mersenneforum.org LLRnet servers for NPLB
 Register FAQ Search Today's Posts Mark Forums Read

 2009-08-17, 22:11 #1167 kar_bon     Mar 2006 Germany 22·23·31 Posts 'funny order' isn't funny at all: processing those ranges has to wait until all pairs are done!
 2009-08-17, 22:14 #1168 IronBits I ♥ BOINC!     Oct 2002 Glendale, AZ. (USA) 21318 Posts BAD server = "nplb.ironbits.net" GOOD server = "noprimeleftbehind.net" Use your own Project's Servers ... You need to ensure everyone is using the GOOD server, for now and future... Last fiddled with by IronBits on 2009-08-17 at 22:16
 2009-08-17, 22:30 #1169 kar_bon     Mar 2006 Germany 54448 Posts in the n-range 950k-960k there're 3 pairs missing in the results! 345 957466 333 957578 315 957595 if that was a 'funny order' why only those 3 pairs? where are they now? for future: sort the knpair-file before uploading to the LLRnet-server! and there seems another outrage of Gary's server again! Last fiddled with by kar_bon on 2009-08-17 at 22:30
2009-08-17, 22:40   #1170
mdettweiler
A Sunny Moo

Aug 2007
USA (GMT-5)

792 Posts

Quote:
 Originally Posted by kar_bon in the n-range 950k-960k there're 3 pairs missing in the results! 345 957466 333 957578 315 957595 if that was a 'funny order' why only those 3 pairs? where are they now? for future: sort the knpair-file before uploading to the LLRnet-server! and there seems another outrage of Gary's server again!
Okay, I think those three results are not the result of a "funny order", but instead somehow messed up due to the power outage. At any rate, there's not really any way we can tell what happend for now; when we process the results at the end, we can verify exactly which results are missing and re-test any that are missing.

Regarding another outage: I just checked and saw that port 7000 went offline. I restarted it. It seems to have crashed a few minutes after it started. That's not particularly abnormal; I've noticed that LLRnet servers tend to sometimes be a bit unstable soon after being restarted after events such as a power outage. At any rate, it's back on now and will probably be OK going forward.

 2009-08-17, 23:05 #1171 gd_barnes     May 2007 Kansas; USA 72×11×19 Posts Oh my goodness. This is just totally unbelievable happening all on one day that I thought I had everything caught up. First: Max, please don't speculate on what "might" be the problem before investigating. That only creates more problems! No, the pairs from MooMoooo's reservation were not loaded out of order. n=970K-1M was loaded right at the end of the server as it should be. I would never knowingly have David load pairs out of n-value order. Second, Max, as you know we need that "loop thing" running in order to keep the servers from crashing again right after they have gone down and have to be restarted. Can you make sure that port G7000 has that running? Since this is a new server, could you have forgotten to put that script in there? Third, Karsten, no I did not get any pairs from G7000 at all. It was only G8000. Any guess as to why some are being rejected would be just that...a guess. Fourth: Probably the reason I couldn't connect to David's machines for so long is that AT THAT TIME, I needed to change nplb.ironbits.net to noprimeleftbehind.net. That is most likely what the problem was. I am now able to connect to nplb.ironbits.net so that is resolved on my end. That said, I'll get to moving my machines over to noprimeleftbehind.net in a little while. And finally: Everyone, when you have time, please change the server on David's machines as follows: Old: nplb.ironbits.net New: noprimeleftbehind.net I will take care of changing that in several places as necessary throughout our drives and server listings. Karsten, I'll have a more detailed look at what is going on with the G7000 rejected pairs and the G8000 missing pairs. I'll post my findings when I have a clear answer. Gary Last fiddled with by gd_barnes on 2009-08-17 at 23:38
 2009-08-17, 23:27 #1172 gd_barnes     May 2007 Kansas; USA 72·11·19 Posts OK, I have the specifics on the port G7000 rejected pairs but cannot say why it happened. Apparently Karsten's machine returned the same 2 pairs twice. Why the server allowed this is anyone's guess. Likely the power outage caused them to be handed out twice in error. We do know that the 2 rejected results for that server are not missing as you'll see. Rejected results: Code: user=kar_bon [2009-08-17 16:21:51] 2011*2^117249-1 is not prime. Res64: 429A46B42C83A580 Time : 32.0 sec. user=kar_bon [2009-08-17 16:21:53] 2011*2^117255-1 is not prime. Res64: D4FD5733AF3AEB58 Time : 33.0 sec. "Real" results.txt: Code: user=kar_bon [2009-08-17 16:21:39] 2011*2^117249-1 is not prime. Res64: 429A46B42C83A580 Time : 20.0 sec. user=kar_bon [2009-08-17 16:21:46] 2011*2^117255-1 is not prime. Res64: D4FD5733AF3AEB58 Time : 26.0 sec. No worries there Karsten. Nothing missing. Apparently the server handed them out twice because of the power outage. I don't think this is fixable. The quick turnaround time of the small tests likely makes such an issue more likely. Looking into the missing pairs on port G8000 now... Gary Last fiddled with by gd_barnes on 2009-08-17 at 23:30
 2009-08-17, 23:44 #1173 gd_barnes     May 2007 Kansas; USA 72·11·19 Posts David, Can you tell me what web page address from the 1st post of this thread should go here: A daily general report for all of Ironbit's servers is at http://nplb.ironbits.net. I tried http://noprimeleftbehind.net but that just goes to the hourly status report just like the 2nd link in that 1st post does. We need the one that goes to the blue page that shows the min-n and max-n remaining in a server as well as # of pairs remaining. Thanks, Gary
 2009-08-17, 23:52 #1174 IronBits I ♥ BOINC!     Oct 2002 Glendale, AZ. (USA) 21318 Posts There are two, I suggest you use the 1st one. There is the noprimeleftbehind.net version here http://noprimeleftbehind.net/index.p...rver_status_ib There is the one I do with vbscript so I can keep an eye on the actual servers that run here right now... http://nplb.ironbits.net
 2009-08-18, 00:48 #1175 gd_barnes     May 2007 Kansas; USA 72·11·19 Posts I really like the vertical formatting on the 2nd one. I thought you might have a 'noprimeleftbehind' web page name version that I should change it to. I'll stick with that one for now unless you have any objection.
2009-08-18, 01:32   #1176
mdettweiler
A Sunny Moo

Aug 2007
USA (GMT-5)

141418 Posts

Quote:
 Originally Posted by gd_barnes OK, I have the specifics on the port G7000 rejected pairs but cannot say why it happened. Apparently Karsten's machine returned the same 2 pairs twice. Why the server allowed this is anyone's guess. Likely the power outage caused them to be handed out twice in error. We do know that the 2 rejected results for that server are not missing as you'll see. Rejected results: Code: user=kar_bon [2009-08-17 16:21:51] 2011*2^117249-1 is not prime. Res64: 429A46B42C83A580 Time : 32.0 sec. user=kar_bon [2009-08-17 16:21:53] 2011*2^117255-1 is not prime. Res64: D4FD5733AF3AEB58 Time : 33.0 sec. "Real" results.txt: Code: user=kar_bon [2009-08-17 16:21:39] 2011*2^117249-1 is not prime. Res64: 429A46B42C83A580 Time : 20.0 sec. user=kar_bon [2009-08-17 16:21:46] 2011*2^117255-1 is not prime. Res64: D4FD5733AF3AEB58 Time : 26.0 sec. No worries there Karsten. Nothing missing. Apparently the server handed them out twice because of the power outage. I don't think this is fixable. The quick turnaround time of the small tests likely makes such an issue more likely. Looking into the missing pairs on port G8000 now... Gary
Ah, as I thought. Yep, nothing to worry about there.
Quote:
 Originally Posted by gd_barnes I really like the vertical formatting on the 2nd one. I thought you might have a 'noprimeleftbehind' web page name version that I should change it to. I'll stick with that one for now unless you have any objection.
BTW, that one should have an analogue at:

http://llrnet.noprimeleftbehind.net/

Last I checked, that one redirects to http://nplb.ironbits.net/. As of this posting llrnet.noprimeleftbehind.net has not been "fixed" yet (and still goes to David's personal web page), but once everything is back up it should work.

 2009-08-18, 11:49 #1177 AMDave     Jan 2006 deep in a while-loop 2×7×47 Posts Stats database recovery should now be complete Watching the logs and the stats closely for a while.

 Similar Threads Thread Thread Starter Forum Replies Last Post mdettweiler No Prime Left Behind 228 2018-12-26 04:50 gd_barnes No Prime Left Behind 0 2009-08-10 19:21 gd_barnes Conjectures 'R Us 39 2008-07-15 10:26 em99010pepe No Prime Left Behind 229 2008-04-30 19:13 em99010pepe No Prime Left Behind 19 2008-03-26 06:19

All times are UTC. The time now is 09:17.

Wed Nov 25 09:17:21 UTC 2020 up 76 days, 6:28, 4 users, load averages: 1.50, 1.53, 1.42