© 2004
Canonical Ltd.
•
Terms of use
•
Data privacy
•
Contact Launchpad Support
•
Blog
•
Careers
•
System status
•
0e1f616
(Get the code!)
After some discussion with rharper on IRC[0], our working hypothesis is that the events that are being processed are as a result of the sgdisk command that does the partition table modification[1], _not_ the pt_update call (which boils down to `partx --update`[2]). As a result, we think we are racing the re-read of the partition table; this is consonant with the different ordering of the udev events on failure.
To test this hypothesis, I'm going to create a cloud-init package which performs its own partition table re-read (which should trigger udev events for those partitions) and `udevadm settle` after the resize. This _should_ mean that the symlink is correctly re-created. If this succeeds, we can examine what change we need to make to cloud-utils.
Tobi, I'll upload that package to the same PPA as the others, and will let you know here when it's ready. We'll want an image with that cloud-init as well as the debugging systemd units.
[0] Which should eventually show up here: https:/
[1] https:/
[2] https:/