September 2021 - openstack-discuss

[TripleO] Gate blocker - please hold rechecks
by Ronelle Landy 02 Oct '21

02 Oct '21

Hello All, We have a gate blocker for tripleo at: https://bugs.launchpad.net/tripleo/+bug/1945682 This tox error is impacting tox jobs on multiple tripleo-related repos. A resolution is being worked on by infra. Please hold rechecks if you are rechecking for this failure. Please note that third-party OVB jobs are also down - following an update to DNS on a network (related bug: https://bugs.launchpad.net/tripleo/+bug/1945678) We will update this list when the error is cleared. Thanks for your patience!

2 3

[oslo] oslo.metrics package for Fedora
by Hirotaka Wakabayashi 01 Oct '21

01 Oct '21

Hello Oslo Team, I am a Fedora packager. I want to package oslo.metrics for Fedora because my package uses oslo.messaging that requires oslo.metrics as you know. oslo.messaging package repository already exists in Fedora. I will take over it from the former package maintainer. the oslo.metrics repository doesn't existso I need to make it. If any concerns with it, please reply. I can update the version as soon as the new version releases by using Fedora's release monitoring system. Thanks in advance, Hirotaka Wakabayashi

4 4

[cinder][nova][requirements] RFE requested for os-brick
by Brian Rosmaita 01 Oct '21

01 Oct '21

Hello Requirements Team, The Cinder team recently became aware of a regression in os-brick that prevents Nova from attaching NVMe volumes [0]. A fix has been merged to master (yoga development) [1] and has been backported to stable/xena [2]. We've proposed a release of os-brick 5.0.1 from stable/xena [3], and are petitioning for an Requirements Freeze Exception to include os-brick 5.0.1 in the xena release. Thank you for your attention to this matter. [0] https://bugs.launchpad.net/os-brick/+bug/1945323 [1] https://review.opendev.org/c/openstack/os-brick/+/811944 [2] https://review.opendev.org/c/openstack/os-brick/+/812010 [3] https://review.opendev.org/c/openstack/releases/+/812048

2 1

[neutron] Drivers meeting agenda - 01.10.2021
by Lajos Katona 30 Sep '21

30 Sep '21

Hi, The agenda for tomorrow's drivers meeting is at [1]. We have 1 RFE to discuss: * https://bugs.launchpad.net/neutron/+bug/1885261 Add stateless firewall support to OVS firewall ** It is Liu who would like to implement this, see logs of last team meeting at [2] ** Liu can't participate as there's some celebration in China, and he is on holiday, but still we can discuss it to have better understanding. [1] https://wiki.openstack.org/wiki/Meetings/NeutronDrivers#Agenda [2] https://meetings.opendev.org/meetings/networking/2021/networking.2021-09-28… See you at the meeting tomorrow. Lajos Katona (lajoskatona

1 0

Open stack Swift
by Midhunlal Nb 30 Sep '21

30 Sep '21

Hi team, I am using Openstack rocky setup with 1 controller node 2 compute node and 1 sStorage node,Now i am planning to configure a storage backup service Swift. Where do I need to configure swift,It needs an extra Node?Please explain the requirements. Thanks & Regards Midhunlal N B +918921245637

3 2

how dpdk improve guest performance
by Satish Patel 30 Sep '21

30 Sep '21

Folks, If I am running dpdk in my compute nodes then sure it will reduce kernel overhead but how does it help to improve guest network performance? Inside guest vm it's going to use kernel to copy packet to userspace. in my example we run high performance haproxy which handles 300k to 500k connections but I have noticed performance is poor and lots of packet drops. If I run the same workload on SRIOV vms it works fine and I notice less kernel context switching inside the guest VM. I think when you run DPDK in compute then it only help compute but not your guest network performance. How do I improve guest performance also with dpdk based compute?

1 0

[nova] WSGI Apache vs Eventlet HTTP Servers
by Ammad Syed 30 Sep '21

30 Sep '21

Hi, I was reviewing below document. https://docs.openstack.org/nova/wallaby/user/wsgi.html Is apache2 wsgi provide better performance then eventlet http servers for nova api ? If yes, can you guys please share the sample or reference apache-nova-api file for wsgi ? - Ammad

1 0

[GPU] Request for suggestions for OpenStack deployment
by Jay See 30 Sep '21

30 Sep '21

Hi, We are planning to buy some servers with GPU cards from DELL(R740/R750)/HPE/Supermicro with NVIDIA A100 40GB (GPU). I have the following questions: 1. What is the best ratio for Memory to GPU? or It doesn´t matter. 2. Anyone using A100? A100 has any known issues with OpenStack deployment? 3. If you have set up with similar server with GPU, could you share the server configuration? It will help us in deciding / identifying the components we have missed. Thanks in advance. Regards, ~ Jayachander.

1 0

DC DR high availability - Snapshot taking feature and user experience
by KK CHN 30 Sep '21

30 Sep '21

List, I am in the process of setting up a DC and DR sites with openstack(ussuri, qemu kvm with debian as base OS). I would like to know what all are the options in DC and DR sites for High availability . What the architecture methods to follow at DC and DR so that, If VM crashes at one physical machine at DC ( for example 3 controllers in DC), or Host machine crashes( RAM, Hardware failure etc. ) / machine power cable detached accidentally so that no services down / VMs and applications not down. How can achieve this ? share the best practices here. Also If I take snapshots of the running VMs ( what will the user experience ? will they in freeze / logged out from applications right now they are logged in ? ) . Can we avoid the service unavailable for users while taking snapshots ? To DR we are rsyncing these snapshots after converting each VM image to qcow2 then rsyncing . ( All these we are performing on Controller node . Generally 100 GB VM will output 16 GB qcow2 image for example. ) its taking 2 minutes for snapshot creation and 20 to 30 Minutes for qcow2 conversion . ) How many snapshots and conversion to qcow2( qemu-img convert) can be performed on this controller machine where we performing this operation . ( Can we apply parallel processing for this and how ? ) . Seet he controller specs where we perform this. The controller spec ( 20 core CPUs with 1.5 TB RAM total 160 processors ) Then copying to DR ( which is 200 KM away from our DC takes 4 minutes with rsync). Then we populate each VM with these copied qcow2 images and attaching the Network ips and nat there. Is this the practice or any other good way to perform this for a robust DC and DR setup . Kindly share your thoughts. Regards, Kris

1 0

[all][tc] Technical Committee next weekly meeting on Sept 30th at 1500 UTC
by Ghanshyam Mann 29 Sep '21

29 Sep '21

Hello Everyone, Technical Committee's next weekly meeting is scheduled for Sept 30th at 1500 UTC. If you would like to add topics for discussion, please add them to the below wiki page by Wednesday, Sept 29th, at 2100 UTC. https://wiki.openstack.org/wiki/Meetings/TechnicalCommittee#Next_Meeting -gmann

1 1