17:15:30 #startmeeting ovn_community_development_discussion 17:15:31 Meeting started Thu Sep 10 17:15:30 2020 UTC and is due to finish in 60 minutes. The chair is mmichelson. Information about MeetBot at http://wiki.debian.org/MeetBot. 17:15:32 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 17:15:34 The meeting name has been set to 'ovn_community_development_discussion' 17:15:36 <_lore_> hi all 17:16:57 Hi everyone. NOt sure who all is here today, but I can give a brief update 17:16:57 Hi 17:17:11 Hi 17:17:51 I sent out an email last week announcing a soft freeze for OVN 20.09.0 for 11 September. So far I haven't got any replies for specific patches that people would like in that release. I've spent the week looking at what appear to be the most impactful patches, as well as the ones that are easiest to get merged. 17:18:16 I have a few more patches in my queue, but if anyone has specific ones they want in 20.09.0, please send a response to my email. 17:18:50 mmichelson: I have a couple, I'll reply to your email. 17:19:07 dceara, thanks! 17:19:47 complicating matters, I'm going to be out starting Monday for 2 weeks. So someone will need to volunteer to create the 20.09 branch once we have patches merged after the soft freeze. 17:20:02 I'll be back on 28 September, so I can actually push the release out when I get back. 17:21:15 And that's all from me 17:22:41 Thanks mmichelson. I can have quick update. 17:23:36 I got my i-p flow installation series and nbcfg measurement improvement series merged. Thanks mmichelson and numans for the reivews! 17:24:03 I will do some more reviews this week. 17:24:08 That's my update. 17:24:33 <_lore_> can I go next? very quick update also from me 17:24:45 _lore_: sure 17:24:52 <_lore_> this week I worked mainly downstream backporting upstream fixes 17:24:58 Hello 17:25:08 <_lore_> then I resumed my iPXE work rebasing my old patch 17:25:42 <_lore_> the I started working on ovn-scale-test to run switch_per_node task in a single iteration to reuse the ssh connection 17:25:45 <_lore_> *then 17:26:16 <_lore_> the main issue there is the time estimation done by rally since w the current code it is performed per iteration 17:26:58 <_lore_> dceara suggested we can use a new feature in rally called like merge_atomic_timer that should merge timer w the same name 17:27:26 <_lore_> but it require to upgrade rally (out current version in quite old, like 5y) 17:27:42 <_lore_> so we need to understand explore this 17:27:53 <_lore_> zhouhan: even your version is quite old, right? 17:28:50 _lore_: yes, I think so. 17:29:11 <_lore_> I guess is aligned with our version more or less 17:29:19 <_lore_> have you ever tried to update it? 17:29:38 _lore_: I can remember clearly but just followed the document of ovn-scale-test to install 17:29:48 s/can/can't 17:29:58 <_lore_> from jakub? 17:30:25 _lore_: from ovn-scale-test repo 17:31:03 <_lore_> ok 17:31:10 <_lore_> then I will look into it 17:31:29 _lore_: and I never tried updating once it worked. So please explore and update the document for the steps if there is any change. 17:31:32 <_lore_> last I have a pending ovs-raft patch waiting for reviews 17:31:46 <_lore_> zhouhan: ack, will do 17:32:22 _lore_: thanks! could you remind the ovs-raft patch link if you have it? 17:32:36 <_lore_> I trying to open ovs patchwork 17:32:37 <_lore_> :) 17:32:48 <_lore_> but it seems very slow 17:32:52 _lore_: ok :) 17:32:59 <_lore_> not sure why 17:33:38 <_lore_> zhouhan: "[PATCH v2] ovsdb: raft: add some debugging information to cluster/status command" 17:33:48 <_lore_> I do not have the link at the moment 17:33:59 _lore_, yeah looks like patchworks is slow 17:34:00 _lore_: I see. thanks! 17:34:08 or down I think 17:34:11 <_lore_> that's all from my side 17:34:14 <_lore_> thx 17:34:29 #link https://patchwork.ozlabs.org/project/openvswitch/patch/95b2a271bc6f04b91bed76847b70359b4234eeef.1599127790.git.lorenzo.bianconi@redhat.com/ 17:35:01 patchwork is not responsive for some reason. 17:36:10 I can go real quick next 17:36:56 I spent much of this week on reviews. Thanks to mmichelson and the patchworks, it is much easier to find the ovn patches and to review. 17:37:15 I also addressed comments from mmichelson for the expr cache patch series. 17:37:21 and I applied it today. 17:38:00 I'm also looking into some ovn issues seen in openshift deployments 17:38:22 mmichelson, thanks for the reviews. 17:38:24 That's it from me. 17:39:23 I have a brief update too. 17:40:47 I sent a few patches trying to make Chassis/Chassis_private processing work when the system-id changes. But as pointed by numans and zhouhan there's no bullet proof way to support this without external help from operators. I'll send a patch to properly document this and cleanup the chassis.c code. 17:42:30 I also sent a patch today to add a "default conflict resolution" behavior when we have multiple openflows with the same match and different actions. This is needed to avoid inconsistent behaviors when users configure overlapping ACLs (I know it's a misconfig but sometimes it can't be avoided easily) 17:42:32 #link http://patchwork.ozlabs.org/project/ovn/patch/1599743157-13626-1-git-send-email-dceara@redhat.com/ 17:43:03 dceara, I'm planning to give that a look today 17:44:03 Last thing is the series I sent last week for bypassing conntrack, it might not be worth merging patch 2/2 because it conflicts with numans' fix for load balancers but the first patch can be applied independently and optimizes openflow count for ACLs (as suggested by zhouhan) 17:44:11 #link http://patchwork.ozlabs.org/project/ovn/list/?series=199101 17:44:29 That's it from me this week. Thanks! 17:44:54 mmichelson: thanks! 17:45:24 dceara: thanks. I am looking at the "default conflict resolution" patch, and have some comments. 17:45:48 zhouhan: sure, do you want to discuss here or on the mailing list? 17:46:29 dceara: let me reply in ML 17:46:44 zhouhan: ok, thanks 17:47:26 OK, does anybody else have an update to give? 17:50:08 All right, I guess that's everyone then 17:50:08 dceara: just replied :) 17:50:39 Have a good day/evening/morning everyone. As I mentioned, I'll be out the next two weeks. Bye 17:50:50 bye everyone! 17:50:58 #endmeeting