You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
This repository has been archived by the owner on Jan 16, 2024. It is now read-only.
On KVM, one of the events to trigger recontextualization in VM is the change of the context CD-ROM.
On Debian/Ubuntu/RH, the udev generates 2 nearly same events about the CD-ROM change. In configuration of context packages, we do have a silly filtering by event sequential number to catch only a one of those events (as there is nothing else how to distinguish them):
which relies on the fact that the events have contiguous sequential numbers, where one will be even and one odd.
After a stress testing we found a rare cases when such events don't have contiguous sequential numbers. They don't match the udev configuration and don't trigger recontextualizatoin. In following example we see both sequential numbers are odd (X, X+2):
On systemd platforms, we trigger reconfiguration service with --no-block and the second event doesn't trigger another run, as the service is already running. On minority of non-systemd Debian platforms (Devuan), two reconfiguration services are started, but the second one doesn't detect any changes and terminates.
On KVM, one of the events to trigger recontextualization in VM is the change of the context CD-ROM.
On Debian/Ubuntu/RH, the udev generates 2 nearly same events about the CD-ROM change. In configuration of context packages, we do have a silly filtering by event sequential number to catch only a one of those events (as there is nothing else how to distinguish them):
addon-context-linux/src/lib/udev/rules.d/65-context.rules##deb.one
Lines 15 to 21 in bfba520
which relies on the fact that the events have contiguous sequential numbers, where one will be even and one odd.
After a stress testing we found a rare cases when such events don't have contiguous sequential numbers. They don't match the udev configuration and don't trigger recontextualizatoin. In following example we see both sequential numbers are odd (X, X+2):
It should be evaluated how current systems behave if we drop the filtering.
The text was updated successfully, but these errors were encountered: