|
|
Up |
|
|
  |
Author: John SorosJohn Soros Date: Aug 25, 2008 23:49
Ok, still, it would be great to know how to set the time, as my time is way off (by more than 4 hours).
Best++
John
On Mon, 25 Aug 2008 15:44:39 -0700
"ron minnich" gmail.com> wrote:http://www.unixtimestamp.com/
> I just realized that even one timesync is too much. You should not
> run any at all. The hardware clock is set from Linux and I don't even
> allow it to be set. It makes no sense to do that.
>
> So don't let timesync run.
>
> thanks
>
> ron
>
|
| |
|
| | 8 Comments |
|
  |
Author: Alex LeeAlex Lee Date: Aug 27, 2008 19:34
On Tue, Aug 26, 2008 at 1:44 AM, John Soros gmail.com> wrote:
> Ok, still, it would be great to know how to set the time, as my time is way off (by more than 4 hours).
> Best++
> John
Hi again,
A couple of updates: I'm getting the same cron messages and time
problems that John is seeing. I've made sure that timesync isn't
running. My lguest time is about 3.5 hours ahead of host time.
As for the load issue: The full cpu usage that I was seeing earlier
actually stops after a several minutes.
After booting, ^T^Tp shows genrandom as Running. Several minutes later
the cpu usage drops to near idle, and genrandom goes to Wakeme. Then
every now and then (I'm not sure if anything in particular triggers
it, or if just randomly happens) I see the cpu usage jump again, and I
can see that genrandom is once again Running. Then after several
minutes it stops again.
When genrandom is running, stats shows a load of around 3000.
Otherwise, load is always around 2000.
|
| Show full article (1.23Kb) |
|
| | 6 Comments |
|
  |
Author: ron minnichron minnich Date: Aug 27, 2008 19:41
On Wed, Aug 27, 2008 at 7:28 PM, Alex Lee uchicago.edu> wrote:
> The problem looks similar to this:
> http://9fans.net/archive/2006/03/588 -- except that the lguest
> instance works fine while genrandom is running. Is this genrandom
> behavior anything out of the ordinary?
>
I don't think so. Genrandom make me want to look at using the hardware
RNGs a few years ago, but then Intel killed firmware hub and the whole
idea kind of went away.
ron
|
| |
| 1 Comment |
|
  |
Author: ron minnichron minnich Date: Aug 27, 2008 19:39
On Mon, Aug 25, 2008 at 11:44 PM, John Soros gmail.com> wrote:
> Ok, still, it would be great to know how to set the time, as my time is way off (by more than 4 hours).
What's the time on your host (Linux) look like? Are you sure it's not
a time zone setup issue?
Is it always four hours? It should not be drifting.
ron
|
| |
| no comments |
|
  |
Author: John SorosJohn Soros Date: Aug 28, 2008 01:26
Hi again, 9fans
On Wed, 27 Aug 2008 21:28:42 -0500
"Alex Lee" uchicago.edu> wrote:
> On Tue, Aug 26, 2008 at 1:44 AM, John Soros gmail.com> wrote:
>> Ok, still, it would be great to know how to set the time, as my time is way off (by more than 4
>> hours). Best++
>> John
...
|
| Show full article (1.80Kb) |
| 1 Comment |
|
  |
Author: Richard MillerRichard Miller Date: Aug 28, 2008 01:30
> As for the load issue: The full cpu usage that I was seeing earlier
> actually stops after a several minutes.
> ...
> Is this genrandom
> behavior anything out of the ordinary?
/dev/random normally runs at 100%% cpu after booting, and whenever
its queue of random bytes is getting empty. Since it uses the
clock to produce entropy, this might take longer than usual if
your clock is behaving strangely.
|
| |
| no comments |
|
  |
Author: erik quanstromerik quanstrom Date: Aug 28, 2008 05:00
>
> When genrandom is running, stats shows a load of around 3000.
> Otherwise, load is always around 2000.
>
what are the other two processes that are constantly
running?
- erik
|
| |
| no comments |
|
  |
Author: erik quanstromerik quanstrom Date: Aug 28, 2008 05:12
>
> I don't think so. Genrandom make me want to look at using the hardware
> RNGs a few years ago, but then Intel killed firmware hub and the whole
> idea kind of went away.
>
via has release the padlock documentation.
unfortunately one needs to enable floating point
and sse to use the padlock rng.
on reading that, i decided that software rng
wasn't as slow or as important as i'd remembered.
there's also the problem of ensuring that the
hardware rng is at least as good as plan 9's.
- erik
|
| |
| no comments |
|
  |
|
|
  |
Author: ron minnichron minnich Date: Aug 28, 2008 08:46
this is weird. So, to recap, your timezone is set correctly, and yet
you are four hours off.
A useful thing to do is cat /dev/time and see how it changes.
The time from lguest is simple: you read a 64-bit # which is time.
It's just like Xen that way.
Also, try this to test another issue:
date && sleep 60 && date
Two things: should take 60 seconds by the watch and the two dates
should report 60 seconds apart.
Do they?
ron
|
| |
| no comments |
|
|