could you arrange the test listing so that it is organized from lowest N to highest? perhaps we could have a bold print for tests where factors have been found?
Please go to this url and see if you or someone you know has the IP of the old server hardcoded into their configuration.
Eventually, those ips will no longer be piping the data over to the new server and those clients will drop off. I'd like to give people time to fix their configurations (sb.pns.net or sbp.pns.net) so they don't lose work.
Also take a look at this url to see tests that are over 90 days old and still not completed. Altho some of them have been checked once already, this does show that these users and some of their computers are very slow (possibly due to a virus, or other DC application).
Also added this url to show people who have greater than 20 tests pending, which should make it easier to track down a runaway client.
--
Mike Garrison
Last edited by Alien88; 05-26-2004 at 12:27 AM.
could you arrange the test listing so that it is organized from lowest N to highest? perhaps we could have a bold print for tests where factors have been found?
I figured it would be easier for people to go through the list and find their username, rather than lowest N to highest. As for bold print for tests where factors have been found, that's a little too complicated for me to write right now. The page is basically to just give people a clue that something is wrong with one of their machines.Originally posted by Keroberts1
could you arrange the test listing so that it is organized from lowest N to highest? perhaps we could have a bold print for tests where factors have been found?
Maybe it's possible to make the table sortable by columns as compromise?
One other thing:
How is e.g.
possible? Shouldn't the test already have expired long ago?Code:lucasdavenport 67607*2^3653091+1 May 28 2003 Apr 29 02:55 0 %
As long as you send partial blocks back, you can have a test quite a long time.
Right. But there have to be a lot of blocks reported as the test was handed out 11 months ago. Approx. 7 months since then, the expiration time was 10 days, lateron 30 days.
Thus, there should be at least 25 blocks. Even when rounding off, this means there have to be 2501 blocks to be at 0%...
Unless something "strange" occured - like the reassigned test thingy in the 'deduction' thread or the user started over several times...
What happens when the user clears the cache of the client and restarts testing the same k/n pair using registry?
test assigned: long ago, progress: low ???
I do not say that is the case, but I just wondered what would have happened on the server side.
Yeah, i'm really not sure what the deal is with some of the really old tests.
as for what you said nuri.. i'm not quite sure. maybe dave can answer it.
Maybe your client is making progress reports, but isn't actually making any progress. In other words maybe it starts up, reads a cache file, starts working long enough to make a progress report saying progress = X, but then crashes before writing a new cache file... then it gets restarted by, say, a cron job...
The server should probably verify that you're actually making progress when deciding what to expire =)
One dual ppro, still working on tests from October 2003... 23% and 24% respectively
I think I'll pop them on a faster computer to finish them off. I can't stand waiting for another year to find out if one is prime or not.
the one from me is just the 600mhz celeron-o-fury compaq should be done in a week or 2 i think
Is there anyway to tell where the system with the hardcoded client resides physically? I'm about at a loss for locating the one system under my name that is hardcoded.
I think kugano can provide you with the IP address(es) with which the client contacted the server.
P.S.: You should enter the signature as an image to be shown as one...
I think machines like that would be better doing sieving or double-chekcing. Even factoring would be more efficent on them, if you have sufficient RAM.Originally posted by allio
One dual ppro, still working on tests from October 2003... 23% and 24% respectively
The biggest problem I see with PRP on very slow machines is if the k/n pair you're testing is prime and it takes, let's say, 12 months to complete, the project would be testing one "useless" k during all that time.
Maybe a second deadline, which would raise as n raises, would be solution, even if there are progess reports?
I wish I could... but when connecting through a data pipe, as is the case when a machine's using the old server's IP, the server sees the client as coming from the old SB IP. The original client's IP doesn't propagate up through the pipe to the new server. The only hope is that Mike's data pipe program (presumably datapipe.c?) keeps logs that show the original clients' IPs. Somehow I doubt it. I'll ask him when he gets back from Ann Arbor in a couple days.I think kugano can provide you with the IP address(es) with which the client contacted the server.
I can tell you, though, that it's a Windows XP machine running client version 1.2.5. If that helps at all.
Thanks for the info. Every little bit helps.
I just noticed that one of my machines (probably running linux, v1.2.5) was on the list of computers w/ old tests, has my test been completed by someone else, and if so is there any way that the computer be forced to get a new k-n, i would have thoought it would have happened since the computer was running mprime for 2 sets (well over the 30 day window) then i decided to return my attention back to SoB, it is very possible that the machine might have turned off yesterday since it hadn't report in today and the web servers on that machine are also down
How about restarting previous k/n when the client grabs another one. Been reading thru the forum on how to do it. Must have missed it. Thanks.
e
Last edited by engracio; 06-23-2004 at 07:26 AM.
With a linux-client just delete the cache-file and then it will get a new test.
But even if someone has done the test already if you are somewhat into the test, you might as well finish it as it will save a double test.
I am about 37% done but like i said i am not at home and as far as i know the computer has shutdown due to a power outage that lasted longer than my ups protects for
if you have a processor that takes an extremely long time to finish any particular test perhaps it would be best suited for sieving or factoring? Perhaps you could just leaqve it on double check tests. You still have a chance of finding a prime running the double check. So far with the eror rate we've encountored you may even have a better chance of finding a prime there.
it is an athlon xp 1900 underclocked to 1200mhz w/ 256m of ram
There is a weird one in the oldTests page. Something is not right, maybe it is just me.
v**** 27653•2^5449857+1 81.174.142.193 Jan 18 23:20 Jul 8 16:31 97 % 252
It has been running six months, but it is 97% complete with a cEMs/s of only 252. My PC's cEMs/s rate is 2000 times that and it takes around six days to process a kn pair. Even with the lower n it should not already be 97% complete.
Ok. I have 27 test pending in the 20 test or more post. What exactly does that mean? If it's a problem that needs to be taken care of maybe you could post in this thread since I don't get in here that often.
"I am responsible for everything except my very responsiblity"
www.amdrocks.com
Team AmdRocks The World!!!
Having a lot of tests running is no problem. You have no tests in the "Tests That Are Over 90 Days Old" page. Keep up the good work!
This page is just to timely see "runaway clients", which - for whatever reason - request new tests on a very short basis (often 5 - 20 minutes).Originally posted by SkullHead
Ok. I have 27 test pending in the 20 test or more post. What exactly does that mean?
When there are no problems, people can be proud when they are on that list.
That'd be me then!Originally posted by Mystwalker
When there are no problems, people can be proud when they are on that list.
"I am responsible for everything except my very responsiblity"
www.amdrocks.com
Team AmdRocks The World!!!