bhyve clock issue

Ilario Febi ifebi at schema31.it
Tue Apr 26 13:50:57 UTC 2016


Yes,

I found errors in /var/log/messages and fix adding:

kern.timecounter.hardware=TSC-low
into /etc/sysctl.conf file

Thanks for support.



Il 14/04/16 18:13, Neel Natu ha scritto:
> Hi,
>
> Did you see anything interesting in the ntpd log file around the time the large skew happened?
>
> Best
> Neel
>
>> On Apr 13, 2016, at 3:44 AM, Ilario Febi <ifebi at schema31.it> wrote:
>>
>> Hi All
>>
>> I have a strange issue about the clock of some FreeBSD Bhyve guests,
>> maybe someone of you can help me to diagnose it and work on a
>> solution.
>>
>> In detail, sometimes (without an apparent regularity) one of four
>> bhyve guest's clock goes exactly 7 minutes and 10 seconds in the
>> future.
>>
>> This clock stays unaligned for about 20 minutes, then, without any
>> human intervention, the clock realign to proper condition.
>>
>> My configuration is:
>>
>> envrm02 (bhyve HOST, 10.2-RELEASE-p14)
>> -> og-qa (bhyve guest - 10.2-RELEASE-p14)
>> -> og-ci (bhyve guest - 10.2-RELEASE-p14)
>> -> og-test (bhyve guest - 10.2-RELEASE-p14)
>> -> og-prod-rm (bhyve guest - 10.2-RELEASE-p14)
>> All the og-* machines just have a running copy of tomcat.
>>
>> I built a simple test to catch the error with python/fabric... the test is:
>>
>> fab -R ENV,OG,OG-PROD -- date | awk -F ': ' '/2016/{print $2, $1}'
>>
>> which simply run the command "date" on all hosts.
>>
>> So, the result is:
>> @Mac-mini ~/p/f/> ./check_clock.sh
>> Thu Apr  7 17:38:00 CEST 2016 [envrm02] out
>> Thu Apr  7 17:38:00 CEST 2016 [og-qa] out
>> Thu Apr  7 17:38:00 CEST 2016 [og-ci] out
>> Thu Apr  7 17:38:00 CEST 2016 [og-test] out
>> Thu Apr  7 17:38:00 CEST 2016 [og-prod-rm] out
>>
>> @Mac-mini ~/p/f/> ./check_clock.sh
>> Thu Apr  7 18:38:00 CEST 2016 [envrm02] out
>> Thu Apr  7 18:45:10 CEST 2016 [og-qa] out
>> Thu Apr  7 18:38:00 CEST 2016 [og-ci] out
>> Thu Apr  7 18:38:00 CEST 2016 [og-test] out
>> Thu Apr  7 18:38:00 CEST 2016 [og-prod-rm] out
>>
>> @Mac-mini ~/p/f/> ./check_clock.sh
>> Thu Apr  7 18:41:05 CEST 2016 [envrm02] out
>> Thu Apr  7 18:48:15 CEST 2016 [og-qa] out
>> Thu Apr  7 18:41:05 CEST 2016 [og-ci] out
>> Thu Apr  7 18:41:05 CEST 2016 [og-test] out
>> Thu Apr  7 18:41:05 CEST 2016 [og-prod-rm] out
>>
>> @Mac-mini ~/p/f/> ./check_clock.sh
>> Thu Apr  7 18:46:51 CEST 2016 [envrm02] out
>> Thu Apr  7 18:46:52 CEST 2016 [og-qa] out
>> Thu Apr  7 18:46:52 CEST 2016 [og-ci] out
>> Thu Apr  7 18:46:53 CEST 2016 [og-test] out
>> Thu Apr  7 18:46:53 CEST 2016 [og-prod-rm] out
>>
>> The configuration is the same for all hosts and is this:
>> @Mac-mini ~/p/f/> fab -R ENV,OG,OG-PROD -- cat /etc/rc.conf | grep ntpd
>> [envrm02] out: ntpd_enable="YES"
>> [og-qa] out: ntpd_enable="YES"
>> [og-qa] out: ntpd_sync_on_start="YES"
>> [og-ci] out: ntpd_enable="YES"
>> [og-ci] out: ntpd_sync_on_start="YES"
>> [og-test] out: ntpd_enable="YES"
>> [og-test] out: ntpd_sync_on_start="YES"
>> [og-prod-rm] out: ntpd_enable="YES"
>> [og-prod-rm] out: ntpd_sync_on_start="YES"
>>
>> @Mac-mini ~/p/f/> fab -R ENV,OG,OG-PROD -- cat /etc/ntp.conf | grep -v 'out: #'
>> [envrm02] Executing task '<remainder>'
>> [envrm02] run: cat /etc/ntp.conf
>> [envrm02] out:
>> [envrm02] out: server 0.freebsd.pool.ntp.org iburst
>> [envrm02] out: server 1.freebsd.pool.ntp.org iburst
>> [envrm02] out: server 2.freebsd.pool.ntp.org iburst
>> [envrm02] out:
>> [envrm02] out: restrict default limited kod nomodify notrap nopeer noquery
>> [envrm02] out: restrict -6 default limited kod nomodify notrap nopeer noquery
>> [envrm02] out: restrict 127.0.0.1
>> [envrm02] out: restrict -6 ::1
>> [envrm02] out: restrict 127.127.1.0
>>
>> [og-qa] Executing task '<remainder>'
>> [og-qa] run: cat /etc/ntp.conf
>> [og-qa] out:
>> [og-qa] out: server 0.freebsd.pool.ntp.org iburst
>> [og-qa] out: server 1.freebsd.pool.ntp.org iburst
>> [og-qa] out: server 2.freebsd.pool.ntp.org iburst
>> [og-qa] out:
>> [og-qa] out: restrict default limited kod nomodify notrap nopeer noquery
>> [og-qa] out: restrict -6 default limited kod nomodify notrap nopeer noquery
>> [og-qa] out: restrict 127.0.0.1
>> [og-qa] out: restrict -6 ::1
>> [og-qa] out: restrict 127.127.1.0
>>
>> [og-ci] Executing task '<remainder>'
>> [og-ci] run: cat /etc/ntp.conf
>> [og-ci] out:
>> [og-ci] out: server 0.freebsd.pool.ntp.org iburst
>> [og-ci] out: server 1.freebsd.pool.ntp.org iburst
>> [og-ci] out: server 2.freebsd.pool.ntp.org iburst
>> [og-ci] out:
>> [og-ci] out: restrict default limited kod nomodify notrap nopeer noquery
>> [og-ci] out: restrict -6 default limited kod nomodify notrap nopeer noquery
>> [og-ci] out: restrict 127.0.0.1
>> [og-ci] out: restrict -6 ::1
>> [og-ci] out: restrict 127.127.1.0
>>
>> [og-test] Executing task '<remainder>'
>> [og-test] run: cat /etc/ntp.conf
>> [og-test] out:
>> [og-test] out: server 0.freebsd.pool.ntp.org iburst
>> [og-test] out: server 1.freebsd.pool.ntp.org iburst
>> [og-test] out: server 2.freebsd.pool.ntp.org iburst
>> [og-test] out:
>> [og-test] out: restrict default limited kod nomodify notrap nopeer noquery
>> [og-test] out: restrict -6 default limited kod nomodify notrap nopeer noquery
>> [og-test] out: restrict 127.0.0.1
>> [og-test] out: restrict -6 ::1
>> [og-test] out: restrict 127.127.1.0
>>
>> Some observation:
>> * The clock of envrm02 (the HOST) always seems to be correct.
>> * The issue happens on a (apparently) random guest (not always the same one).
>> * The issue happens always on one host at a time.
>>
>> The bhyve command line is, roughly, this one:
>>
>> /usr/sbin/bhyveload -m 2048 -d /san_storage/VMfs/$machine/$machine.img $machine
>> /usr/sbin/bhyve -c 2 -m 2048 -A -H -P -s 0,hostbridge -s
>> 2,ahci-hd,/san_storage/VMfs/$machine/$machine.img -s 4,virtio-net,$tap
>> -s 31,lpc -l com1,stdio $machine
>>
>> Thanks in advance.
>> _______________________________________________
>> freebsd-virtualization at freebsd.org mailing list
>> https://lists.freebsd.org/mailman/listinfo/freebsd-virtualization
>> To unsubscribe, send any mail to "freebsd-virtualization-unsubscribe at freebsd.org"

-- 
*Ing. Ilario Febi

Schema 31 S.p.a. Socio Unico*
*System Administrator

*BO - FI - ROMA - PA
ITALY
Tel: +39. 06.98.358.472
Cell: +39. 366.69.45.384
Fax: +39. 055.71.880.466
Società del Gruppo *SC31 ITALIA*


More information about the freebsd-virtualization mailing list