mooney | moshele: has the ovs support merged yet? | |
| 11:08:30 | moshele | sean-k-mooney: the ovs patches https://mail.openvswitch.org/pipermail/ovs-dev/2017-March/330151.html are close to be merge | |
| 11:08:32 | sean-k-mooney | its targeted to the 2.8 release correct | |
| 11:08:44 | moshele | sean-k-mooney: yes | |
| 11:08:44 | sean-k-mooney | ok cool. | |
| 11:08:58 | moshele | sfinucan: thanks : ) | |
| 11:09:14 | openstackgerrit | Balazs Gibizer proposed openstack/nova master: Short circuit notifiations when not enabled https://review.openstack.org/428260 | |
| 11:09:45 | sfinucan | moshele: np. It's a very high level review because I'm not up to speed on the context. Would appreciate a better overview in the commit message | |
| 11:09:55 | ygl | bauzas: how to proceed now | |
| 11:10:35 | sfinucan | moshele: In fact, a 'Supported VIF types' document in os-vif wouldn't be any harm. Perhaps add one and populate it with details of this ovs_accelerated vif | |
| 11:10:55 | sfinucan | i.e. what it is, why you'd want to use it (vs. other vif types) and perhaps what the minimum requirements are | |
| 11:11:11 | openstackgerrit | John Garbutt proposed openstack/nova master: Add description to policies in simple_tenant_usage.py https://review.openstack.org/449742 | |
| 11:18:30 | openstackgerrit | John Garbutt proposed openstack/nova master: Add description to policies in pause_server.py https://review.openstack.org/449478 | |
| 11:25:39 | ygl | hi all | |
| 11:25:54 | ygl | my nova-scheduler filter is failing and not launching the instance | |
| 11:26:05 | ygl | it is returning 0 hosts | |
| 11:28:44 | ygl | it is ocata setup | |
| 11:34:50 | sean-k-mooney | moshele: once you add unit tests to the nova review it looks good to me but i left 1 question regarding the designer but its minior though | |
| 11:35:10 | moshele | sean-k-mooney: thank :) | |
| 11:35:20 | sean-k-mooney | moshele: for this to work you will also need a bump of os-vif right | |
| 11:35:30 | sean-k-mooney | ie we will need a new release | |
| 11:35:58 | moshele | sean-k-mooney: ok | |
| 11:37:31 | johnthetubaguy | sfinucan: self.stub_out doesn't use mox right? there seem to be lots of folks trying to remove the use of that. | |
| 11:42:10 | bauzas | ygl: again, like I said, please look at n-cpu.logs | |
| 11:42:24 | bauzas | ygl: find the request-id by looking at the n-api.logs | |
| 11:42:32 | ygl | bauzas: ok | |
| 12:00:35 | BobBall_ | johnthetubaguy: Can you comment on https://review.openstack.org/#/c/444230/4/nova/virt/xenapi/vif.py@275 pls? I think that Rackspace may depend on this magic value but I'm not certain | |