QA (public)

Had meeting with EngInfra TL 2022-12-07 mflores. Prg CoLo will start migrating services 2023-03, bugzilla, gitlab, virtualization clusters. s390 and PowerPC will be moved as well, likely 2023-05. They should be offline for some days and then usable again after setup in Prg CoLo. x86_64+aarch64 is ordered as new. Nbg new DC will also be setup in that time. 40 racks for everything from NUE1 that does not fit/move to FC labs. Monitoring: Prg CoLo will have switches and firewalls. They shall be configured as IaC, maybe with salt or terraform. After that monitoring is planned, but I consider it doubtful if this will work out.

2022-12-08: Decided with mgriessmeier, nsinger, mflores to order 4x ARM machines for Prg CoLo to have redundancy for each o3+osd, i.e. 2xARM@o3, 2xARM@osd

Right now waiting for DC being ready for us to use or waiting for any pending questions

-> subtasks

I would like to track this outside our current backlog as we don't need to conduct that much work now.

Combining #121720 and #130955 as there is too much overlap

I wrote a message in https://mailman.suse.de/mailman/private/qa-team/2023-June/005988.html

Hi all,
Be advised that there are plans to fully empty the old Nuremberg
datacenter at the old office location "Maxtorhof" aka. NUE1 until end of
this year. This means moving services and machines to other locations or
decomissioning services or machines that are not needed anymore.
The SUSE QE Tools team will organize, execute and lead any necessary
actions concerning LSG QE services and machines as far as we know of.

How will you be impacted by this? In the best case you will only see
short outages of services during the actual migrations. Maybe you will
need to reach specific machines by new domains (FQDNs). Likely over the
next weeks and months individual services will have outages and
performance degradations. In the worst case critical machines that no
one considered will be lost and services need to be recovered with
careful and tricky reverse engineering. Good planning and reviews of
plans can mitigate that risk :)

According to current plans we want to setup new openQA workers in the
following weeks and the service openqa.suse.de and according virtual
machine will move to PRG2 (the new Prague datacenter) on 2023-07-17.
Expect an outage on https://openqa.nue.suse.com and
https://openqa.suse.de on that day.

The equivalent migration will be conducted for
https://openqa.opensuse.org at beginning of 2023-08.

Find more details in
https://progress.opensuse.org/issues/121720

Have fun,
Oliver

and an according copy in https://suse.slack.com/archives/C02CANHLANP/p1687787065732719

All remaining tasks done \o/

After more than 2.5 years starting with #100455 we concluded the work for the big multi-datacenter migration. With the great work of the QE Tools team members and with all your enduring patience we came to that achievement :)

We already have new ongoing tasks based on new hardware as well as older hardware that can be repurposed for better use. So back to work everyone! :)

The saga included 158 subtasks, pre-planning tasks, planning tasks, coordination tasks as well as actual hands-on work. Overall we migrated/decommissioned/reinstalled/repurposed 100+ physical machines plus many more virtual ones. All covering much more then just openQA but a variety of QE machines or old QA SLE and QAM machines. We helped to clean out and fully decommission 7 server rooms or labs all while ensuring operability as much as possible. For some hosts, like PowerPC, the downtime was actually longer than a year for a variety of reasons but for the most critical services the accumulated downtime was in the range of hours.