PRPnet server bugs and barfing problem - Page 10

rogue · 2009-08-08, 18:23

There is nothing weird in the code where it is hanging. The debug message is written before the send(). It is possible that the send() hung, but I really don't know. AFAICT, something happened in the network communication that I'm not capturing. If the send() fails then the server would output a message.

As for the `À|ë¨ in the message on the client side, that is my bad. I'm not passing an appropriate parameter to the output string.

MyDogBuster · 2009-08-09, 13:52

Okay guys, I don't know what this problem falls under. I took a power hit last night from an electrical storm. When the power came back on an hour later, 3/4th's of my candidates file was gone. A big hole in the center. I rebuilt it and started things up again. 2 hours later I took another power hit and was down for 2 hours. When I came back up 3/4 of my candidates file was missing again, this time the whole backend was gone. This particular candidates file is 4.2MB large, 150K candidates.

What is strange is that I have another PRPNet server running a different test on the same machine. The second candidates file was not affected either time. The second candidates file is only 353KB, 13K candidates.

Is reading the candidates file into memory somehow causing part of the file to disappear when something happens to memory? There were no abnormal messages anywhere either time.

rogue · 2009-08-09, 16:04

Quote:

Originally Posted by MyDogBuster

Okay guys, I don't know what this problem falls under. I took a power hit last night from an electrical storm. When the power came back on an hour later, 3/4th's of my candidates file was gone. A big hole in the center. I rebuilt it and started things up again. 2 hours later I took another power hit and was down for 2 hours. When I came back up 3/4 of my candidates file was missing again, this time the whole backend was gone. This particular candidates file is 4.2MB large, 150K candidates.

What is strange is that I have another PRPNet server running a different test on the same machine. The second candidates file was not affected either time. The second candidates file is only 353KB, 13K candidates.

Is reading the candidates file into memory somehow causing part of the file to disappear when something happens to memory? There were no abnormal messages anywhere either time.

You were missing stuff in the middle of the file? It sounds like a bug, yet you didn't see any messages and the candidates weren't written anywhere else. I suspect a memory leak or maybe you are using too much memory. How much RAM is being used by the larger one? Is memory being cached to disk?

MyDogBuster · 2009-08-09, 17:54

Quote:

You were missing stuff in the middle of the file? It sounds like a bug, yet you didn't see any messages and the candidates weren't written anywhere else. I suspect a memory leak or maybe you are using too much memory. How much RAM is being used by the larger one? Is memory being cached to disk?

Ram used = 40548K not cached to disk

It looked like it was 3MB (3000K) of memory missing both times. The machine has 3GB of memory. The candidates were not written anywhere else.

I understand that I would have lost any changes still in memory since the last update because of the power blip, but I didn't expect 3/4 of the file to go missing. I actually noticed it only because I went into the directory to see if I have any new primes, when I noticed the candidates file went from 4500K to 1500K in size.

Should be an easy test to reconstruct. Build a large candidates file, start the server and a client or 2 and pull the plug.

rogue · 2009-08-09, 18:50

Quote:

Originally Posted by MyDogBuster

Ram used = 40548K not cached to disk

It looked like it was 3MB (3000K) of memory missing both times. The machine has 3GB of memory. The candidates were not written anywhere else.

I understand that I would have lost any changes still in memory since the last update because of the power blip, but I didn't expect 3/4 of the file to go missing. I actually noticed it only because I went into the directory to see if I have any new primes, when I noticed the candidates file went from 4500K to 1500K in size.

Should be an easy test to reconstruct. Build a large candidates file, start the server and a client or 2 and pull the plug.

The candidates file is only open when reading (at startup) and writing (when it saves). It seems very unusual that power would have gone out twice in the middle of saving, which should only take a few seconds.

henryzz · 2009-08-09, 19:03

Quote:

Originally Posted by rogue

The candidates file is only open when reading (at startup) and writing (when it saves). It seems very unusual that power would have gone out twice in the middle of saving, which should only take a few seconds.

i seem to remember that MyDogBuster had his server writing after every single change or something ridiculous like that to help test the server in all situations

MyDogBuster · 2009-08-09, 20:07

Quote:

i seem to remember that MyDogBuster had his server writing after every single change or something ridiculous like that to help test the server in all situations

It was at 0 (instant update), but I changed it to 10 minutes before all this started. Makes no sense to me either. I'm watching that file like a hawk now.

MyDogBuster · 2009-08-09, 22:50

Well, good news. I've gone thru two more power blips and didn't lose anything. Getting a little tired with the power company switching grids, buts its 90 degrees with 100% humidity here. Air conditioner heaven.

mdettweiler · 2009-08-11, 00:51

I've loaded 140K-150K into G3000 again. Since no major bugs seem to still be open (excepting the weird problems Ian had, which don't seem to have occurred anywhere else), if this run goes well then we should be OK to proceed with the next range.

mdettweiler · 2009-08-11, 14:54

All servers have been upgraded to PRPnet 2.2.5. Meanwhile, though, I had another strange freeze on the server (this was while it was still running 2.2.4):

Code:

[2009-08-11 13:36:34 GMT] Message coming on socket 5
[2009-08-11 13:36:34 GMT] socket 5 <<<< FROM sm5ymt@pekhult.se _6 sm5ymt
[2009-08-11 13:36:34 GMT] sm5ymt@pekhult.se connecting from 91.149.43.243
[2009-08-11 13:36:34 GMT] socket 5 <<<< GETGREETING
[2009-08-11 13:36:34 GMT] socket 5 >>>> ############
[2009-08-11 13:36:34 GMT] socket 5 >>>> Welcome to the CRUS G3000 PRPnet beta test server! :-D
[2009-08-11 13:36:34 GMT] socket 5 >>>> Server is running PRPnet v2.2.3
[2009-08-11 13:36:34 GMT] socket 5 >>>> ############
[2009-08-11 13:36:34 GMT] socket 5 >>>> OK.
[2009-08-11 13:36:34 GMT] socket 5 <<<< QUIT
[2009-08-11 13:36:34 GMT] closing socket 5
[2009-08-11 13:36:34 GMT] Message coming on socket 5
[2009-08-11 13:36:34 GMT] socket 5 <<<< FROM sm5ymt@pekhult.se _6 sm5ymt
[2009-08-11 13:36:34 GMT] sm5ymt@pekhult.se connecting from 91.149.43.243
[2009-08-11 13:36:34 GMT] socket 5 <<<< RETURNWORK 2.2.4
[2009-08-11 13:36:34 GMT] socket 5 <<<< WorkUnit: 51255*6^144850+1 1249996661
[2009-08-11 13:36:34 GMT] socket 5 >>>> ERROR: Workunit 51255*6^144850+1 not found on server
[2009-08-11 13:36:34 GMT] socket 5 <<<< End of Message
[2009-08-11 13:36:34 GMT] socket 5 >>>> INFO: 0 of 1 test results were accepted
[2009-08-11 13:36:34 GMT] socket 5 >>>> End of Message
[2009-08-11 13:36:34 GMT] socket 5 <<<< QUIT
[2009-08-11 13:36:34 GMT] closing socket 5
[2009-08-11 13:36:34 GMT] Message coming on socket 5
[2009-08-11 13:36:34 GMT] socket 5 <<<< FROM sm5ymt@pekhult.se _6 sm5ymt
[2009-08-11 13:36:34 GMT] sm5ymt@pekhult.se connecting from 91.149.43.243
[2009-08-11 13:36:34 GMT] socket 5 <<<< GETWORK 2.2.4 1
[2009-08-11 13:36:34 GMT] socket 5 <<<< llr
[2009-08-11 13:36:34 GMT] socket 5 <<<< phrot
[2009-08-11 13:36:34 GMT] socket 5 <<<< pfgw
[2009-08-11 13:36:34 GMT] socket 5 <<<< End of Message
[2009-08-11 13:36:34 GMT] socket 5 >>>> ServerVersion: 2.2.4
[2009-08-11 13:36:34 GMT] First check candidate 22, 166753*6^144869+1
[2009-08-11 13:36:34 GMT] socket 5 >>>> ServerType: 1
[2009-08-11 13:36:34 GMT] socket 5 >>>> WorkUnit: 166753*6^144869+1 1249997794 166753 6 144869 1
[2009-08-11 13:36:34 GMT] sm5ymt@pekhult.se (_6) at 91.149.43.243: Sent 166753*6^144869+1
[2009-08-11 13:36:34 GMT] socket 5 >>>> End of Message
[2009-08-11 13:36:34 GMT] socket 5 <<<< QUIT
[2009-08-11 13:36:34 GMT] closing socket 5
[2009-08-11 13:36:34 GMT] Message coming on socket 5
[2009-08-11 13:36:34 GMT] socket 5 <<<< FROM sm5ymt@pekhult.se _6 sm5ymt
[2009-08-11 13:36:34 GMT] sm5ymt@pekhult.se connecting from 91.149.43.243
[2009-08-11 13:36:34 GMT] socket 5 <<<< RETURNWORK 2.2.4
[2009-08-11 13:36:34 GMT] socket 5 <<<< WorkUnit: 51255*6^144850+1 1249996661
[2009-08-11 13:36:34 GMT] socket 5 >>>> ERROR: Workunit 51255*6^144850+1 not found on server
[2009-08-11 13:36:34 GMT] socket 5 <<<< End of Message
[2009-08-11 13:36:34 GMT] socket 5 >>>> INFO: 0 of 1 test results were accepted
[2009-08-11 13:36:34 GMT] socket 5 >>>> End of Message
[2009-08-11 13:36:34 GMT] socket 5 <<<< QUIT
[2009-08-11 13:36:34 GMT] closing socket 5
[2009-08-11 13:36:34 GMT] Message coming on socket 5
[2009-08-11 13:36:34 GMT] socket 5 <<<< FROM sm5ymt@pekhult.se _6 sm5ymt
[2009-08-11 13:36:34 GMT] sm5ymt@pekhult.se connecting from 91.149.43.243
[2009-08-11 13:36:34 GMT] socket 5 <<<< GETGREETING
[2009-08-11 13:36:34 GMT] socket 5 >>>> ############
[2009-08-11 13:36:34 GMT] socket 5 >>>> Welcome to the CRUS G3000 PRPnet beta test server! :-D
[2009-08-11 13:36:34 GMT] socket 5 >>>> Server is running PRPnet v2.2.3
[2009-08-11 13:36:34 GMT] socket 5 >>>> ############
[2009-08-11 13:36:34 GMT] socket 5 >>>> OK.
[2009-08-11 13:36:34 GMT] socket 5 <<<< QUIT
[2009-08-11 13:36:34 GMT] closing socket 5
[2009-08-11 13:36:39 GMT] Message coming on socket 5
[2009-08-11 13:36:39 GMT] socket 5 <<<< FROM sm5ymt@pekhult.se _6 sm5ymt
[2009-08-11 13:36:39 GMT] sm5ymt@pekhult.se connecting from 91.149.43.243
[2009-08-11 13:36:51 GMT] socket 5 <<<< GETGREETING
[2009-08-11 13:36:51 GMT] socket 5 >>>> ############
[2009-08-11 13:36:51 GMT] socket 5 >>>> Welcome to the CRUS G3000 PRPnet beta test server! :-D
[2009-08-11 13:36:51 GMT] socket 5 >>>> Server is running PRPnet v2.2.3
[2009-08-11 13:36:51 GMT] socket 5 >>>> ############
[2009-08-11 13:36:51 GMT] socket 5 >>>> OK.
[2009-08-11 13:36:51 GMT] socket 5 <<<< QUIT
[2009-08-11 13:36:51 GMT] closing socket 5
[2009-08-11 13:36:51 GMT] Message coming on socket 5
[2009-08-11 13:36:51 GMT] socket 5 <<<< FROM sm5ymt@pekhult.se _6 sm5ymt
[2009-08-11 13:36:51 GMT] sm5ymt@pekhult.se connecting from 91.149.43.243
[2009-08-11 13:36:51 GMT] socket 5 <<<< GETGREETING
[2009-08-11 13:36:51 GMT] socket 5 >>>> ############
[2009-08-11 13:36:51 GMT] socket 5 >>>> Welcome to the CRUS G3000 PRPnet beta test server! :-D
[2009-08-11 13:36:51 GMT] socket 5 >>>> Server is running PRPnet v2.2.3
[2009-08-11 13:36:51 GMT] socket 5 >>>> ############
[2009-08-11 13:36:51 GMT] socket 5 >>>> OK.
[2009-08-11 13:36:51 GMT] socket 5 <<<< QUIT
[2009-08-11 13:36:51 GMT] closing socket 5
[2009-08-11 13:36:51 GMT] Message coming on socket 5
[2009-08-11 13:36:51 GMT] socket 5 <<<< FROM gbarnes017@gmail.com humpford gd_barnes
[2009-08-11 13:36:51 GMT] gbarnes017@gmail.com connecting from 76.92.253.11
[2009-08-11 13:36:51 GMT] socket 5 <<<< GETGREETING
[2009-08-11 13:36:51 GMT] socket 5 >>>> ############
[2009-08-11 13:36:51 GMT] socket 5 >>>> Welcome to the CRUS G3000 PRPnet beta test server! :-D
[2009-08-11 13:36:51 GMT] socket 5 >>>> Server is running PRPnet v2.2.3
[2009-08-11 13:36:51 GMT] socket 5 >>>> ############
[2009-08-11 13:36:51 GMT] socket 5 >>>> OK.
[2009-08-11 13:36:51 GMT] socket 5 <<<< QUIT
[2009-08-11 13:36:51 GMT] closing socket 5
[2009-08-11 13:36:51 GMT] Message coming on socket 5
[2009-08-11 13:36:51 GMT] socket 5 <<<< FROM sm5ymt@pekhult.se _6 sm5ymt
[2009-08-11 13:36:51 GMT] sm5ymt@pekhult.se connecting from 91.149.43.243
[2009-08-11 13:36:51 GMT] socket 5 <<<< RETURNWORK 2.2.4
[2009-08-11 13:36:51 GMT] socket 5 <<<< WorkUnit: 74612*6^144857+1 1249996678
[2009-08-11 13:36:51 GMT] socket 5 >>>> ERROR: Workunit 74612*6^144857+1 not found on server
[2009-08-11 13:36:52 GMT] socket 5 <<<< End of Message
[2009-08-11 13:36:52 GMT] socket 5 >>>> INFO: 0 of 1 test results were accepted
[2009-08-11 13:36:52 GMT] socket 5 >>>> End of Message
[2009-08-11 13:36:53 GMT] socket 5 <<<< QUIT
[2009-08-11 13:36:53 GMT] closing socket 5
[2009-08-11 13:36:53 GMT] Message coming on socket 5
[2009-08-11 14:41:14 GMT] Accepted force quit.  Waiting to close sockets before exiting
[2009-08-11 14:42:24 GMT] Accepted force quit.  Waiting to close sockets before exiting
[2009-08-11 14:42:54 GMT] Accepted force quit.  Waiting to close sockets before exiting
(at this point I had to kill the server)

Oddly enough, on numerous occasions in the log excerpt above, it seems that various clients connected to the server, grabbed the greeting, and exited. Mark, do you know why any clients would want to do that?

Also, there are a lot of candidates being returned that aren't found on the server; I'm not sure if this is something odd, or if there just happened to be a number of expired tests right around the time when I checked the server.

At any rate, though, at 13:36:53 GMT it received a connection from a client, then promptly froze. It remained that way until over an hour later when I had to -SIGKILL the server.

It's now restarted with v2.2.5.

P.S.: I just thought of something. There weren't any changes to the server in 2.2.5, right? If so, then it seems I just wasted my time changing the servers over.

MyDogBuster · 2009-08-11, 16:48

Max, you might want to check to see if some of the candidates file is missing right around where Lennart's returned tests failed. Also look in the completed work file. They have to be somewhere. I think you'll find the server was looping because some of the file is gone.

2009-08-09, 13:52	#101
MyDogBuster May 2008 Wilmington, DE 2852₁₀ Posts	Okay guys, I don't know what this problem falls under. I took a power hit last night from an electrical storm. When the power came back on an hour later, 3/4th's of my candidates file was gone. A big hole in the center. I rebuilt it and started things up again. 2 hours later I took another power hit and was down for 2 hours. When I came back up 3/4 of my candidates file was missing again, this time the whole backend was gone. This particular candidates file is 4.2MB large, 150K candidates. What is strange is that I have another PRPNet server running a different test on the same machine. The second candidates file was not affected either time. The second candidates file is only 353KB, 13K candidates. Is reading the candidates file into memory somehow causing part of the file to disappear when something happens to memory? There were no abnormal messages anywhere either time. Last fiddled with by MyDogBuster on 2009-08-09 at 13:53

2009-08-11, 16:48	#110
MyDogBuster May 2008 Wilmington, DE 2²·23·31 Posts	Max, you might want to check to see if some of the candidates file is missing right around where Lennart's returned tests failed. Also look in the completed work file. They have to be somewhere. I think you'll find the server was looping because some of the file is gone. Last fiddled with by MyDogBuster on 2009-08-11 at 17:05

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
PRPNet server for personal use	johnadam74	Software	2	2016-01-01 15:58
New SR5 PRPnet server online	ltd	Sierpinski/Riesel Base 5	15	2013-03-19 18:03
First PSP PRPnet 4.0.6 server online	ltd	Prime Sierpinski Project	9	2011-03-15 04:58
PRPnet 3.1.3 stress-test server	mdettweiler	No Prime Left Behind	40	2010-01-30 18:05
First pass PRPNet server out of work?	opyrt	Prime Sierpinski Project	6	2009-09-24 18:14

2009-08-08, 18:23	#100
rogue "Mark" Apr 2003 Between here and the 18D0₁₆ Posts	There is nothing weird in the code where it is hanging. The debug message is written before the send(). It is possible that the send() hung, but I really don't know. AFAICT, something happened in the network communication that I'm not capturing. If the send() fails then the server would output a message. As for the `À\|ë¨ in the message on the client side, that is my bad. I'm not passing an appropriate parameter to the output string.

2009-08-09, 22:50	#107
MyDogBuster May 2008 Wilmington, DE 101100100100₂ Posts	Well, good news. I've gone thru two more power blips and didn't lose anything. Getting a little tired with the power company switching grids, buts its 90 degrees with 100% humidity here. Air conditioner heaven.

2009-08-11, 00:51	#108
mdettweiler A Sunny Moo Aug 2007 USA (GMT-5) 3×2,083 Posts	I've loaded 140K-150K into G3000 again. Since no major bugs seem to still be open (excepting the weird problems Ian had, which don't seem to have occurred anywhere else), if this run goes well then we should be OK to proceed with the next range.