strange issues with the system clock in all VMs, and dom0
-
upgraded from 4.1 to 4.2.2 in the spring, clock was previously normal
-
noticed the clock goes at ~ double speed on 4.2.2
-
reinstalled to carefully check that I didn’t do anything unusual 1st time. only option I chose atypically was to not install the Fedora template
-
problem persists (and is not fun to handle, of course)
-
sys-net is debian-xfce, the systemd service runs, timedatectl confirms NTP is on, timezone is correct
-
dom0 timedatectl also confirms the timezone, the date and the time are correct
-
importantly, the RTC clock time in dom0 timedatectl is correct, and reveals the difference between what the system clock believes is the correct time and what the RTC clock is maintaining
I can’t help but think this is some kind of issue with my CPU (Skylake) in combination with the new kernel version (6.6.x) being used on Qubes 4.2.x (as I say, 4.1 was completely ok immediately before upgrading)
I am hoping instead that I’ve made a stupid error instead, but it’s not easy to see what it is, as the sys-net VM is standard (and not minimal), the qvm-clock-sync command service is running, but simply not frequently enough to stop processes/applications quitting or complaining about the time being too far in the future.
I tried checking the possible system clock sources in dom0, chose the alternate option from /sys/devices/system/clocksource (only xen and tsc are available) by editing the kernel command line on system boot, and adding clocksource=tsc as a line at the end. checking in dom0 /sys/devices/system/clocksource demonstrated I added the line correctly, but the mega fast system clock remained.
Any clues? seems strange to me that no-one else apparently had this problem, from what I could find.