Port-xen archive

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Old Index]

Re: CVS commit: src/sys/arch/xen/xen



At Thu, 22 May 2025 16:22:43 -0700, "Greg A. Woods" <woods%planix.ca@localhost> wrote:
Subject: Re: CVS commit: src/sys/arch/xen/xen
>
> We'll know in a week if it fixes anything.

Sadly it didn't really fix anything.

Time still goes wonky, here somewhat before 650,000 seconds uptime:

[Fri May 30 03:31:34 2025][ 646993.1938510] xen_rtc_set: Setting to 1748601093.332013000 s at systime 647006841230263 ns, gst 646966565398370 ns (nanouptime: 646993193850970 ns, initial: 13647379293 ns) diff(gst-(nt+init)): -40.275831893 s
[Fri May 30 03:31:45 2025][ 647004.1835228] xen_rtc_set: Setting to 1748601104.321685000 s at systime 647017830902126 ns, gst 646977554393322 ns (nanouptime: 647004183522833 ns, initial: 13647379293 ns) diff(gst-(nt+init)): -40.276508804 s
[Fri May 30 03:31:56 2025][ 647015.1731999] xen_rtc_set: Setting to 1748601115.311362000 s at systime 647028820579202 ns, gst 646988543625304 ns (nanouptime: 647015173199909 ns, initial: 13647379293 ns) diff(gst-(nt+init)): -40.276953898 s
[Fri May 30 03:32:07 2025][ 647025.1648753] xen_rtc_set: Setting to 1748601125.303037000 s at systime 647038812254578 ns, gst 646999534435018 ns (nanouptime: 647025164875285 ns, initial: 13647379293 ns) diff(gst-(nt+init)): -39.277819560 s
[Fri May 30 03:32:17 2025][ 647026.1644901] xen_rtc_set: Setting to 1748601126.302652000 s at systime 647039811869421 ns, gst 647009533390743 ns (nanouptime: 647026164490128 ns, initial: 13647379293 ns) diff(gst-(nt+init)): -30.278478678 s
[Fri May 30 03:32:26 2025](XEN) [2025-05-30 10:31:44.726] Watchdog timer fired for domain 0
[Fri May 30 03:32:26 2025](XEN) [2025-05-30 10:31:44.726] Hardware Dom0 shutdown: watchdog rebooting machine




There are also still lots of event counters firing at what I think are
excessive rates:

# vmstat -e | fgrep xen | fgrep -v xenev0
vcpu0 xen missed hardclock                            118751769  969 intr
vcpu0 xen local_ns one tick or more behind global_ns  127585891 1041 intr
vcpu0 xen global_ns prevented from running backwards  128939054 1052 intr
vcpu1 xen missed hardclock                             88227613  720 intr
vcpu1 xen local_ns one tick or more behind global_ns   25709199  209 intr
vcpu1 xen global_ns prevented from running backwards   36421443  297 intr
vcpu2 xen missed hardclock                            106878886  872 intr
vcpu2 xen local_ns one tick or more behind global_ns   14847880  121 intr
vcpu2 xen global_ns prevented from running backwards   18676556  152 intr
vcpu3 xen missed hardclock                            106876926  872 intr
vcpu3 xen local_ns one tick or more behind global_ns   14090038  115 intr
vcpu3 xen global_ns prevented from running backwards   18005325  147 intr
vcpu4 xen missed hardclock                                63950    0 intr
vcpu4 xen global_ns prevented from running backwards   79344978  647 intr
vcpu5 xen missed hardclock                                71467    0 intr
vcpu5 xen global_ns prevented from running backwards  124023990 1012 intr
vcpu6 xen missed hardclock                                61283    0 intr
vcpu6 xen global_ns prevented from running backwards   34744740  283 intr
vcpu7 xen missed hardclock                                63656    0 intr
vcpu7 xen global_ns prevented from running backwards   20660958  168 intr


I do think that while it was running OK that there might have been
slightly less wandering skew in the global system time, but that's just
a feeling, not based on any solid measurements or stats.

So, I think the change is good, though maybe not necessary, and it
clearly doesn't fix or work around the underlying Xen kernel problem.

--
					Greg A. Woods <gwoods%acm.org@localhost>

Kelowna, BC     +1 250 762-7675           RoboHack <woods%robohack.ca@localhost>
Planix, Inc. <woods%planix.com@localhost>     Avoncote Farms <woods%avoncote.ca@localhost>

Attachment: pgp8g4UHy21Ij.pgp
Description: OpenPGP Digital Signature



Home | Main Index | Thread Index | Old Index