15:00:30 <haleyb> #startmeeting neutron_dvr
15:00:31 <openstack> Meeting started Wed Feb  3 15:00:30 2016 UTC and is due to finish in 60 minutes.  The chair is haleyb. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:00:32 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
15:00:34 <haleyb> #chair Swami
15:00:35 <openstack> The meeting name has been set to 'neutron_dvr'
15:00:36 <openstack> Current chairs: Swami haleyb
15:00:57 <carl_baldwin> o/
15:01:26 <haleyb> let's get started, hopefully we didn't step on each other's wiki edits Swami
15:01:40 <Swami> haleyb: just made a small change
15:01:45 <haleyb> #topic Announcements
15:02:06 <haleyb> Neutron midcycle is coming up
15:02:15 <haleyb> https://etherpad.openstack.org/p/neutron-mitaka-midcycle
15:03:06 <haleyb> obondarev, Swami - were you two still planning on attending?  I know we had added a DVR topic for discussion
15:03:14 <obondarev> o/
15:03:23 <Swami> haleyb: I need to get approval, I am waiting.
15:03:25 <obondarev> sorry for being late
15:03:38 <obondarev> I'm waiting for approval too
15:04:40 <haleyb> ok, we can talk later then
15:04:55 <haleyb> i do have approval, but it's *cold* up there :)
15:05:29 <Swami> haleyb: in the worst case we can still work remotely
15:05:58 <haleyb> Swami: right, we can have a remote midcycle for DVR
15:06:35 <haleyb> #topic Live Migration
15:06:42 <Swami> haleyb: the good news is that the live migration patch seems to fix the floatingip ssh issue
15:06:53 <haleyb> the tempest failure?
15:07:09 <Swami> haleyb: not the tempest failure, I have not tested all test cases.
15:07:32 <haleyb> Swami: which issue, since there are a few seemingly
15:07:34 <Swami> haleyb: But it seems to fix the critical bug that we have. - Block migration with floatingip ssh fails.
15:08:15 <obondarev> Swami: oh, so it helps?
15:08:22 <Swami> I need to do more testing on all use cases.
15:08:27 <Swami> obondarev: yes.
15:08:32 <obondarev> Swami: great!
15:09:00 <Swami> obondarev: haleyb: still the patches are WIP. But it would be great if I can get some early feedback.
15:09:07 <obondarev> Swami: so what is the channel of nova telling neutron to configure fip namespace on dest host?
15:09:28 <haleyb> Swami: so this passed in jenkins?  or your testing?
15:09:49 <Swami> obondarev: Nova does not mandate that. But all nova mentioned was to call the "setup_networks_on_host"
15:10:13 <obondarev> Swami: ok, I need to look at the patches
15:10:18 <Swami> Since we internally use the 'host' on floatingip, we can now use 'dest_host' along with the existing host.
15:10:44 <Swami> obondarev: This patch is based on your input, were to add the new destination as part of the 'portbinding:profile'
15:10:59 <obondarev> Swami: ah, good to hear it works
15:11:01 <Swami> obondarev: I am not using the portbinding table extensions for this patch.
15:11:26 <Swami> obondarev: I have also discussed with kevinbenton and this and he is also ok with our approach.
15:11:54 <Swami> This approach will reduce the work on the api. If we introduce the extension, then we might have to change the api to add the new destination host.
15:12:21 <haleyb> https://review.openstack.org/#/c/275420/ https://review.openstack.org/#/c/260738/ and https://review.openstack.org/#/c/275073/ - any reviews in addition to those ?
15:12:33 <haleyb> that's nova/server/agent
15:12:38 <Swami> haleyb: at this time, these are the only three patches.
15:13:38 <Swami> haleyb: thanks for your early review comment and refactor on the nova patch. That looked good.
15:13:49 <haleyb> great, good job on getting those started
15:14:26 <haleyb> i'm sure there's some more cleanups there, but it looks good
15:15:34 <Swami> I only had one concern, in the nova for a particular case, it is calling 'setup_networks_on_host" and then immediately it calls "portbinding".
15:16:24 <Swami> So there is not enough time delay between a pre-migration phase and portbinding phase there, but let us evaluate all the test cases and see how it goes.
15:17:37 <haleyb> Swami: would be good to get some nova core's on that to get feedback, as well as kevin
15:17:51 <Swami> haleyb: as that's what I am hopping.
15:18:23 <Swami> haleyb: also I found a note in the nova midcycle etherpad to fix the "migration_instance_start".
15:18:58 <Swami> But when I looked into the code, it seemed that "setup_networks_on_host" is more reasonable since it is ahead in the call.
15:19:47 <Swami> carl_baldwin: did you have any update from the nova midcylce on this particular issue.
15:20:01 <Swami> s/midcylce/midcycle
15:20:22 <carl_baldwin> Swami: I didn't follow it closely enough.  armax might have an update to give.  Sorry.
15:20:35 <Swami> carl_baldwin: thanks
15:21:26 <Swami> that's all for the live migration.
15:21:56 <haleyb> #topic Bugs
15:22:19 <Swami> We had three new bugs that was filed this week.
15:22:37 <Swami> #link https://bugs.launchpad.net/neutron/+bug/1540259
15:22:40 <openstack> Launchpad bug 1540259 in neutron "uselist should be True to DVRPortBinding orm.relationship" [Low,In progress] - Assigned to ZongKai LI (lzklibj)
15:22:52 <Swami> This bug is a straight forward one and also has a patch for it.
15:23:13 <Swami> #link https://review.openstack.org/274550
15:23:29 <Swami> All ready review is in progress for this patch.
15:23:37 <Swami> So nothing more to discuss here.
15:23:41 <Swami> let us move on.
15:23:57 <Swami> #link https://bugs.launchpad.net/neutron/+bug/1541406
15:23:58 <openstack> Launchpad bug 1541406 in neutron "IPv6 prefix delegation does not work with DVR" [Undecided,New]
15:24:17 <Swami> This is a new bug that was filed yesterday.
15:24:22 <haleyb> ritesh found this, i reproduced it yesterday
15:24:28 <Swami> haleyb: yes
15:24:39 <Swami> haleyb: is it confirmed then
15:25:02 <haleyb> yes, l3-agent is using wrong namespace in the PD case
15:25:42 <Swami> haleyb: is it using the snat namespace or the router namespace.
15:25:53 <haleyb> router, should be using snat
15:26:10 <Swami> ok,
15:26:16 <haleyb> or it needs more work too...
15:26:21 <Swami> got it.
15:26:58 <Swami> The next one in the list is
15:27:02 <Swami> #link https://bugs.launchpad.net/neutron/+bug/1538369
15:27:03 <openstack> Launchpad bug 1538369 in neutron "re factor add_router_interface in l3_dvr_db.py" [Wishlist,New] - Assigned to Manjeet Singh Bhatia (manjeet-s-bhatia)
15:27:16 <Swami> This is not a bug, but a refactor request.
15:27:32 <Swami> So nothing more on this bug.
15:28:04 <Swami> The next one is the DVR+HA
15:28:33 <Swami> This patch needs review #link https://review.openstack.org/#/c/143169/
15:28:50 <obondarev> It had some issues with scheduling staff, need to look through it again
15:29:03 <Swami> obondarev: thanks.
15:29:44 <Swami> The next high one in the list is
15:29:47 <Swami> #link https://bugs.launchpad.net/neutron/+bug/1462154
15:29:48 <openstack> Launchpad bug 1462154 in neutron "With DVR Pings to floating IPs replied with fixed-ips" [High,In progress] - Assigned to ZongKai LI (lzklibj)
15:30:06 <Swami> This is also under review for a while and having slow progress.
15:30:22 <carl_baldwin> I need to look again at this.  Last time I looked, I was very confused about how routes were added.
15:31:04 <Swami> carl_baldwin: thanks, yes I agree that this patch is not very clear.
15:31:37 <haleyb> https://review.openstack.org/#/c/246855/ is patch
15:31:59 <Swami> haleyb: thanks
15:32:04 <Swami> The next one is the
15:32:07 <Swami> #link https://bugs.launchpad.net/neutron/+bug/1522824
15:32:08 <openstack> Launchpad bug 1522824 in neutron "DVR multinode job: test_shelve_instance failure due to SSHTimeout" [High,In progress] - Assigned to shihanzhang (shihanzhang)
15:33:24 <haleyb> That bug and a number of others all fall under https://review.openstack.org/#/c/215467
15:33:47 <Swami_> haleyb: yes.
15:33:58 <Swami_> It needs one more core blessings.
15:34:07 <haleyb> needs another recheck now that nodepool is happy
15:35:07 <Swami_> haleyb: ok.
15:35:20 <haleyb> just triggerd  recheck
15:35:43 <Swami_> haleyb: thanks
15:35:47 <Swami_> The next one on the list is
15:35:50 <Swami_> #link https://bugs.launchpad.net/neutron/+bug/1445255
15:35:51 <openstack> Launchpad bug 1445255 in neutron "DVR FloatingIP to unbound allowed_address_pairs does not work" [Low,In progress] - Assigned to Brian Haley (brian-haley)
15:36:08 * haleyb re-assigns :)
15:36:13 <Swami_> This patch needs review #link https://review.openstack.org/#/c/254439/
15:36:44 <haleyb> Swami: ot
15:36:53 <haleyb> it's in merge conflict, i'll take a look
15:37:11 <Swami_> haleyb: yes, I will push it after addressing the issue.
15:37:20 <haleyb> sounds good
15:37:37 <Swami_> That's all I had for bugs this week.
15:38:31 <haleyb> #topic Gate failures
15:38:58 <haleyb> https://bugs.launchpad.net/bugs/1538387
15:38:59 <openstack> Launchpad bug 1538387 in neutron "fdb_chg_ip_tun throwing exception because fdb_entries not in correct format" [High,In progress] - Assigned to Kevin Benton (kevinbenton)
15:39:22 <haleyb> https://review.openstack.org/272986 is out for review, been some comments
15:39:51 <Swami_> haleyb: will take a look at it.
15:40:35 <haleyb> there was also a follow-on by kevin regarding "marshalling" of fdb entries, carl_baldwin had a good comment so it needs an update
15:40:54 <haleyb> https://review.openstack.org/274605
15:41:24 <haleyb> https://launchpad.net/bugs/1454921 is another causing exceptions
15:41:25 <openstack> Launchpad bug 1454921 in neutron "OVS DVR: KeyError: 'gateway_mac'" [Medium,In progress] - Assigned to Hong Hui Xiao (xiaohhui)
15:41:55 <haleyb> https://review.openstack.org/#/c/272025/ out for review
15:42:19 <Swami_> haleyb: This bug seem to have been created a while back.
15:42:31 <haleyb> obondarev: can you take another look at that review ^^
15:42:33 <Swami_> I remember that I fixed a bug similar to this one a while back.
15:42:38 <Swami_> I will take a look at it again.
15:42:45 <obondarev> haleyb: sure
15:43:12 <haleyb> Swami_: yes, i think we just never noticed, but the exception has been in the logs for a while, agent just kept on running
15:43:34 <Swami_> haleyb: ok, thanks
15:44:50 <haleyb> The single-node DVR job has been tracking the CVR one pretty closely, so nothing there.  Once the migration issues are fixed we can look at the multinode jobs further
15:44:59 <Swami_> haleyb: obondarev: on migrate volume error, do we have more details on what does that test really do.
15:45:45 <Swami_> haleyb: Sometimes the graph kind of does not give you the right picture.
15:45:47 <obondarev> Swami_: I'm afraid I don't
15:46:58 <Swami_> need to understand more on those failing tests.
15:48:54 <haleyb> #topic Performance/Scalability
15:49:05 <haleyb> obondarev: all changes are merged :)
15:49:12 <obondarev> haleyb: right :)
15:49:26 <haleyb> what will you do with all the spare time?
15:49:32 <obondarev> I think I will go with a couple of follow ups
15:49:33 <Swami_> obondarev: nice work
15:49:36 <obondarev> minot however
15:49:42 <obondarev> minor*
15:49:48 <obondarev> Swami_: thanks
15:50:38 <haleyb> #topic Open Discussion
15:51:47 <Swami_> since we have almost reduced most of the single node failures we have to target on the multinode failures. May be the live migration patch can help, but we should see if there are any other test cases that are not aligned with live migration and still fails in upstream.
15:53:16 <carl_baldwin> obondarev: Great work on that one.
15:53:47 <Swami_> I don't have any other topic for today.
15:53:48 <obondarev> carl_baldwin: thanks. And thanks for your reviews folks
15:54:42 <haleyb> Swami_:  yes, agreed, let's just file bugs as we find issues
15:55:03 <Swami_> haleyb: will do
15:55:31 <Swami_> haleyb: obondarev: did you receive the emails from the openstack summit on the dvr talks submission
15:55:55 <obondarev> Swami_: yeah, I did
15:56:12 <haleyb> I did forget to mention earlier that I will miss the next two meetings, school break week and heading on vacation - I'll coordinate with Swami_  to chair alone in my absence
15:56:23 <Swami_> obondarev: thanks
15:56:30 <Swami_> haleyb: no problem.
15:56:39 <obondarev> Swami_: thank you
15:57:07 <haleyb> anything else?
15:58:00 <Swami_> no
15:58:35 <haleyb> well, good work everyone on getting things into much better shape
15:58:38 <haleyb> #endmeeting