PRPNet 5.5 Released

[Happy5214]

A PRPNet server I set up to find first primes for various Riesel k's is still handing out candidates for k's after primes have already been found with servertype=1 (Sierpinski/Riesel). The tests are at the n=110k range right now, so the extra tests aren't a huge imposition right now, but a fix would be appreciated before I get too much deeper in this project.

[rogue]

Originally Posted by Happy5214

A PRPNet server I set up to find first primes for various Riesel k's is still handing out candidates for k's after primes have already been found with servertype=1 (Sierpinski/Riesel). The tests are at the n=110k range right now, so the extra tests aren't a huge imposition right now, but a fix would be appreciated before I get too much deeper in this project.

There are two possibilities. You have onekperinstance=0. This is a bug, which I think is fixed, but not in the current build available at source force. The prime for that k is for n < the prime found. I would guess the first. Change onekperinstance to 1. The server will still send out candidates for the same k to multiple clients, but will only do so if all k have at least one candidate set out to a client.

[Happy5214]

Originally Posted by rogue

There are two possibilities. You have onekperinstance=0. This is a bug, which I think is fixed, but not in the current build available at source force. The prime for that k is for n < the prime found. I would guess the first. Change onekperinstance to 1. The server will still send out candidates for the same k to multiple clients, but will only do so if all k have at least one candidate set out to a client.

Correct, it is the first. I only have one client running on it, so onekperinstance=1 (which I assume would run each k in sequence in my case) is unpalatable, though it's something I might have to consider once the terms grow bigger. I'm running from the latest SVN, so it hasn't hit there yet either.

[rogue]

Originally Posted by Happy5214

Correct, it is the first. I only have one client running on it, so onekperinstance=1 (which I assume would run each k in sequence in my case) is unpalatable, though it's something I might have to consider once the terms grow bigger. I'm running from the latest SVN, so it hasn't hit there yet either.

I committed a few code changes including what I think will fix the issue. I have not tested it though because I'm not running that kind of server right now. It is in PrimeWorkSender::SelectCandidates(). For Sierpinski/Riesel servers it will execute a different SQL statement.

[Happy5214]

Originally Posted by rogue

I committed a few code changes including what I think will fix the issue. I have not tested it though because I'm not running that kind of server right now. It is in PrimeWorkSender::SelectCandidates(). For Sierpinski/Riesel servers it will execute a different SQL statement.

The changes appear to have fixed the problem. Thanks!

[Trilo]

Hi, is there anyway to run multithreaded LLR instances using prpnet? I tried adding ThreadsPerTest=1 in llr.ini but it appears that prpnet rewrites the file upon bootup and the changes do not get saved.

[rogue]

Originally Posted by Trilo

Hi, is there anyway to run multithreaded LLR instances using prpnet? I tried adding ThreadsPerTest=1 in llr.ini but it appears that prpnet rewrites the file upon bootup and the changes do not get saved.

If there is a command line option then add via prpclient.ini

[lalera]

hi,
https://www.mersenneforum.org/showpo...&postcount=142

[rogue]

I have found and fixed a bug in the server that impacts Sierpinski/Rlesel project. This has a big impact servers with large numbers of sub minute PRP tests, but could affect any Sierpinski/Riesel server. The changes to source have been committed and the Windows exes have been updated over at sourceforge.

[Happy5214]

I'm getting a build failure on Linux:

In file included from /usr/include/fcntl.h:301,
                 from Socket.h:29,
                 from prpclient.cpp:22:
In function ‘int open(const char*, int, ...)’,
    inlined from ‘int main(int, char**)’ at prpclient.cpp:462:19:
/usr/include/x86_64-linux-gnu/bits/fcntl2.h:50:24: error: call to ‘__open_missing_mode’ declared with attribute error: open with O_CREAT or O_TMPFILE in second argument needs 3 arguments
   50 |    __open_missing_mode ();
      |    ~~~~~~~~~~~~~~~~~~~~^~
make: *** [makefile:40: prpclient.o] Error 1

[rogue]

Originally Posted by Happy5214

I'm getting a build failure on Linux:

In file included from /usr/include/fcntl.h:301,
                 from Socket.h:29,
                 from prpclient.cpp:22:
In function ‘int open(const char*, int, ...)’,
    inlined from ‘int main(int, char**)’ at prpclient.cpp:462:19:
/usr/include/x86_64-linux-gnu/bits/fcntl2.h:50:24: error: call to ‘__open_missing_mode’ declared with attribute error: open with O_CREAT or O_TMPFILE in second argument needs 3 arguments
   50 |    __open_missing_mode ();
      |    ~~~~~~~~~~~~~~~~~~~~^~
make: *** [makefile:40: prpclient.o] Error 1

I have never seen this compiler error before. For now, delete or comment out these lines in prpclient.cpp.

   int lockFile;
   lockFile = open("client.lock", O_WRONLY | O_CREAT | O_EXCL);
   if (lockFile <= 0)
      TerminateWithError("Only one copy of the PRPNet client can be run from this folder at a time.");