Automated PRP using LLRNet - Page 6

ltd · 2008-06-07, 21:49

I also checked the client cause i had some memory that there is a problem
with the number of reserved WU.

I found in the original client config(not written by me) the following info about the WU chache size:

-- to ask several WU job to the server at a time, set the variable below
-- to something greater than 1 (but no more than 100)
WUCacheSize =

So please can you decrease the WUcachesize to something below 100 to see if there is really a problem.

ltd · 2008-06-07, 22:17

The restart did not help.

I will now remove most of the jobs that were send out to MyDogBuster to see if that helps.

ltd · 2008-06-07, 22:26

Deleting all the jobs from the queue did help at least for my client.
I also see that there is one new job for Konrad127123, Birdman2584 and 135 for MyDogBuster.

The number is still increasing so there seems to be something wrong with the client that lets run it wild.

Edit:
We are now at 270 reserved. It will take some time to clean up all the dropped tests. So there will be some gaps in our tests in the next week at least.

Edit2:
I have send a message to DoBuster to stop his client as it looks like its never stopping to request new WU. I hope that the message arrives before the server stops to hand out new work to others.

I will make the next cleanup in the morning. (It is nearly 1 in the morning here and i am a little bit tired.

ltd · 2008-06-08, 09:36

Update on the runaway client.

Reason for the problem was a script with an error that lead to the problem
that the llrnet client was started twice from the same directory.

This is solved now and I have cleaned up the DB as good as possible.

There might be some dead pairs that we need to release once more
later or a handfull of tests that will be send out twice(producing double checks)

vaughan · 2008-06-26, 07:51

Is the Sierpinski port open? Is the server handing out Sierpinski Base 5 LLR tasks? I keep getting this message:

Successfully connected to server!
recv error res=-1, errno=0
Could not log on to the server
LOGGING OUT

ltd · 2008-06-26, 08:25

As I am at work I can not try to fetch work but my internal scripts show
that the Sierpinski server has handed out a new pair to birdman at 10:03 CEST.
So I asume that the server is still working fine.

vaughan · 2008-06-27, 05:24

Lars I am still unable to connect.

ltd · 2008-06-27, 11:04

I have tested the server and it is handing out tests without problems.
Is it possible that you are behind a firewall and have the communication blocked for the sierpinski llrnet client. I can see that you have no problems with the riesel side so my only idea is that a firewall blocks the access from the sierpinski folder. Or on a second thought is it possible that you have some
restrictions to the accessrights for the sierpinski folder that blocks the client from writing to a file?

vaughan · 2008-06-27, 12:42

Yes you are right. That particular PC is behind AVG-Grisoft's Security Suite. I'll have to check how to allow access for the project. I have attached another PC to get Riesel tasks.

vaughan · 2008-07-02, 02:32

I set the cache to 5 a couple of days ago and so far have seen at least 2 cases where a message appears in the DOS box with words along the lines of:
The server refused your new result :
either someone else computed it already,
either the server is now configured to
work on other numbers.
The server refused your new result :
either someone else computed it already,
either the server is now configured to
work on other numbers.
Starting probable prime test of 164852*5^268140+1
Starting probable prime test of 164852*5^268140+1
Using generic reduction FFT length 64K

Using generic reduction FFT length 64K

and then it starts the new task.
I don't know why the messages get repeated either.

I'm not impressed seeing as there are only a handful of us crunching numbers on this project.

ltd · 2008-07-02, 05:13

Hi vaughan,

i checked the log and the number in your example is returne only by you on 2008-07-01 11:29:10.

One question about your setup on that machine. Do you run one or two instances of llr on that machine? If you have more instances of llr running on a machine each must use its own!!!!!! directory. This is the case also if both instances run for the same project!!!!!

2008-06-07, 22:26	#58
ltd Apr 2003 1404₈ Posts	Deleting all the jobs from the queue did help at least for my client. I also see that there is one new job for Konrad127123, Birdman2584 and 135 for MyDogBuster. The number is still increasing so there seems to be something wrong with the client that lets run it wild. Edit: We are now at 270 reserved. It will take some time to clean up all the dropped tests. So there will be some gaps in our tests in the next week at least. Edit2: I have send a message to DoBuster to stop his client as it looks like its never stopping to request new WU. I hope that the message arrives before the server stops to hand out new work to others. I will make the next cleanup in the morning. (It is nearly 1 in the morning here and i am a little bit tired. Last fiddled with by ltd on 2008-06-07 at 22:45

2008-07-02, 02:32	#65
vaughan Jan 2005 Sydney, Australia 517₈ Posts	Completed task not accepted I set the cache to 5 a couple of days ago and so far have seen at least 2 cases where a message appears in the DOS box with words along the lines of: The server refused your new result : either someone else computed it already, either the server is now configured to work on other numbers. The server refused your new result : either someone else computed it already, either the server is now configured to work on other numbers. Starting probable prime test of 1648525^268140+1 Starting probable prime test of 1648525^268140+1 Using generic reduction FFT length 64K Using generic reduction FFT length 64K and then it starts the new task. I don't know why the messages get repeated either. I'm not impressed seeing as there are only a handful of us crunching numbers on this project.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
LLRnet and PRPnet servers for automated LLR	mdettweiler	Twin Prime Search	235	2021-05-13 21:13
Automated LLR testing with LLRnet	mdettweiler	No Prime Left Behind	24	2011-11-04 19:20
Automated primality testing with LLRnet	mdettweiler	Conjectures 'R Us	18	2008-03-04 00:06
Automated PRP discussions	ltd	Sierpinski/Riesel Base 5	20	2006-09-02 22:19
Automated P-1 thoughts.	nucleon	Marin's Mersenne-aries	3	2004-03-25 02:45

2008-06-07, 21:49	#56
ltd Apr 2003 2²·193 Posts	I also checked the client cause i had some memory that there is a problem with the number of reserved WU. I found in the original client config(not written by me) the following info about the WU chache size: -- to ask several WU job to the server at a time, set the variable below -- to something greater than 1 (but no more than 100) WUCacheSize = So please can you decrease the WUcachesize to something below 100 to see if there is really a problem.

2008-06-07, 22:17	#57
ltd Apr 2003 772₁₀ Posts	The restart did not help. I will now remove most of the jobs that were send out to MyDogBuster to see if that helps.

2008-06-08, 09:36	#59
ltd Apr 2003 2²×193 Posts	Update on the runaway client. Reason for the problem was a script with an error that lead to the problem that the llrnet client was started twice from the same directory. This is solved now and I have cleaned up the DB as good as possible. There might be some dead pairs that we need to release once more later or a handfull of tests that will be send out twice(producing double checks)

2008-06-26, 07:51	#60
vaughan Jan 2005 Sydney, Australia 5×67 Posts	Is the Sierpinski port open? Is the server handing out Sierpinski Base 5 LLR tasks? I keep getting this message: Successfully connected to server! recv error res=-1, errno=0 Could not log on to the server LOGGING OUT

2008-06-26, 08:25	#61
ltd Apr 2003 772₁₀ Posts	As I am at work I can not try to fetch work but my internal scripts show that the Sierpinski server has handed out a new pair to birdman at 10:03 CEST. So I asume that the server is still working fine.

2008-06-27, 05:24	#62
vaughan Jan 2005 Sydney, Australia 5·67 Posts	Lars I am still unable to connect.

2008-06-27, 11:04	#63
ltd Apr 2003 2²×193 Posts	I have tested the server and it is handing out tests without problems. Is it possible that you are behind a firewall and have the communication blocked for the sierpinski llrnet client. I can see that you have no problems with the riesel side so my only idea is that a firewall blocks the access from the sierpinski folder. Or on a second thought is it possible that you have some restrictions to the accessrights for the sierpinski folder that blocks the client from writing to a file?

2008-06-27, 12:42	#64
vaughan Jan 2005 Sydney, Australia 5×67 Posts	Yes you are right. That particular PC is behind AVG-Grisoft's Security Suite. I'll have to check how to allow access for the project. I have attached another PC to get Riesel tasks.

2008-07-02, 05:13	#66
ltd Apr 2003 304₁₆ Posts	Hi vaughan, i checked the log and the number in your example is returne only by you on 2008-07-01 11:29:10. One question about your setup on that machine. Do you run one or two instances of llr on that machine? If you have more instances of llr running on a machine each must use its own!!!!!! directory. This is the case also if both instances run for the same project!!!!!