LLRnet servers for NPLB

[IronBits]

As to the Port numbers you guys are floating around...
seems silly to me at this end of the stick...
Port 400 to you, is actually port 9000 to the server. /me hugs the rock solid smoothwall router
You can call the port anything you like, I'll always make it something that fits within my server port range that satisfies me and ensures there are no collisions with other well known port ranges.
Also, after watching this server run, under the worst load where we did almost 13,000 in one day, I believe the server could handle twice that load easily.
Just so long as we remember that there are NO proxy servers allowed to connect to it, we should never go down.
It was rock solid before the proxy problem, and it has been rock solid since the proxy went away.

One or two servers should be plenty with the amount of computers we have collectively.
Switching port numbers around for the sake of trying to make them mean something is more work than it's worth.

Port 400, come get whatever it's serving up...

If you need to add another Server/Port, sure, we can use 5000 as we have used it before and it is still there waiting.
I can create stats for it like I do on port 400 if you like...

It would be nice if we could move CRUS stuff to Gary, as it just sits there doing nothing 99% of the time, and I'd prefer to work with the active stuff.

Max, seeing that you have control over Gary's servers, you can create your own personal llrnet servers over there right?

This will allow me to focus on nplb hourly/daily stats for our active crews, while I try to teach myself some php and mysql and hone my vbscript skillz.

[mdettweiler]

Originally Posted by gd_barnes

Ugh, and I was sure that we had this settled already.

Not to beat a dead horse but what you're trying to avoid on port 4000 is what is going to happen on port 400. When any port dries, it means lost processing time because it's bound to happen at some point when at least one person is not right there at his machine to switch it over to the other port.

I just now reread our PM exchange. Here is how it went:

Me:
You:
It ended there without any kind of additional suggestion to use port 8000 for k=800-1001. I guess part of my point is to stop the whole "new server" thing and just stick with what we already have to avoid lost processing time and more new servers. We've had enough servers already to choke a horse. (Ian's phrase that I kind of like, lol) Any new server can be used as a roving server.

What I suppose I didn't pick up on is: What do you mean by a "pseudo-plan for port #'s and k-ranges". I wasn't aware of any such plan. If that's the case, then we need to go all the way with it: Port 400 for k=400-600, port 600 for k=600-800, and port 800 for k=800-1001 but that would be 2 new servers. bleh!

As for switching JobMaxTime, that's a 5 min. task twice. As for loading n>600K pairs, that could be done now in the existing ports.

One final thing on this: Don't forget that Ian and me have to switch upwards of 40 cores any time a port changes. It's a pain. Others a little less but still an annoyance.

I'll send you the k=400-1001 for n=600K-1M file shortly.


Gary

Okay, I see what you mean. I guess we'll just stick with G4000--and since I've already got G8000 all set up in the Perl scripts and all, I'll just leave it there in case we ever need it later as a "roving" server for rallies or whatnot.

I guess I had sort of assumed that, considering as how we had originally picked the port 400 number based on the k-range (and how I, by extension, picked port 4000 based on that), that we would continue the numbering scheme--but, yeah, I guess it doesn't really matter.

Max

[em99010pepe]

I moved 4 cores to G4000 and the server went down. I'll stay away from it.

Edit: It's up again.

[gd_barnes]

Originally Posted by mdettweiler

Okay, I see what you mean. I guess we'll just stick with G4000--and since I've already got G8000 all set up in the Perl scripts and all, I'll just leave it there in case we ever need it later as a "roving" server for rallies or whatnot.

I guess I had sort of assumed that, considering as how we had originally picked the port 400 number based on the k-range (and how I, by extension, picked port 4000 based on that), that we would continue the numbering scheme--but, yeah, I guess it doesn't really matter.

Max

Honestly, I wasn't ever aware that we picked port 400 and 4000 based on the k-range. After all, we originally had port 300 for k=400-1001.

Do you now feel you wasted a lot of time with the Perl scripts or were they just all copied over from port 4000? If so, I'll tell you what we could do with port 8000: Since we will need another port for our k=1005-2000 for n=350K-500K effort, we could use it for that. Since you've got it set up, we don't want to set up yet another one for it. The n=50K-350K effort for that k-range would have too fast of tests to put in a server; well, at least for n<~200K or so.

Here is one thing that I am considering: Putting on a big push on sieving for k=1005-2000 after the first of the year so that we can start looking for smaller top-5000 primes in late Jan. or early Feb. Ian is already at P=900G and will stop at P=1T. I'm speculating that we'll need to get it to P=4T-6T to start LLRing.

My reasoning: The 5000th prime will likely be at n=350K by Feb. I'd like to get as many primes for that k-range in the top-5000 database for future historical reference. Sure, NPLB will get them and then lose them shortly thereafter but not only will they remain in the database as long as the site exists, we'll quickly move up to a range where the primes won't drop off so quickly.

Also, I suspect that many folks with lesser resources will enjoy going back to the days of when we first started the project where they had a good chance of finding a top-5000 prime in a week's time. To put it in perspective, n=700K will take 8 times as long to find a prime as n=350K! That is: 4 times as long to test with 1/2 as much chance of finding a prime. Hence why the score is 8 times as high at the top-5000 site.


Gary

[gd_barnes]

Originally Posted by IronBits

Max, seeing that you have control over Gary's servers, you can create your own personal llrnet servers over there right?

This will allow me to focus on nplb hourly/daily stats for our active crews, while I try to teach myself some php and mysql and hone my vbscript skillz.

Hey now: I can shut him down at any point. lol Anyway, yeah, he does all the server stuff on my machines. I'm still a little amazed by it that someone I've never met is controlling parts of my machines. It's kind of funny when you think about it. Alas, they are only prime cruncher machines. Nothing else is on them.

I know you're a bit bugged that the CRUS servers get little work on them. (Actually no work for months if I remember right.) I think it's only port 6; correct? I think I mentioned previously that we'd get some resources on them after our 1st drive is done. That is still the plan.

Regardless, we can get them moved to my machines if Max is up to the task. Max, I have no problem if David would prefer that we move the CRUS server(s) to my machine(s).

Initially, I thought this would be a good effort for port 8000 but then I realized that you have that one intigrated in NPLB stuff so probably not such a good choice. Regardless of that, if you're up to the task of setting up a port or two on my machine for CRUS, go right ahead.


Gary

[MyDogBuster]

Ian is already at P=900G and will stop at P=1T

Actually I sent you p=800 - p=1000 earlier tonight.

I can start sieving again if you want to take it to the 4T - 6T range.

I just didn't know where you wanted me to stop.

[henryzz]

Originally Posted by gd_barnes

Also, I suspect that many folks with lesser resources will enjoy going back to the days of when we first started the project where they had a good chance of finding a top-5000 prime in a week's time. To put it in perspective, n=700K will take 8 times as long to find a prime as n=350K! That is: 4 times as long to test with 1/2 as much chance of finding a prime. Hence why the score is 8 times as high at the top-5000 site.

definitely
bring on the primes!!!

[mdettweiler]

Originally Posted by gd_barnes

Honestly, I wasn't ever aware that we picked port 400 and 4000 based on the k-range. After all, we originally had port 300 for k=400-1001.

Do you now feel you wasted a lot of time with the Perl scripts or were they just all copied over from port 4000? If so, I'll tell you what we could do with port 8000: Since we will need another port for our k=1005-2000 for n=350K-500K effort, we could use it for that. Since you've got it set up, we don't want to set up yet another one for it. The n=50K-350K effort for that k-range would have too fast of tests to put in a server; well, at least for n<~200K or so.

Okay, I like that idea--let's do that. You are partially correct in both your statements regarding the Perl scripts: yes, I was able to copy over everything from port 4000, but I had to do each block of code at a time, by copying it into a different file, doing a search/replace for 4000 and 8000, repsectively, and then pasting it back into the main file immediately after the respective port 4000 code. It probably took me about 5-10 minutes. Not too terribly bad, but...it's still at least a small hassle to add or remove servers to my scripts.

Meanwhile, I've marked port 8000 as "for future use" rather than "NPLB 7th Drive" on the status page, to reflect its new purpose.

Here is one thing that I am considering: Putting on a big push on sieving for k=1005-2000 after the first of the year so that we can start looking for smaller top-5000 primes in late Jan. or early Feb. Ian is already at P=900G and will stop at P=1T. I'm speculating that we'll need to get it to P=4T-6T to start LLRing.

My reasoning: The 5000th prime will likely be at n=350K by Feb. I'd like to get as many primes for that k-range in the top-5000 database for future historical reference. Sure, NPLB will get them and then lose them shortly thereafter but not only will they remain in the database as long as the site exists, we'll quickly move up to a range where the primes won't drop off so quickly.

Hmm, I see--sort of the same thing we did with the early part of the 1st Drive. Good idea. Of course, I presume that we'll still be starting on LLRing n=50K-350K even before sieving is ready for the top-5000-range work?

Also, I suspect that many folks with lesser resources will enjoy going back to the days of when we first started the project where they had a good chance of finding a top-5000 prime in a week's time. To put it in perspective, n=700K will take 8 times as long to find a prime as n=350K! That is: 4 times as long to test with 1/2 as much chance of finding a prime. Hence why the score is 8 times as high at the top-5000 site.

Indeed. As one of those folks with lesser resources, I am looking forward to this new effort with great interest!

Max

[mdettweiler]

Originally Posted by gd_barnes

Hey now: I can shut him down at any point. lol Anyway, yeah, he does all the server stuff on my machines. I'm still a little amazed by it that someone I've never met is controlling parts of my machines. It's kind of funny when you think about it. Alas, they are only prime cruncher machines. Nothing else is on them.

LOL--actually, Gary and I do have one private server running on there, separate from the NPLB status pages. That was in fact the first one I set up, so that I could get a feel for how all this server stuff worked.

[mdettweiler]

Originally Posted by em99010pepe

I moved 4 cores to G4000 and the server went down. I'll stay away from it.

Edit: It's up again.

Hmm...that's odd. I checked the console output log and didn't see anything--but then again, the terminal application only keeps about 500 lines of output before discarding them. Oh well, at least whatever it was is now fixed.

Maybe you caught it in the middle of a prune? I'm not sure, but from what I've seen of the server I would imagine that it would be momentarily unresponsive to queries during such times.

[gd_barnes]

Originally Posted by em99010pepe

I moved 4 cores to G4000 and the server went down. I'll stay away from it.

Edit: It's up again.

I think we're confident that it should be good now. If you still decide to move 4 cores to port 4000 a little later on, let me know and I'll move my quad back to port 400.