Next Thursday we'll be upgrading to OpenStack version 'Queens' starting
at 15:00 UTC.
The upgrade will take a couple of hours. During the upgrade process,
Horizon (and associated OpenStack APIs) will be disabled. There may
also be brief network interruptions during the upgrade.
Toolforge and existing VMs should be largely unaffected if all goes well
-- there may be some service interruption if we encounter networking issues.
- Andrew + the WMCS team
Hello all,
For almost a year, the Wikidata development team has been working on the
task of redesigning and migrating the wb_terms table, which had become too
big and unsustainable over the years.
You can read the tale of our journey on this blog post: Come to Terms with
Changes
<https://phabricator.wikimedia.org/phame/post/view/195/coming_to_terms_with_…>
If you’re a tool maintainer and your tool queries directly the Labs
database replicas, you can read more details
<https://lists.wikimedia.org/pipermail/wikidata/2020-March/013901.html>
about the next steps and how to update your code.
Congratulations to all the developers involved in this big project!
Cheers,
--
Léa Lacroix
Project Manager Community Communication for Wikidata
Wikimedia Deutschland e.V.
Tempelhofer Ufer 23-24
10963 Berlin
www.wikimedia.de
Wikimedia Deutschland - Gesellschaft zur Förderung Freien Wissens e. V.
Eingetragen im Vereinsregister des Amtsgerichts Berlin-Charlottenburg unter
der Nummer 23855 Nz. Als gemeinnützig anerkannt durch das Finanzamt für
Körperschaften I Berlin, Steuernummer 27/029/42207.
On 2020-03-03, Brooke completed the automatic migration phase of the
2020 Kubernetes migration by moving the last workloads from the legacy
Kubernetes cluster to the 2020 Kubernetes cluster [0].
All Toolforge tools using `webservice --backend=kubernetes ...` and/or
manually maintained Kubernetes objects are now running on the 2020
Kubernetes cluster. The Toolforge admin team is in the process of
tearing down the legacy cluster and cleaning up various documentation
and tooling related to it [1].
This project involved a lot of hard work that most of the Toolforge
community did not see. Brooke and Arturo started planning things over
a year ago [2] to ensure that the Toolforge admin team would be able
to complete this migration with a minimum amount of disruption to
tools and their maintainers. Along the journey they researched
Kubernetes best practices and recommendations, read and re-read
numerous tutorial and how-to docs, and designed a completely new
process to automate the deployment of Kubernetes in Toolforge. They
also sought and received help from other Toolforge admins, Wikimedia
Foundation staff, and technical volunteers. This was a truly
collaborative effort.
I am very happy to say that in my opinion we have a well automated and
monitored Kubernetes cluster in Toolforge today. There are many more
features that we will continue to work on as we try to make Kubernetes
use in Toolforge easier for everyone, but we can only do that work
because we now have this solid base to build on. I look forward to
announcements of many more features in the coming months.
Thank you to our alpha and beta testers who found more edge cases and
made good suggestions for simplifying things. Thank you all for your
patience and understanding when things did not go quite as planned
during this process. And finally thank you in advance for the edits
that will be made to help pages on Wikitech and elsewhere as we all
work on bug #1 (improving documentation).
[0]: https://phabricator.wikimedia.org/T246519
[1]: https://phabricator.wikimedia.org/T246689
[2]: https://phabricator.wikimedia.org/T214513
Bryan, on behalf of the Toolforge admin team and the Wikimedia Cloud
Services team
--
Bryan Davis Technical Engagement Wikimedia Foundation
Principal Software Engineer Boise, ID USA
[[m:User:BDavis_(WMF)]] irc: bd808
Time for another hardware repair task! A lot of these fragile hosts are
slated for replacement within the year but in the next time we need to
nurse them along.
On Friday I'm going to drain cloudvirt1006 for a battery replacement.
Starting around 15:00 UTC the following VMs will be copied to another
host, which means a short period of downtime and a reboot:
phabricator-stage-1001
icinga-dev-01
tool-buster
dwl-test
deployment-echostore01
deployment-sessionstore03
dashiki-02
mwv-builder-03
images
af-nb-fe
traffic-ncredir
deployment-logstash03
media-streaming
bstorm-nfs-test
maps-wma
cvn-app9