Since Windows by default has NNTP (a time protocol that allows your clock to be adjusted automatically to the server it talks to), your clock can change forward or backward depending on the interval of this.

A little correction: it's NTP :)

NTP: Network Time Protocol
NNTP: Network News Transfer Protocol

Yet another addition to this: Windows (at least XP and older) can behave jerky like this because it supports only SNTP (Simple Network Time Protocol, a simplified part of NTP, allows only setting the computer local time so some (precise in the moment) value) out of the box.

To overcome the problem of jumping some (milli)seconds (or even minutes) here-and-there, I'd reccomend installing a full NTP driver, which is capable of setting the speed of the computer's local time to pretty precise value and keeping it. (Most unices and Linux has this out-of-the-box for years already. I wonder what does Vista say to this?)

Peter

Since Windows by default has NNTP (a time protocol that allows your clock to be adjusted automatically to the server it talks to), your clock can change forward or backward depending on the interval of this.

A little correction: it's NTP :)

NTP: Network Time Protocol
NNTP: Network News Transfer Protocol

a little more detailed correction: as Peter says windoze uses simple NTP - where you grab the time every day or so to correct the local clock.

In contrast, NTP run with a full NTP client keeps the clock within 1/8th sec at all times and adjusts within that zone by running the local clock a few parts per million fast or slow from time to time - called 'slewing' the clock rather than stepping it. If you are really running NTP you don't see clock steps in normal operation - tho if it slews during benchmarks you might see an increase/decrease in your benchmarks if you look at them on a parts per million resolution ;-)

There is a free NTP binary for windows (no Vista as yet). Download here from a German firm that sell clocks that you can use to set up a time standard. You don't have to buy a clock to download the software, very proper because the ntp source is GPL'd. I think it is very decent of the gmbh to have a special download page in English... that goes beyond the requirements of the GPL.

The source is on http://www.ntp.org/ and the brave can compile it locally.

Linux distros will have it as a package -- look for either ntp or ntpd. On debian for example, package ntp then asks you to choose to install either ntp-simple or ntp-refclock.

Having said all that, time stepping is not the sole cause of this error - I have ntp on my Linuces and still see this error. My impression is that it happens when the cpu is heavily loaded and the client doesn't get to run at all for a few seconds. Try calculating pi to a few million places and bye bye application... Suspend all projects beforehand, and the application survives.

River~~

Client_state.xml is being written to every multiple times a second regardless of what your write to disk is set to.

Don't worry about disk writes though. If your disk can't take many disk writes, it's not a good disk. And it won't reach the amount of writes a database disk gets a second.

But not so if it is a flash disk- they are better than they used to be but each bit on a flash disk has only a finite number of writes.

boinc really should honour the setting, even for the state file, IMO. Hopefully (for the disk) the disk will buffer the writes, but that is bad too as it increases the odds of the data being inconsistent with the project files if the machine crashes...

R~~

Just now I'm dealing with a BOINC Mgr on a computer with broken internet connection. I think it hangs on each attempt to connect to a project, but I can't even make it stop trying to connect, because I can't select "Disable network activity". As soon as it starts responding again, it does another RPC and hangs again!

Also, the CPU WAS IDLE, probably the apps kept exiting because of the hanged core client waiting for DNS. Who knows for how many hours it has been like that.

This situation was bugging me often, for months already. Few days ago I've found Boinc on my computer (which was running without internet connection) in bad loop, starting apps and finding them exited few moments later, doing this again and again a day and half long. BoincManager was absolutely hanged, Boinc was unresponsive to all means of communication (BoincManager, BoincView, boinc_cmd), neither restarting Boinc (+Mgr) nor trying to reboot multiple times helped.

Last few months my Boinc logs very often contained entries for two hosts in form "Can't resolve hostname [myhost.mydo.main] host not found" and "Can't resolve hostname [myhost.mydo.main] valid name, no data record of requested type". I was blaming my computer's DNS settings (company laptop traveling a lot out of it's home domain), but finally tried to find and remove exact these myhost.mydo.main entries (for the mentioned 2 hosts) from the remote_hosts.cfg file (with intention of replacing them with exact IPs as often as necessary) - the bad restart loop was suddenly cured.

Now I'm pretty sure Boinc has hard problems with occasional DNS (and/or internet connectivity) issues.

Peter