Dear SWITCHengines users,
Compute nodes in Lausanne are experiencing issues communicating with the message queuing service of SWITCHengines. This means that control plane operations such as starting/stopping/resizing VMs are currently unavailable. Running VMs are not impacted. Underlying connectivity is verified, and we continue to investigate.
All the best,
Ann Harding
Dear SWITCHengines users,
We have observed an outage to the SWITCHengines storage infrastructure in Zürich. Access to object storage and volumes in Zürich is impacted. We are investigating and working on a resolution and will update when more information is available or at latest within 2 hours.
Best Regards,
Ann
Dear SWITCHengines users,
Due to a network hardware failure, we have an emergency maintenance on the control plane for SWITCHengines from 21:00 tonight, march 11th.
While running VMs are not impacted, during this time it will not be possible to change or manage your VM for up to an hour. This is a precautionary measure to protect the control plane as we complete work to mitigate the hardware failure.
All the best,
Ann Harding
Dear SWITCHengines Users / Admins
We are not only busy improving SWITCHengines as a platform but we are also constantly working on making the actual daily work as end users with it easier. We are happy to announce the following improvements:
Machine Types / Flavors
Life Cycle management of images
GPU set-up improvements
Handling of project deletions (deletion automation with grace period)
Machine Types / Flavors
What is the difference between an m1.xlarge and c1.4-16 flavor? Yes, it’s challenging to understand what the nomenclature means - and we too get confused with the naming.
The flavors names are a historically grown collection of VM sizes and they are difficult to understand. We are working to clean things up over time. Until then, we added a necessary documentation where you find an overview what kind of flavors we serve and what the recommended configurations are.
Most users will not see the flavor choice anymore, as we have released a new version of our customized SWITCHengines Quickstart that only exposes the number of CPUs and the amount of RAM that a VM should have, as well as the desired root volume size (and if you’d like to have the root volume on SSD backed storage).
Docs: https://help.switch.ch/engines/documentation/flavors/ <https://help.switch.ch/engines/documentation/flavors/>
Life Cycle management of images
We have changed the way we handle the image life cycle for our Linux images. Until now, when we updated our images, the outdated images were deactivated and no longer accessible to our customers. This has implications on automation processes that rely on a known (and available) version of an image. Based on user feedback, we have re-engineered this process.
With the recent SWITCHengines software upgrades, we can now hide old images but still make them accessible via their image ID. This leads to an uncluttered user experience for regular users (they only see the most recent images) while preserving access to older images where the image ID is known.
We still reserve the right to delete images definitively which are not used anymore, and which have a certain age. If you need an image for an undefined amount of time, you should upload it as a private image.
GPU setup improvements
We have also simplified the access to GPUs. Until now, we had separate, large Ubuntu images with Nvidia drivers installed. This has led to confusion (a GPU image doesn’t by default have a GPU enabled VM), bloat and duplication of images.
Now we provide an installation script to our standard Ubuntu images (Bionic/Focal) that allows you to install the NVIDIA GPU drivers within minutes. This is a lightweight solution and provides you with the needed flexibility and is also the recommended way of installing the software. The script installs the Nvidia drivers, Docker and nvidia-docker2 that allows you to use the GPU on the VM and lets you run a container that can access the GPU.
Docs: https://help.switch.ch/engines/documentation/gpu_support/ <https://help.switch.ch/engines/documentation/gpu_support/>
Handling of projects set to delete after a given time
Our Admin UI (https://engines.admin.switch.ch <https://engines.admin.switch.ch/>) has allowed you to specify an end date of a project.
Based on your feedback, we have made the following changes:
Warning emails are sent out 28 days, 7 days and 1 day in advance of the deletion
If a project is set to “allow extension” the user receiving these mails can extend the project duration on their own by using a web form
This should make for a smoother user experience (and less work to reset the end dates for projects that go on longer than expected)
We are constantly working on other new features and the removal of pain points to improve the user experience and the stability of SWITCHengines . Please don’t hesitate to contact us (engines-support(a)switch.ch) if you have some notes about the above points or about further improvements. As you know, we are open to customer feedback and we try to fulfill as much as possible.
Best regards,
The SWITCHengines people at SWITCH
--
SWITCH
Jens-Christian Fischer, Teamlead Managed Applications & Services
Werdstrasse 2, P.O. Box, 8021 Zurich, Switzerland
phone +41 44 268 15 15, direct +41 44 268 15 71
jens-christian.fischer(a)switch.ch
http://www.switch.ch
