Re: Weird failure in explain.out with OpenBSD

On 2021-11-11 02:15, Thomas Munro wrote:
> On Thu, Nov 11, 2021 at 1:15 PM Michael Paquier <michael(at)paquier(dot)xyz> wrote:
>> morepork uses 6.9 now, but it has been recently upgraded from 5.4 or
>> a version close to that, right? I am wondering if 7.0 may help
>> regarding this issue. Postgres is not wrong here.
>
> I dunno. Clocks on virtualised systems and even metal seem to be a
> minefield of quirks and heuristics. Some discussion, may or may not
> be relevant:
>
> https://marc.info/?l=openbsd-tech&m=161657532610882&w=2

Hm, interesting.

> Some starter questions for Mikael would be: could you please check
> which clock source it's using?,

It's disabling TSC:

cpu1: disabling user TSC (skew=-4465)

so it seems to be using the acpitimer?

acpitimer0 at acpi0: 3579545 Hz, 24 bits

> is it under a hypervisor, and if so
> which?,

VMWare ESXI 6.7 latest updates installed.

> what is the CPU model?, what are other kernels choosing when
> running as guests on the same hypervisor (if applicable)?

AMD Ryzen 5 3600X

> Did
> anything interesting happen on either the guest or host operating
> system (if not bare metal) some time around 2021-11-10 21:59 CET?

Nope, not that I can see.

> I'm no expert on this stuff but something tells me that a 124ms leap
> needs a different explanation than the sub-µs difference reported
> earlier...

It's running ntpd to keep time in sync if that matters?

/Mikael