September 2020 - openstack-discuss

[election][swift] PTL candidacy for Wallaby
by Tim Burke 26 Sep '20

26 Sep '20

I'd like to announce my candidacy for Swift PTL for the Wallaby cycle. This past year, Swift celebrated ten years running in production. Much has changed in that time: new features have been developed and polished, versions of Python have come and gone, and clusters have grown to staggering capacities. However, our commitment to operational excellence remains the same. Recently (particularly in the last several months), I've noticed our contributors increasingly have an operator's mindset. We look for more and better ways to measure Swift. We seek to reduce client impacts from config reloads and upgrades. We take greater ownership over the health and performance of our clusters. To a large extent, we're all operators now. The benefits have been enormous. We've improved performance; we've upgraded without disrupting any client requests; we've migrated clusters to Python 3 to position them well for the next ten years. Through it all, clients put ever more data into Swift. The increases in demand bring almost incomprehensible scales. We now see individual clusters sustaining tens of thousands of requests every second. We see containers with a billion objects. We see expansions that are as large as many whole clusters were just a few years ago. This is our next great challenge: how do we move away from a world where expansions are a rarity that may require a bit of a scramble and into a world of constant expansion? How can we effectively manage clusters with thousands of nodes? How do we shift from thinking in terms of petabytes to exabytes? I can't wait to see how we rise to meet this challenge. Tim Burke

1 0

The question about OpenStack instance fail to boot after unexpected compute node shutdown
by Yule Sun 26 Sep '20

26 Sep '20

I saw your message on the website http://lists.openstack.org/pipermail/openstack-discuss/2019-April/005635.ht… . And i have been troubled by the same problem for a long time . I tried to change the parameters resume_guests_state_on_host_boot=True and instance_usage_audit=True in the nova.conf ,but it didn't work . I wonder if you have solved this problem, looking forward your reply, Thank you and best regards. And my instance error looks like this . [cid:e8770762-d2b2-4630-9ce6-4d357dc92cee]

2 1

[election][zaqar] PTL candidacy for Wallaby
by hao wang 26 Sep '20

26 Sep '20

Hi, I want to propose my candidacy and continue serving as Zaqar PTL in the Wallaby cycle. For the Victoria release, the Zaqar team continues to keep the Zaqar stable and reliable, and also some great works have been done. We designed and implemented the Encrypted message in the queue to enhance the security of Zaqar. We also fixed some bugs to meet the python3 requirements. In the Wallaby cycle, I would like to continue serving this project and bring more contributors and the user requires. There are some major works I want to do in W cycle: 1.Finish the Topic resource work, to support all backends. 2.Consider to support more backends that have widely used. 3.Continue to support more security algorithms for encrypted messages. 4.Implement some new requirements from users. 5.keep Zaqar more reliable as usual. Thanks to everyone to support my work and keep contributing to Zaqar. I will do my best in the new cycle.

1 0

[openstack-ansible][election] Candidacy
by Dmitriy Rabotyagov 26 Sep '20

26 Sep '20

Hey everyone! This cycle I've decided to take part in elections and run for PTL of the OpenStack-Ansible project. I feel that OSA nowadays is a solid and stable project, and we're doing a great job to get all features in time and deploy only modern software with our deployments, as well as providing deployers with a high range of different supported options and flexibility. It's rather important to keep up with the progress. And there are main things I'd love to focus on during the upcoming release: # Integration tests coverage for all roles Despite migrating tests for most of the projects to the integrated repo, we still have several projects not covered with proper testing (like Zun). I also would love to pay attention to passing appropriate tempest scenarios for all projects instead of running some default ones. # Cleaning up bugtracker I think one of our highest priorities should be project stability. And taking care of bugs might help us heavily with that. We haven't been paying enough attention to our Launchpad for a while and I think we should change that by cleaning up stale bugs, so we could focus on real ones. Returning bug triage to the meeting might be also an efficient way to handle them. # Roles cleanup We have a lot of things left in code, like workarounds for upgrades, upstream bugs, support for obsolete operating systems, etc. Removal/replacement of deprecated OpenStack options should be also included. # Usage of the new pip resolver Our venv build process relies on provided constraints heavily, which isn't valid in case of the new pip resolver usage, so we will need to make appropriate adjustments into the process to keep the ability to define specific package requirements and use modern pip. Thank you for taking into consideration my nomination.

1 0

Re: [magnum][infra] Can We Delete the OpenDev Fedora Atomic Images Mirror
by feilong＠catalyst.net.nz 25 Sep '20

25 Sep '20

1 0

OpenstackSwift - rclone Copy/Sync Large Objects set (150Mil objects - 80TB) from Tenant PROD1 to PROD2
by prem shivakumar 25 Sep '20

25 Sep '20

Hi Open-stack users, Our current setup: PROD1 Container1 Dir1/Dir2/YYYYMMDD.tar Want to move to: PROD2 Container2_YYYYMM Dir1/Dir2/YYYYMMDD.tar I got all the objects metadata(Object names) using “swift list Container1” I am currently using rclone where I am passing metadata(Object names) info of “Container1”. All the Auth variables exist in rclone.conf. Example command: rclone copy PROD1: Dir1/Dir2/YYYYMMDD.tar PROD2:Container2_YYYYMM/Dir1/Dir2/ YYYYMMDD.tar Currently, I am facing slowness in copying files from PROD1 to PROD2. On an average, It takes about 5s to copy the object from PROD1 to PROD2. If I count the time against 150M its more than 10000Days. I have also tried “rclone sync with —fast-list” it is failing as it is unable to keep the listing in the memory. In this case, is there any easier and faster solution to copy data from Container1 to Container2? Thank you.

1 0

[magnum][infra] Can We Delete the OpenDev Fedora Atomic Images Mirror
by Clark Boylan 25 Sep '20

25 Sep '20

I'm in the process of removing our fedora-30 images and noticed that we are still mirroring Fedora Atomic 28 and 29 images here: https://mirror.dfw.rax.opendev.org/fedora/atomic/stable/. Are these images in use anywhere? I would like to clean them up as they are even more stale than the fedora-30 nodepool images. Using http://codesearch.openstack.org I don't see anything using them but that only searches the master branches of our git repos. Clark

3 3

[election][kolla] PTL candidacy for Wallaby
by Mark Goddard 25 Sep '20

25 Sep '20

Hi, I'd like to nominate myself to serve as the Kolla PTL for the Wallaby cycle. I have been PTL for the last 3 cycles, and would like the opportunity to continue to lead the team. I am pleased with the current state of the project. We have an active community, produce interesting and useful features, and regularly receive positive feedback. As with many contributors in a less certain world, I have felt the squeeze on upstream this cycle. I'd like to thank those who have stepped up to help when downstream (or holiday) means I'm unavailable, and those who generally help out with project admin tasks. This cycle we added two new meetings to the calendar. Kolla Klub started well, with fantastic attendance and some great discussions. Over time engagement has been more variable. I think we need to assess whether to continue regular meetings, and if so, how to keep it interesting. The Kolla Kall is more development oriented, and has been useful for focussing on some specific issues. For this meeting I think we need to be sure that it is the best use of people's time, and be a bit more dynamic about content and attendance. One area where I feel we could improve as a project is in integrating contributors in non-EU/US time zones. Language, culture and time zones can make this a challenge, but everyone benefits from a connected global community. We made a start on improving our documentation this cycle, let's keep pushing on that. Thanks for reading, Mark Goddard (mgoddard)

1 0

[InteropWG] Looking for k8s Ready OpenStack / Open Infra and BMaaS Add-on Conformance Candidates
by prakash RAMCHANDRAN 25 Sep '20

25 Sep '20

Hi all, Here is summary of Friday alternate week calls of Interop(next Oct 9th) join us on meetpad - refer https://etherpad.opendev.org/p/interop We are looking forward to prep work for PTG on October23 Monday & October 30 Friday Interop slots ( https://ethercalc.openstack.org/7xp2pcbh1ncb ) 1. Seeking Candidate for "Kubernetes Ready OpenStack" Branding - need Ideation from different Project teams with variations you seek (see eg.) eg. Distros RHSOP with Cinder CSIVIO with Swift CSI eg. Hosted version Canonical OpenStack Distro with Manila CSISUSE OpenStack with Octavia Load Balances Ingress Controller Add your ideas to - https://etherpad.opendev.org/p/2020-Wallaby-interop-brainstorming 2. Seeking Candidates for OpenStack ready Bare Metal as a Service Add-ons IronicMaaS Add your ideas to - https://etherpad.opendev.org/p/2020-Wallaby-interop-brainstorming All Vendors and PTLs interested in pursuing ideas for their next releases OpenStack or Open Infra release please add your ideas to etherpads above.Reply to this chain and we will try consolidate Interop in Bare metal & K8s era. ThanksPrakash / MarkFor InteropWG

1 0

[tripleo]] tripleo victoria rc1
by Marios Andreou 25 Sep '20

25 Sep '20

hello TripleO folks o/ we don't normally/aren't expected to cut release candidates. However after discussion with and guidance from Wes I decided to propose an rc1 as practice since I haven't used the new releases tool before [4]. The proposal is up at [1] but I've set -1 there on request from reviewers; we can't merge that until [2] merges first. If you are interested please check [1] and add any comments. In particular let me know if you want to include a release for one of the independent repos [3] (though for validations-common/libs we need to reach out to the validations team they are managing those AFAIK). thanks, hope you enjoy your weekend! [1] https://review.opendev.org/#/c/754346/ [2] https://review.opendev.org/754347 [3] https://releases.openstack.org/teams/tripleo.html#independent [4] https://releases.openstack.org/reference/using.html#using-new-release-comma…

1 0