have whonixcheck look out for timekeeping watchdog kernel messages

Information

ID: 480
PHID: PHID-TASK-h6a4l5jbzoakfa6r675m
Author: Patrick
Status at Migration Time: resolved
Priority at Migration Time: Normal

Description

This…

[ 2743.843690] timekeeping watchdog: Marking clocksource 'tsc' as unstable, because the skew is too large:

… or similar it could look in dmesg.

whonixcheck or whonixcheck --verbose should add a test for that?

It could dmesg for “timekeeping watchdog” and if detected, show a warning.

Kernel messages wrt timekeeping can indicate issues which generate lots of confusion, inconvenience and even time related fingerprinting issues. See VMs with Whonix-Tor networking timeout and do not recover · Issue #1779 · QubesOS/qubes-issues · GitHub (new information to be added there soon). Which took a while to debug.

Implementation details:

  • #whonixcheck runs as user user. /var/log/kern.log is owned by group adm. So whonixcheck has no access to it. To avoid adding user user to group adm, dmesg gets used instead, which [strangely but never mind] is accessible as standard user.
  • dmesg has the added advantage, that it only shows kernel messages since last reboot. So it would not show any messages that may have occurred much earlier that were fixed by kernel or virtualizer upgrades.

Comments


marmarek

2016-03-21 12:57:32 UTC


Patrick

2016-03-21 14:16:27 UTC


Patrick

2016-03-21 14:21:18 UTC