| Re: [chrony-dev] Chrony freezing |
[ Thread Index |
Date Index
| More chrony.tuxfamily.org/chrony-dev Archives
]
On Fri, 13 Nov 2009, Miroslav Lichvar wrote:
On Fri, Nov 13, 2009 at 08:19:27AM -0800, Bill Unruh wrote:
On Fri, 13 Nov 2009, Miroslav Lichvar wrote:
Where do you propose we should add it? If it is right before the
second read, it will solve the problem with blocking forever, but
chronyd will still be unresponsive for the second which I'm worried
about more as it degrades NTP performance.
This is only used for reading the rtc and for determining the rate and offset
of the rtc. Ie, it does not impact ntp performance, except insofar as it make
characterising rtc more difficult.
I meant performance for NTP clients, if the request hits chrony just
when it's in the blocking read, the reply will be extremely
asymmetric. This might be mitigated by the SO_TIMESTAMP support we've
recently added.
Good point.
But most systems ignore chrony's rtc
anyway, and start the system using hwclock to set the system time from the
rtc. And the newer hwclock's also have an ability to roughly determine the
rate of the rtc and offset and correct for those. Ie, the rtc performance of
chrony is to some extent becoming superfluous.
Yeah, that's why I don't use the feature, I'd have to hack the system
scripts. And most of my machines run 24/7 anyway.
What about making always two readings, i.e. don't disable the interrupt
after first one, wait for the interrupt again and use that? (see the patch)
Precisely what it does now. And that freezes the system sometimes.
It only needs to do this in "normal" operation.
Ok, please try the patch. It should never freeze in the read, it will
just stop tracking rtc if it stops generating the interrupts.
OK, will do.
Does the early interrupt happen always? If yes we could use a
heuristic, if the first interrupt is repeatedly 1 second +- 10% from
the second one, we can turn the hack off.
No idea anymore. Things have changed. The problem was there with older kernels.
In fact the rtc was completely whacky because of the switchover to hpet
support. I put in a kernel bug report on it which caused some work to be done
on the rtc driver. (can't find it right now).
--
William G. Unruh | Canadian Institute for| Tel: +1(604)822-3273
Physics&Astronomy | Advanced Research | Fax: +1(604)822-5324
UBC, Vancouver,BC | Program in Cosmology | unruh@xxxxxxxxxxxxxx
Canada V6T 1Z1 | and Gravity | www.theory.physics.ubc.ca/
---
To unsubscribe email chrony-dev-request@xxxxxxxxxxxxxxxxxxxx with "unsubscribe" in the subject.
For help email chrony-dev-request@xxxxxxxxxxxxxxxxxxxx with "help" in the subject.
Trouble? Email listmaster@xxxxxxxxxxxxxxxxxxxx.