PRPnet beta test server

[Brucifer]

I do have to say though that prpnet linux client is pretty decent as far as willingness to give up time slices for my desktop. Runs smoother to me in that regard than llrnet.

[Razor_FX_II]

Runs very nice on Vista 64bit and doesn't interfere with my Boinc GPUGRID work.

[AMDave]

Max,

I sent you an email about a record in the GB-2000 file:
results_20090522_0001_GB_nplb_2000.txt

[AMDave]

Originally Posted by Brucifer

I do have to say though that prpnet linux client is pretty decent as far as willingness to give up time slices for my desktop. Runs smoother to me in that regard than llrnet.

That is excellent feedback for JP.
I hope someone will pass that on.

[AMDave]

sorry.
I meant that is great feedback for rogue.

[gd_barnes]

Max,

Can you post a status as to where we are at on the PRPnet server?

I would like to put a high priority on getting all of the bugs worked out of it, if there are any left. :-)

In my mind, the 3 main things to do are:

1. Check the residues of the mini-drive vs. the results that I sent you.

2. Make sure the stats interface is working correctly.

3. Make sure all results are being copied off correctly.


Once those 3 things are done, then I'd like to get it rolled out.


Thanks,
Gary

[mdettweiler]

Originally Posted by gd_barnes

Max,

Can you post a status as to where we are at on the PRPnet server?

I would like to put a high priority on getting all of the bugs worked out of it, if there are any left. :-)

In my mind, the 3 main things to do are:

1. Check the residues of the mini-drive vs. the results that I sent you.

2. Make sure the stats interface is working correctly.

3. Make sure all results are being copied off correctly.


Once those 3 things are done, then I'd like to get it rolled out.


Thanks,
Gary

1. I still have yet to check the residuals for the mini-drive results. However, at the moment my highest priority is to...

2. Get the stats interface working correctly. At first we thought everything was OK (and it was) but then AMDave and I realized that we'd need to set up a separate system for handling doublechecks so that they don't get rejected as duplicates since the first-pass results are already in the DB. We're actively working on that at this time. Rest assured, no results will be lost; we have everything on file, the only thing we have to do is get it into the DB correctly.

3. I have verified that all results are being copied off correctly. The conversion script for producing LLRnet results from PRPnet output files is working great, and the daily files are showing up exactly as they're supposed to. The only thing left to do is produce interim results throughout the day like we do with LLRnet servers, but ATM that's a slightly lower priority than getting the stats imports working.

Interestingly enough, the stats import hangups have nothing to do with PRPnet specifically. They are in fact only due to the fact that we are now doublechecking ranges for which we have the original residuals in the DB, and it would happen just the same even if it was on an LLRnet server.

I'll keep you posted as to our progress.

Max

[Lennart]

Max server is out off work


Lennart

[mdettweiler]

Originally Posted by Lennart

Max server is out off work


Lennart

Holy shmoly! That was fast!

The server dried out a LOT quicker than I'd ever expected. I'll talk to Gary about the next batch of work which he is currently sieving.

Meanwhile, if any of you need something to keep those cores busy, may I recommend the PRPnet server of our sister project, Conjectures 'R Us, at nplb-gb1.no-ip.org port 3000. It is rather easy to configure it as a "fallback server" for when the NPLB server is out of work or cannot be reached--instructions for doing this are included in prpclient.ini. The server is currently loaded with plenty of Sierpinski base 6 work to keep those hungry cores happy.

Max

[gd_barnes]

Max,

We won't roll out anything on any new k-ranges until sieving has reached P=4T in about a week. Although that may be tentative depending on a couple of things that I haven't decided on yet.

Please go ahead and load another double-check range in there. Since Lennart is helping, you might make it a bigger range.

Although getting this going is top priority, it's also important that it be nearly as user friendly as LLRnet. I've heard a comment or 2 to the contrary so if we go with it, we have to give very detailed instructions on running it. Perhaps even creating a "care package" like you did for me on my machines and posting it in our threads would be the way to go.


Gary

[mdettweiler]

Originally Posted by gd_barnes

Max,

We won't roll out anything on any new k-ranges until sieving has reached P=4T in about a week. Although that may be tentative depending on a couple of things that I haven't decided on yet.

Please go ahead and load another double-check range in there. Since Lennart is helping, you might make it a bigger range.

Although getting this going is top priority, it's also important that it be nearly as user friendly as LLRnet. I've heard a comment or 2 to the contrary so if we go with it, we have to give very detailed instructions on running it. Perhaps even creating a "care package" like you did for me on my machines and posting it in our threads would be the way to go.


Gary

Okay, sounds good. I'll load up 300K-350K and upgrade the server to PRPnet 2.1.5 while I'm at it.

Meanwhile, about instructions: I have some basic setup instructions in the first post of this thread, but more refined ones will be posted once PRPnet leaves beta. Until then, my priority is on keeping the server running happily and on getting the rest of the backend stuff figured out. But, yes, I have been keeping that in mind; it looks like PRPnet 2.1.5 will be a stable release that will last for a while as the newest, so I will endeavour to get something up as soon as I can.

Max