Running print(platform.platform()) with Python I detected that the kernel versions of the host agents on ubuntu-latest and ubuntu-18.04 differ, even though they point to the same image in the docs.
According to a list of kernel version they map to different Ubuntu versions
ubuntu-latest -> Linux-4.15.0-1061-azure-x86_64-with-glibc2.2.5 -> ubuntu 18.04
ubuntu-18.04 -> Linux-5.0.0-1023-azure-x86_64-with-glibc2.2.5 -> ubuntu 19.04
Edit: I'm wrong about kernel versions mapping to OS versions. The problem is that latest map to 16.04
⚠Do not edit this section. It is required for docs.microsoft.com ➟ GitHub issue linking.
Thank you for your feedback - I'm taking a look!
It turns out ubuntu-latest actually points to ubuntu-16.04, ubuntu-16.04 and ubuntu-18.04 both seem right though. We had to roll latest back to point to 16.04 due to capacity issues which have since been fixed. We're tracking a work item to make latest point to 18.04 again, that should go out after the holidays.
I'm going to close here since we're tracking the work to fix this on our end.
Closing this issue makes it harder for other folks to see that this is an issue.
That's fair, we can reopen until this is fixed.
From what we can tell this is yet to be fixed, using ubuntu-latest we occasionally get 16.04, and other times 18.04. Explicitly specifying ubuntu-18.04 fails to spin up the pipeline.
What's the current status on this, and is it possible to track?
Bump, still ongoing.
is this still a problem? im currently using ubuntu-latest in the pipeline
Still seeing this.
@alepauly @mitesch @densto88 customers are seeing ubuntu-latest route to 16.04 sometimes, as recently as April 7. Can you guys please investigate what's happening?
At this point we're dealing with a service bug, not a docs issue. If anyone feels like opening an issue on Developer Community then feel free, but we will also run this down internally.
@Chris113113 we looked at service-side telemetry and cannot find any evidence of an ubuntu-latest build running against the Ubuntu16 pool. To dig further, please post a Developer Community ticket with your organization name, project name, and pipeline name. A pointer to a run where it occurred would also help. Thanks!
Most helpful comment
That's fair, we can reopen until this is fixed.