Tool-forge users can ignore this email, it only concerns VPS project owners.
Long ago, the Wikimedia Operations team made the decision to phase out use of Ubuntu servers in favor of Debian. It's a long, slow process that is still ongoing, but in production Trusty is running on an ever-shrinking minority of our servers.
As Trusty becomes more of an odd duck in production, it grows harder to support in Cloud Services as well. Right now we have no planned timeline for phasing out Trusty instances (there are 238 of them!) but in anticipation of that phase-out we'd like to discourage the addition of new Trusty hosts to the cloud.
Step one[1] is to prevent people from creating new Trusty images unless they really, really need them. We would like to remove Trusty from the default available list of base images and make it available for new VMs only via special request on phabricator. The questions for you are:
1) Would that change make your life a lot harder? 2) If yes, can you name a date after which it /won't/ make your life harder?
If the loss of Trusty doesn't worry you, feel free to ignore this email. In the event of a silent (or relatively quiet) response, I'll pull Trusty from the default image list sometime in the next few weeks.
- Andrew (+ the rest of the Cloud team)
[1] https://phabricator.wikimedia.org/T161899
_______________________________________________ Wikimedia Cloud Services announce mailing list Cloud-announce@lists.wikimedia.org (formerly labs-announce@lists.wikimedia.org) https://lists.wikimedia.org/mailman/listinfo/cloud-announce
The documentation (https://wikitech.wikimedia.org/wiki/Help:Toolforge/Grid) says that the only release available to jsub is Trusty. It also says we shouldn't run our jobs on Kubernetes ( https://wikitech.wikimedia.org/wiki/Help:Toolforge/Kubernetes).
Can you clarify how we should run our jobs?
Thanks
On Tue, Oct 10, 2017 at 5:33 PM, Andrew Bogott abogott@wikimedia.org wrote:
Tool-forge users can ignore this email, it only concerns VPS projectowners.
Long ago, the Wikimedia Operations team made the decision to phase outuse of Ubuntu servers in favor of Debian. It's a long, slow process that is still ongoing, but in production Trusty is running on an ever-shrinking minority of our servers.
As Trusty becomes more of an odd duck in production, it grows harderto support in Cloud Services as well. Right now we have no planned timeline for phasing out Trusty instances (there are 238 of them!) but in anticipation of that phase-out we'd like to discourage the addition of new Trusty hosts to the cloud.
Step one[1] is to prevent people from creating new Trusty imagesunless they really, really need them. We would like to remove Trusty from the default available list of base images and make it available for new VMs only via special request on phabricator. The questions for you are:
- Would that change make your life a lot harder?
- If yes, can you name a date after which it /won't/ make your life
harder?
If the loss of Trusty doesn't worry you, feel free to ignore thisemail. In the event of a silent (or relatively quiet) response, I'll pull Trusty from the default image list sometime in the next few weeks.
- Andrew (+ the rest of the Cloud team)
[1] https://phabricator.wikimedia.org/T161899
Wikimedia Cloud Services announce mailing list Cloud-announce@lists.wikimedia.org (formerly labs-announce@lists.wikimedia.org) https://lists.wikimedia.org/mailman/listinfo/cloud-announce _______________________________________________ Wikimedia Cloud Services mailing list Cloud@lists.wikimedia.org (formerly labs-l@lists.wikimedia.org) https://lists.wikimedia.org/mailman/listinfo/cloud
On Thu, Oct 12, 2017 at 6:04 PM, Magog The Ogre magog.the.ogre@gmail.com wrote:
The documentation (https://wikitech.wikimedia.org/wiki/Help:Toolforge/Grid) says that the only release available to jsub is Trusty. It also says we shouldn't run our jobs on Kubernetes (https://wikitech.wikimedia.org/wiki/Help:Toolforge/Kubernetes).
Can you clarify how we should run our jobs?
This is a great clarifying question. The proposed change would only effect the creation of new virtual machines in Cloud VPS projects. Toolforge is a Cloud VPS project, but running a tool on the job grid does not involve creating a new virtual machine. All of the existing grid engine nodes will remain operational.
The intent of https://phabricator.wikimedia.org/T161899 is to make creation of new Trusty virtual machines an explicit opt-in action for each Cloud VPS project. If we do need to create additional grid engine nodes for Toolforge we will be able to do that.
Migration of the Toolforge bastions and Grid Engine nodes to a newer operating system will happen at some point, but not for several months at least. We do not currently know if that migration will be to Debian Stretch, Ubuntu Xenial, or something else. When it does happen, the migration will be well announced and include a period of overlap were both Trusty and the replacement are available to allow tool maintainers a reasonable period to adapt their tools to the newer versions of PHP, Python, etc that will be on the replacement OS.
Bryan