clock gets desynced
|Reported by:||joanna||Owned by:||marmarek|
It seems like the qvm-sync-clock (most likely run by a cron job) has desynced my time. After resume the time was correct, but a few moments later it got set incorrectly to roughly the time when the machine was suspended a few hours back.
It seems like the ntpdate in the netvm must have failed (e.g. due to networking problem), but the qvm-sync-clock script must have missed that, and used the incorrect time from the netvm anyway.
Here's a snippet of my messages log from Dom0:
Feb 9 09:27:25 dom0 kernel: [229015.168850] PM: resume of devices complete after 699.879 msecs Feb 9 09:27:25 dom0 kernel: [229015.169306] Restarting tasks ... done. Feb 9 09:27:25 dom0 kernel: [229015.174246] video LNXVIDEO:00: Restoring backlight state Feb 9 09:27:25 dom0 kernel: [229015.180839] usb 2-1.1: USB disconnect, address 9 Feb 9 09:27:25 dom0 pulseaudio: ratelimit.c: 252 events suppressed Feb 9 09:27:28 dom0 kernel: [229017.403431] pciback 0000:00:19.0: get owner: 8 Feb 9 09:27:28 dom0 kernel: [229017.403952] pciback 0000:00:19.0: get owner: 8 Feb 9 09:27:28 dom0 kernel: [229017.457967] pciback 0000:00:19.0: get owner: 8 Feb 9 09:27:28 dom0 kernel: [229017.458436] pciback 0000:00:19.0: get owner: 8 Feb 9 01:16:23 dom0 pulseaudio: ratelimit.c: 256 events suppressed Feb 9 09:36:55 dom0 kernel: [229586.631896] usb 1-1.6: USB disconnect, address 6
Note how the time was correct, and then suddenly changed to 01:16:23. Later it was reverted back to the correct value after I manually run qvm-sync-clock...
Interestingly I made an experiment: 1) I disabled networking, and 2) run ntpdate restart in the netvm, and 3) checked the $?, thinking it might be 0, and so explaining this problem. But it was correctly set to 1, so this doesn't explain how qvm-sync-clock might got a wrong time from netvm...
Change History (8)
comment:7 Changed 17 months ago by marmarek
- Owner changed from joanna to marmarek
- Status changed from new to accepted