PRPNet released!

[mdettweiler]

I've noticed a very odd error that's shown up a few times today in the CRUS PRPnet server's console output:

[2009-01-05 19:46:36 GMT] Error sending ServerVersion: 1.0.4
 to localhost:3000
[2009-01-05 19:46:36 GMT] Pipe interrupt with code 13.  Processing will continue
[2009-01-05 19:46:36 GMT] Error sending WorkUnit: 98860*6^121721+1 1231184796 9860 6 121621 1
 to localhost:3000
[2009-01-05 19:46:36 GMT] imgunn1654@gmail.com (MyDogBuster) at *.*.*.*: Sent 98860*6^121721+1
[2009-01-05 19:46:36 GMT] Pipe Interrupt with code 13.  Processing will continue
[2009-01-05 19:46:36 GMT] Error sending End of Message
 to localhost:3000

Any idea what's causing this? Also, is this just benign, or will it cause problems at some point?

[rogue]

Originally Posted by mdettweiler

I've noticed a very odd error that's shown up a few times today in the CRUS PRPnet server's console output:

[2009-01-05 19:46:36 GMT] Error sending ServerVersion: 1.0.4
 to localhost:3000
[2009-01-05 19:46:36 GMT] Pipe interrupt with code 13.  Processing will continue
[2009-01-05 19:46:36 GMT] Error sending WorkUnit: 98860*6^121721+1 1231184796 9860 6 121621 1
 to localhost:3000
[2009-01-05 19:46:36 GMT] imgunn1654@gmail.com (MyDogBuster) at *.*.*.*: Sent 98860*6^121721+1
[2009-01-05 19:46:36 GMT] Pipe Interrupt with code 13.  Processing will continue
[2009-01-05 19:46:36 GMT] Error sending End of Message
 to localhost:3000

Any idea what's causing this? Also, is this just benign, or will it cause problems at some point?

I don't know what is causing this. Code 13 is a SIGPIPE, meaning that the socket connection between the client and server was broken. Is the client or server running on Windows?

[mdettweiler]

Originally Posted by rogue

I don't know what is causing this. Code 13 is a SIGPIPE, meaning that the socket connection between the client and server was broken. Is the client or server running on Windows?

The server is running on Linux; however, the client is probably running Windows (based on the fact that in its results, it lists the program as "phrot.p3.exe").

[rogue]

Originally Posted by mdettweiler

The server is running on Linux; however, the client is probably running Windows (based on the fact that in its results, it lists the program as "phrot.p3.exe").

I've seen these kinds of issues with Windows before with other software that uses sockets, but could never find the problem.

[rogue]

I found a bug in Work.cpp. In the Load() function, the code needs to be changed to read as:

if (!memcmp(line, "WorkUnit=", 9))
{
wu = &w_WorkUnit[i];

if (sscanf(line, "WorkUnit=%s %llu %s %s", wu->s_Name, &wu->l_TestID, wu->s_Program, wu->s_Residue) == 4)
{
if (sscanf(wu->s_Name, "%llu*%d^%d%d", &wu->l_k, &wu->i_b, &wu->i_n, &wu->i_c) == 4)
{
if (!strcmp(wu->s_Residue, "notdone") || !strcmp(wu->s_Residue, "inprogress"))
i_CompletedWorkUnits = i;
i++;
}
}
}

otherwise it will count an inprogress test as completed and potentially abandon work in progress by the client.

[rogue]

I've released version 1.0.5. You can d/l it from here. Here is what has been patched:

• Fixed an issue on Windows as Visual Studio does not initialize instance variable if they are not explicity initialized. (server and client)
• Fixed an issue when loading from a save file where the number of completed work units was not correctly updated. (client only)

[mdettweiler]

Rogue,

Recently we've been working on setting up a PRPnet server over at NPLB, and we were discussing how to integrate it into our existing LLRnet-based stats system. Karsten informed me of the -L command line switch to output results in LLRnet format (which I wasn't previously aware of--was this feature in existence from the very beginning, or was it added later?). However, that still doesn't quite solve the problem of trying to generate a server status page for a PRPnet server (like we do with our LLRnet servers at http://nplb-gb1.no-ip.org/llrnet/ and http://nplb.ironbits.net/). The prpserver.candidates file format is so different from LLRnet's knpairs.txt that it would be quite difficult to integrate into the existing status page generator scripts.

However, then I remembered that a while back you'd mentioned you were working on having the PRPnet server itself generate a status page, a la ECMnet. Did you ever get the chance to finish that? If not, when's the ETA for its release?

Thanks,
Max

[Mini-Geek]

Originally Posted by mdettweiler

Karsten informed me of the -L command line switch to output results in LLRnet format (which I wasn't previously aware of--was this feature in existence from the very beginning, or was it added later?).

Originally Posted by kar_bon

because Rogue(Mark) wrote the PRPnet-client/server, i think he's able to include an option like

-L make LLRnet-server ouput-format (3 lines per k/n-pair)

i.e. it's not included yet. Karsten's just commenting on that Rogue could include it.

(posted this in the individual k thread too)

[mdettweiler]

Originally Posted by Mini-Geek

i.e. it's not included yet. Karsten's just commenting on that Rogue could include it.

(posted this in the individual k thread too)

Whoops--my bad. Sorry about that. As I said over at NPLB regarding this--this isn't too big a deal, since my conversion script has been tested to work quite reliably.

Though, of course, it still would be nice to have a built in -L option for outputting LLRnet-format results.

[rogue]

Originally Posted by mdettweiler

Rogue,

Recently we've been working on setting up a PRPnet server over at NPLB, and we were discussing how to integrate it into our existing LLRnet-based stats system. Karsten informed me of the -L command line switch to output results in LLRnet format (which I wasn't previously aware of--was this feature in existence from the very beginning, or was it added later?). However, that still doesn't quite solve the problem of trying to generate a server status page for a PRPnet server (like we do with our LLRnet servers at http://nplb-gb1.no-ip.org/llrnet/ and http://nplb.ironbits.net/). The prpserver.candidates file format is so different from LLRnet's knpairs.txt that it would be quite difficult to integrate into the existing status page generator scripts.

However, then I remembered that a while back you'd mentioned you were working on having the PRPnet server itself generate a status page, a la ECMnet. Did you ever get the chance to finish that? If not, when's the ETA for its release?

Thanks,
Max

The status page I'm working on is a server status. It doesn't have user stats, although user stats shouldn't be too difficult. The main problem with user stats is how you roll them up.

I don't have an ETA right now. I've been busy with other things. The code is mainly written, but not tested. There are a couple of additional bells and whistles I want to included, but haven't done so yet.

[mdettweiler]

Originally Posted by rogue

The status page I'm working on is a server status. It doesn't have user stats, although user stats shouldn't be too difficult. The main problem with user stats is how you roll them up.

I don't have an ETA right now. I've been busy with other things. The code is mainly written, but not tested. There are a couple of additional bells and whistles I want to included, but haven't done so yet.

User stats won't be a problem--we've already got a big fancy NPLB stats site set up at http://stats.ironbits.net/ for our LLRnet servers, and we should be able to import PRPnet results into the database there without a problem after I run them through a little Perl script I wrote for converting PRPnet results to LLRnet format. Thus, the only thing that we'd need from the PRPnet software itself is the server status page.

No problem about not being able to release it yet--we've decided to start up the range were were planning on loading the new PRPnet server with in an LLRnet server for now, and then we can switch it over to PRPnet later on when it's all ready.

Max