08:01:50 <yasufum> #startmeeting tacker 08:01:50 <opendevmeet> Meeting started Tue Sep 20 08:01:50 2022 UTC and is due to finish in 60 minutes. The chair is yasufum. Information about MeetBot at http://wiki.debian.org/MeetBot. 08:01:50 <opendevmeet> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 08:01:50 <opendevmeet> The meeting name has been set to 'tacker' 08:01:55 <masaki-ueno> hi 08:02:20 <takahashi-tsc> hi 08:03:26 <yasufum> #link https://etherpad.opendev.org/p/tacker-meeting 08:04:09 <h_asahina> hi 08:04:20 <yasufum> It's R2 and rc1 was over, so I wonder there are not so much topic today. 08:04:46 <yasufum> Anyway, thanks for your hard work for reviews. 08:05:44 <yasufum> So, let's start from the first item. 08:05:55 <yasufum> #topic Recommended versions of k8s, k8s client, and helm in Tacker Zed Release 08:06:22 <yasufum> yota-kazato: can you share your topic? 08:06:37 <yuta-kazato> yasufum: ok 08:06:57 <yuta-kazato> Before that, I would like to thank everyone for your contributions to rc1 last week! 08:07:29 <yuta-kazato> My topic is Recommended versions of k8s, k8s client, and helm in Tacker Zed Release 08:08:07 <yuta-kazato> In Tacker, we use kubernetes (k8s), k8s client, and Helm for CNF in LCM operations. 08:08:42 <yuta-kazato> However, I think that Tacker community do not recommend which versions to use or support. 08:09:01 <yuta-kazato> There are some problems about their versions. 08:09:41 <yuta-kazato> e.g. Zuul in Zed release uses Helm v3.5.4, which is older than the latest version 3.9.4. 08:10:19 <yuta-kazato> In Devstack, there is the other same problem of k8s versions between k8s client (1.25) and k8s server (1.23). 08:10:30 <yuta-kazato> It should not exceeds the supported minor version skew of +/-1. 08:10:47 <yuta-kazato> As you know, k8s/helm updates and their EoLs are fast cycles.. 08:11:11 <yasufum> sure 08:11:20 <yuta-kazato> Therefore, we need to decide the recommended versions in Zed release and update their versions regularly to provide mutual guarantees. 08:11:42 <yuta-kazato> Does anyone have specific opinions on what has been discussed so far? 08:13:08 <yasufum> I don't have specific one actually... 08:14:14 <masaki-ueno> I think we should determine recommended versions of kubernetes according to the status of k8s python client. 08:16:59 <masaki-ueno> Sorr, I noticed that you mentioned it above.... 08:17:14 <yuta-kazato> Thank you for your opinion! Refering k8s python client's version is good idea. 08:18:01 <yasufum> Where can we I find the dependent version of k8s? 08:18:10 <yasufum> https://pypi.org/project/python-tackerclient/ 08:18:16 <yasufum> here? 08:18:43 <yasufum> sorry, it's wrong 08:18:46 <masaki-ueno> https://github.com/kubernetes-client/python 08:19:16 <masaki-ueno> The above describes the supported version for each version of client libs 08:19:45 <yasufum> ok, thanks 08:21:00 <yuta-kazato> Thanks! 08:22:36 <yuta-kazato> How about providing support version information in Tacker documents. 08:22:41 <yuta-kazato> ? 08:22:57 <yuta-kazato> For example, Kuryr-kubernetes provides their supported versions, too. (But documents are not updated..) 08:23:06 <yuta-kazato> # link 08:23:07 <yuta-kazato> https://docs.openstack.org/kuryr-kubernetes/latest/devref/kuryr_kubernetes_versions.html 08:25:31 <masaki-ueno> +1, and I think we also should update the supported version per one year (= the general EoL cycle of K8s) 08:27:47 <yasufum> yuta-kazato: Do you mean we should provide each supported versions of k8s for several Tacker versions? 08:28:49 <yuta-kazato> yasufum: yes, it is better for users and contributers. 08:30:51 <yuta-kazato> And there is the other problem that k8s v1.25 is already EoL.. 08:30:59 <yuta-kazato> # link 08:30:59 <yuta-kazato> https://kubernetes.io/releases/ 08:31:49 <yuta-kazato> sorry the above information is not correct. 08:32:10 <yasufum> If so, how many old versions we should support? For example, from Train, Ussuri or so. 08:32:27 <yuta-kazato> k8s server "v1.23" is EoL. 08:35:54 <yuta-kazato> yasufum: IMO, it would be enough that Tacker Zed release will provide their support versions. 08:36:34 <yuta-kazato> In addition, "this is just a suggestion", but how about agreeing on support versions for the next release in PTG? 08:37:08 <yasufum> do you have any idea? 08:38:13 <yasufum> s/any idea/any other comment/ 08:39:02 <yuta-kazato> That's all from my side. I would like to hear from Tacker team members, thanks. 08:39:46 <yasufum> I agree to the suggestion basically 08:40:50 <ueha> I think we need to change Kubernetes version [1] and Helm version [2] to see if they pass the test. 08:40:55 <ueha> [1] https://opendev.org/openstack/tacker/src/branch/master/.zuul.yaml#L471 08:41:02 <ueha> [2] https://opendev.org/openstack/tacker/src/branch/master/.zuul.yaml#L531 08:42:29 <ueha> When testing on Zuul, I think it is important that the kuryr-kubernetes support the latest version of Kubernetes when testing on Zuul. 08:42:50 <ueha> The default version currently installed by the kuryr-kubernetes is 1.22.3 [3]. 08:42:51 <ueha> . 08:42:57 <ueha> [3] https://opendev.org/openstack/kuryr-kubernetes/src/branch/master/devstack/settings#L30 08:43:52 <ueha> The Kubernetes environment may need to be created without the use of the kuryr-kubernetes.. 08:43:52 <ueha> we need to be careful it. 08:44:59 <yasufum> yeah 08:45:01 <ueha> That's all from my side. 08:45:29 <yuta-kazato> Thank you for your information! I agree. 08:46:30 <ueha> Thanks for your suggestion too :) 08:47:28 <yasufum> ok, let's continue to discussion for making a decision of the topic at the next PTG. thanks. 08:47:35 <masaki-ueno> You mentioned about kubernetes and helm versions, but how about Prometheus? 08:49:29 <masaki-ueno> Since we added the support of FM and PM in Zed, I think we should determine supported version of Prometheus (and Alertmanager) 08:51:14 <yasufum> sure 08:51:28 <masaki-ueno> Even if Prometheus is a sample of External Monitoring Tool, we should add description which version is available for the documentation. 08:51:48 <yuta-kazato> +1, we should discuss and decide supported version of Prometheus too! thank you for your suggestion:) 08:52:41 <yasufum> It must be the similar issue. So, we should have a slot for the topic at the PTG. 08:53:57 <yuta-kazato> OK! thanks. 08:54:07 <yasufum> I'd appreciate if you suggest your concern on the etherpad. 08:54:43 <yasufum> I'm going to upload the etherpad page for the next antelop :) 08:54:58 <yasufum> please wait for a while. 08:55:52 <yuta-kazato> Thanks a lot. 08:56:15 <yasufum> It seems enough for the topic for now. 08:57:05 <yasufum> yuta-kazato: thanks for the suggestion. 08:57:43 <yasufum> move on to the next topic. 08:58:08 <yasufum> #topic CFP for reducing resources for FTs 08:59:01 <yasufum> It's just a brief suggestion for our the latest serious 08:59:04 <yasufum> issue 08:59:06 <yasufum> ... 08:59:59 <yasufum> We has several failures on zuul gate at the end of R3. 09:00:17 <yasufum> It might be because of some shortage of resources on CI. 09:01:26 <yasufum> Although I have no idea for the problem, we should reduce the amount of FTs to avoid such a problem at the end of antelop. 09:02:10 <yasufum> So, I'd appreciate if you suggest any idea for the issue to discuss at the next PTG. 09:02:19 <yasufum> Do you have any comment? 09:04:08 <yasufum> good 09:05:07 <yasufum> go on to the third topic. 09:05:10 <yasufum> # Backport for stable/zed 09:06:12 <yasufum> We couldn't merge all candidate patches for zed, four patch were over the deadline. 09:07:11 <yasufum> So, I decided to backport them in stable/zed. 09:07:46 <yasufum> Three patches have already merged to master, but just one remained. 09:08:00 <yasufum> #link https://review.opendev.org/c/openstack/tacker/+/857082 09:08:53 <yasufum> I'd like to merge it before the end of R2, and upload stable patches for four patches at once. 09:09:31 <yasufum> That's all. Do you have any comment? 09:10:56 <ueha> I think the above patch is waiting for the document to be added. 09:11:13 <yasufum> yes 09:11:54 <yasufum> The owner agreed to add docs. 09:12:02 <ueha> I’m afraid if we can merge this week. 09:13:06 <ueha> This Friday is a holiday in Japan, so there are only 2 days left. but I will do my best to review it when it is added. 09:14:56 <yasufum> thanks 09:16:56 <yasufum> OK 09:17:55 <yasufum> It might be enough for today's discussion. 09:18:11 <yasufum> So, let's close this meeting if you have no more comment. 09:19:14 <yasufum> Thank you for joining, bye! 09:19:22 <yuta-kazato> thanks, bye. 09:19:31 <ueha> Thanks, bye 09:19:37 <yasufum> #endmeeting