mersenneforum.org > Data What did this poor client do?
 Register FAQ Search Today's Posts Mark Forums Read

 2015-05-11, 20:49 #1 tha     Dec 2002 3×271 Posts What did this poor client do? Hmm, something went wrong here: exponent 10670573
2015-05-11, 21:53   #2
Mark Rose

"/X\(‘-‘)/X\"
Jan 2013

3·977 Posts

Quote:
 Originally Posted by tha Hmm, something went wrong here: exponent 10670573
Whatever it did, it did it consistently.

2015-05-12, 03:04   #3
Serpentine Vermin Jar

Jul 2014

37×89 Posts

Quote:
 Originally Posted by Mark Rose Whatever it did, it did it consistently.
Without even looking at the raw results, my guess was the client reported the same result multiple times. I say that because only one entry shows up in the "official" LL results section. The history there has all the old v4 logs that we merged in recently...

I know on the current system if you try to check in the same result again, it says "hey, knock it off" (well, in so many words). The old system might have known it was a duplicate but still logged the message.

 2015-05-25, 06:31 #4 Dubslow Basketry That Evening!     "Bunslow the Bold" Jun 2011 40
2015-05-25, 15:45   #5
Serpentine Vermin Jar

Jul 2014

37·89 Posts

Quote:
 Originally Posted by Dubslow My most recently completed random DC happened to be one of these. http://www.mersenne.org/report_expon...4381829&full=1 Edit: Multi-reported the PM1 and NF results as well as the LL result.
I checked with George and James about things like this. It seems like in some cases, the old version 4 logs we merged in recently contain multiple sets of the same data. We're pretty sure this is caused by users who, for whatever reason, kept reporting the same data over and over for months, even years in some cases.

With the old version of Prime95, I remember it had that .spl file or whatever, and that would contain the data to be sent off to Primenet on it's next connection. My hypothesis is that something was weird on the user's system where it could append new data to that file but it couldn't be deleted, so it would just report the same thing each time it connected to the server.

Well, that's my working theory on why, in a worst case example, one entry shows up 130+ times or something like that. Most of the duplicates were cases where it showed up just twice, maybe 3 times, and again I like to think it was some problem deleting the file after the contents were sent. Or maybe the server gave a weird response so the client resent the result later.

Whatever happened, I have a plan to clear out those dupe entries so they won't be cluttering things up. I have a feeling it might put a little load on the server when that happens so I may just do it in stages.

At any rate, it doesn't affect more than just cluttering things in most cases. LL tests with the same residue and shift-count are already detected, and duplicate TF/P-1 runs with the same settings don't matter. ECM runs that show as duplicates could affect things but there are other things in there that help tell which ones are dupes and which are actually separate runs using the same bounds and # of curves.

Hope that helps explain where things are at.

2015-05-27, 03:22   #6
Serpentine Vermin Jar

Jul 2014

37·89 Posts

Quote:
 Originally Posted by tha Hmm, something went wrong here: exponent 10670573
George did some rockin' and rollin' today and he finessed the history logs to remove those duplicate entries.

If you look at it now, all of those repetitive entries are gone. Which of course means that people in the future who come across this thread (hi, future people) will wonder what we were talking about.

I think he was also working on getting user names to show up in the history for all the old stuff, not just "anonymous". I can't tell if that's the case here or not, but it looks good anyway.

2015-05-27, 03:40   #7
Dubslow

"Bunslow the Bold"
Jun 2011
40<A<43 -89<O<-88

1C3516 Posts

Quote:
 Originally Posted by Madpoo George did some rockin' and rollin' today and he finessed the history logs to remove those duplicate entries.
I looked at the exponent I linked out of random curiosity. Note the following:

Code:
2015-05-25	Dubslow	C	8E2B934ABD0AB8__
2014-02-10	kdgehman	NF	no factor from 2^70 to 2^71
2014-02-10	kdgehman	NF	no factor from 2^69 to 2^70
2013-07-25	Le Comte	NF	no factor from 2^68 to 2^69
2008-11-27	AlexN	NF-PM1	B1=365000, B2=4380000
2005-02-23	AlexN	NF	no factor to 2^68
2005-02-23	AlexN	NF-PM1	B1=365000, B2=4380000
2005-02-23	AlexN	C	8E2B934ABD0AB8__
2004-09-02	oerwin	NF	no factor to 2^62
In particular, note the dates of the remaining seemingly-duplicate results.

2015-05-27, 15:03   #8
Prime95
P90 years forever!

Aug 2002
Yeehaw, FL

165028 Posts

Quote:
 Originally Posted by Dubslow Code: 2008-11-27 AlexN NF-PM1 B1=365000, B2=4380000 2005-02-23 AlexN NF-PM1 B1=365000, B2=4380000 In particular, note the dates of the remaining seemingly-duplicate results.
Interesting, the 2008 entry was reported to the v5 server. The 2005 entry was from the recently incorporated v4 logs. The change I made removed the dups from the v4 logs.

I'm a bit surprised the v5 server didn't ignore the 2008 PM1 result as it was not needed.

2015-05-27, 18:18   #9
Serpentine Vermin Jar

Jul 2014

63358 Posts

Quote:
 Originally Posted by Prime95 Interesting, the 2008 entry was reported to the v5 server. The 2005 entry was from the recently incorporated v4 logs. The change I made removed the dups from the v4 logs. I'm a bit surprised the v5 server didn't ignore the 2008 PM1 result as it was not needed.
2008-11-27 wasn't too far along after v5 seemed to go live, full time. Perhaps that had something to do with it? Something during that transition period?

2015-05-27, 20:05   #10
Prime95
P90 years forever!

Aug 2002
Yeehaw, FL

2×5×7×107 Posts

Quote:
 Originally Posted by Madpoo 2008-11-27 wasn't too far along after v5 seemed to go live, full time. Perhaps that had something to do with it? Something during that transition period?
Quite possibly. We were (continuing development/fixing bugs) at a fast and furious rate.

Feel free to delete the v5 result log entry for that P-1 run.

2015-05-27, 21:19   #11
Dubslow

"Bunslow the Bold"
Jun 2011
40<A<43 -89<O<-88

722110 Posts

Quote:
 Originally Posted by Prime95 Quite possibly. We were (continuing development/fixing bugs) at a fast and furious rate. Feel free to delete the v5 result log entry for that P-1 run.
There's no evidence either to v5 issues or prime 95 spool file bugs; I would keep it for now.

What were the dates of the deleted entries? Both for this exponent and in general.

 Similar Threads Thread Thread Starter Forum Replies Last Post ahmerali PrimeNet 4 2008-11-11 01:46 mgpower0 Prime Sierpinski Project 54 2008-07-15 16:50 drakkar67 Prime Sierpinski Project 5 2008-07-10 12:37 Jeff Gilchrist NFSNET Discussion 7 2004-02-19 17:41 phalanx Software 4 2003-12-12 07:00

All times are UTC. The time now is 17:18.

Tue May 11 17:18:57 UTC 2021 up 33 days, 11:59, 1 user, load averages: 5.08, 4.72, 4.20