mersenneforum.org  

Go Back   mersenneforum.org > Prime Search Projects > No Prime Left Behind

Reply
 
Thread Tools
Old 2009-11-29, 21:34   #23
mdettweiler
A Sunny Moo
 
mdettweiler's Avatar
 
Aug 2007
USA (GMT-5)

3×2,083 Posts
Default

Quote:
Originally Posted by henryzz View Post
wasnt gary going to do a test to see how long the UPS backup would last
Yes, he said he was going to do that around Thanksgiving. Gary, how'd that go?
mdettweiler is offline   Reply With Quote
Old 2009-11-30, 06:59   #24
gd_barnes
 
gd_barnes's Avatar
 
May 2007
Kansas; USA

2·52·7·29 Posts
Default

We get a fair amount of snow and ice in KS (and MO, which I'm 10 miles from) but that is virtually irrelavent to my outages. The summertime is virtually irrelavent to my outages. What IS relavent are thunderstorms in the spring and early summer. Although KS and OK have many tornadoes, those are mostly concentrated in the open plains. I am in the large metropolitan Kansas City area. Tornados are surprisingly rare here. (Much to my chagrin because I do enjoy storm chasing. )

In other words, the outages are largely random except in the spring and early summer. The problem is that the house that I bought, while not too old (built in 1986), is near a main street that still has above ground power lines unlike most of the rest of the area. I see KCP&L working on them frequently. If they will finally pony up the cost to bury them, I suspect I will have almost no outages. Before I was divorced 5 years ago, I lived 5 miles directly west of here in an area mostly built in 1990-1992. All lines were buried. We probably had a total of 5-6 outages in 14 years. Here I've had an avg. of 2-3 per year.

So it's mostly the above ground power lines and to a much lesser extent the weather that affects how many outages we have in the metropolitan area. I've had outages when the whether was perfect because someone ran their car into an electric pole or a bird landed on a line wrong and shorted it out.

Max, you said the status page (i.e. for the LLRnet servers) is for informational purposes only. That's true but it's highly important information that is needed ASAP for the PRPnet servers. If someone doesn't get Email notification of a prime, then it may not get reported for weeks or ever. I realize that the admins can check the primes on the server but we shouldn't be relying on that. Also, we need to be able to quickly see how many pairs are remaining in a server without having to add up every k-value.

I don't know what you mean by "the DB replicates much of that page's functionality". There's nothing that replicates the LLRnet status page's functionality or usefuleness. I'll reiterate again: The current PRPnet status pages are effectively useless for NPLB.

To drive home the point further: I'm not inclined to move any machines to the new PRPnet server for the 5th drive until we KNOW that the Email notification is working OR we have a status page that shows the primes. I suspect issues like this are why Ian wanted to either continue with an LLRnet server or do the next range manually for the 7th drive. If 2 admins are not inclined to use PRPnet servers, then I'm sure others will avoid them also. The Email notification has to be permanently fixed and/or a status page is needed that everyone can see.


Gary

Last fiddled with by gd_barnes on 2009-11-30 at 08:10
gd_barnes is offline   Reply With Quote
Old 2009-11-30, 07:03   #25
gd_barnes
 
gd_barnes's Avatar
 
May 2007
Kansas; USA

1015010 Posts
Default

Quote:
Originally Posted by mdettweiler View Post
Yes, he said he was going to do that around Thanksgiving. Gary, how'd that go?
I forgot about it. I'll do that now. One problem: The server machine is the only one on it now. I hate to let it shut itself down. I think what I'll do is let it run halfway down and double that time. Hopefully the little indicators of power remaining are reasonably accurate.

I'll report back here after the test is done.

Edit: Hah! I goofed. It makes a shrill beeping sound when it goes into battery mode. The instructions seemed to indicate that you had to hold the button down 1 second to make it go silent. Unfortunately that shuts it off. I subsequently found out that just a quick press silences it. So I just caused my own outage. Ergh! So you may see an outage on the GB servers of 5-10 mins. as I get them all restarted.

Last fiddled with by gd_barnes on 2009-11-30 at 07:17
gd_barnes is offline   Reply With Quote
Old 2009-11-30, 07:53   #26
gd_barnes
 
gd_barnes's Avatar
 
May 2007
Kansas; USA

2×52×7×29 Posts
Default

Max, after accidently turning the backup off after unplugging it (in an effort to silence it), I've now restarted all of the GB servers. You might check them to make sure everything's OK. One thing a bit strange that you might check out: For port 3000 that you added since the last time I restarted all the servers, there is no prpserver244 program. I had to run the more usual prpserver program instead.

OK, everyone, here's the scoop: I'm quite disappointed. With my one server machine running full bore on all 4 cores along with other stuff connected that uses virtually no power like a land line phone, router, modem, and fan, it drained ~80% of the battery backup in ~40 mins.

After spending $250 for a top of the line backup, I'm ready to send it back. Ergh! Anyway, the bottom line is that the backup appears to only be good for < 50 mins. I guess to have a longer backup, I'd have to not run the machine full bore.

The good news is that I can only recall 1 (maybe 2) power outages in 3 years of living here that lasted > 30 mins. From my perspective, that is certainly acceptable so I will leave all 4 cores running.

I guess what bothers me the most is that I had hoped to use it to back up 2 or 3 machines running all 4 cores for at least an hour. As it is, it will stay only on the server machine. I suppose that typical backups are not intended for modern machines running 4 cores full blast.


Gary
gd_barnes is offline   Reply With Quote
Old 2009-11-30, 16:51   #27
mdettweiler
A Sunny Moo
 
mdettweiler's Avatar
 
Aug 2007
USA (GMT-5)

11000011010012 Posts
Default

Quote:
Originally Posted by gd_barnes View Post
Max, after accidently turning the backup off after unplugging it (in an effort to silence it), I've now restarted all of the GB servers. You might check them to make sure everything's OK. One thing a bit strange that you might check out: For port 3000 that you added since the last time I restarted all the servers, there is no prpserver244 program. I had to run the more usual prpserver program instead.
Okay, looks good. FYI, don't use prpserver244 any more; that was only something temporary that I was using when Mark and I were debugging problems with the server. In fact, that has a few somewhat major bugs in it that are now fixed in the latest version (2.4.6). Fortunately none of those bugs came back to bite us in the short time they were running 2.4.4 just now. I've restarted all the servers with 2.4.6.

Quote:
OK, everyone, here's the scoop: I'm quite disappointed. With my one server machine running full bore on all 4 cores along with other stuff connected that uses virtually no power like a land line phone, router, modem, and fan, it drained ~80% of the battery backup in ~40 mins.

After spending $250 for a top of the line backup, I'm ready to send it back. Ergh! Anyway, the bottom line is that the backup appears to only be good for < 50 mins. I guess to have a longer backup, I'd have to not run the machine full bore.

The good news is that I can only recall 1 (maybe 2) power outages in 3 years of living here that lasted > 30 mins. From my perspective, that is certainly acceptable so I will leave all 4 cores running.

I guess what bothers me the most is that I had hoped to use it to back up 2 or 3 machines running all 4 cores for at least an hour. As it is, it will stay only on the server machine. I suppose that typical backups are not intended for modern machines running 4 cores full blast.
Hmm...yeah, I guess that sounds about right for a UPS of that size. Nonetheless, yes, given the specs I'd read for that model, it seemed like it should have more capacity than that...but then, what do I know about interpreting UPS specs?

BTW, about the status pages and PRPnet that you mentioned a couple posts up: first of all, the status pages do NOT run the email notification. That's the DB. Nontheless, yes, the status pages are the next-best thing to email notification when checking for primes. I'll see if I can get something cooked up within the next day or two. Also, I'll try to figure out why PRPnet's bult-in email notification (which was supposed to hold us over in the interim until we got DB import turned on for G3000) wasn't working.
mdettweiler is offline   Reply With Quote
Old 2009-11-30, 17:16   #28
mdettweiler
A Sunny Moo
 
mdettweiler's Avatar
 
Aug 2007
USA (GMT-5)

3×2,083 Posts
Default

Behold! We now have a PRPnet status page somewhat like the one we have for the GB LLRnet servers:
http://nplb-gb1.no-ip.org/prpnet/
The first thing the script does is convert the results it's dealing with to LLRnet format, which saved me a lot of work on my end and also ensured that things are largely the same for the end user. The main difference is in the part that displays the #, and first and last k/n pairs remaining. I changed it to a more general "lines in prpserver.candidates", since PRPnet uses two lines to denote assigned pairs, thus meaning that this is close to, but not exactly, the # of k/n pairs remaining. (The exact number would be rather difficult to determine, though this should be close enough for most purposes.) Also, PRPnet sorts its prpserver.candidates file (its equivalent of LLRnet's knpairs.txt, for the uninitiated) by k primary and n secondary, regardless of what order it's actually going to hand out the pairs. Thus, the first and last k/n pairs shown will reflect that. (Again, this is still largely close enough for most purposes, as the lowest n for the first k will probably be rather close to the true lowest n. If you need more granular detail, you can get it per k on the built-in PRPnet status pages.)

Note that as of now, the first and last lines have the text "N 0 active" on them. That's part of PRPnet's prpserver.candidates format, and I haven't gotten around yet to writing code to parse that out for this status page. I'll get around to that eventually, though it's only a cosmetic difference so not quite my highest priority.

Probably the biggest benefit of this status page is the fact that it keeps track of all primes found in a file, a la the LLRnet status pages. That should simplify prime-checking greatly until we can get email notification working again.

(BTW: yes, I know G2000 isn't on the status page yet as of this writing. I'll do that next.)
mdettweiler is offline   Reply With Quote
Old 2009-11-30, 18:21   #29
Brucifer
 
Brucifer's Avatar
 
Dec 2005

313 Posts
Default

Hey Max the page looks great!! Thanx!! Much much simpler than looking through a bunch of log files.
Brucifer is offline   Reply With Quote
Old 2009-11-30, 19:12   #30
Mini-Geek
Account Deleted
 
Mini-Geek's Avatar
 
"Tim Sorbera"
Aug 2006
San Antonio, TX USA

17×251 Posts
Default

Quote:
Originally Posted by mdettweiler View Post
Also, PRPnet sorts its prpserver.candidates file (its equivalent of LLRnet's knpairs.txt, for the uninitiated) by k primary and n secondary, regardless of what order it's actually going to hand out the pairs. Thus, the first and last k/n pairs shown will reflect that. (Again, this is still largely close enough for most purposes, as the lowest n for the first k will probably be rather close to the true lowest n. If you need more granular detail, you can get it per k on the built-in PRPnet status pages.)
If you'd prefer it sorted by length, (which is practically by n for NPLB's purposes) this is a simple change in PRPnet's source: prpserver.cpp line 1446, change "theCandidate = g_CandidatesByBKCN[i];" to "theCandidate = g_CandidatesByLength[i];". I've tried this, and AFAICT it works perfectly.
Not sure if it's worth making the change each version, but there it is.

Last fiddled with by Mini-Geek on 2009-11-30 at 19:13
Mini-Geek is offline   Reply With Quote
Old 2009-11-30, 19:57   #31
MyDogBuster
 
MyDogBuster's Avatar
 
May 2008
Wilmington, DE

B1A16 Posts
Default

Nice job Max. Now I don't have to be re-trained which at my age is next to impossible.
MyDogBuster is online now   Reply With Quote
Old 2009-11-30, 21:22   #32
gd_barnes
 
gd_barnes's Avatar
 
May 2007
Kansas; USA

2·52·7·29 Posts
Default

Excellent. That's a good stop gap page until we can get something that is more exact. The main useful thing is the primes file.
gd_barnes is offline   Reply With Quote
Old 2009-11-30, 21:34   #33
gd_barnes
 
gd_barnes's Avatar
 
May 2007
Kansas; USA

100111101001102 Posts
Default

Quote:
Originally Posted by Mini-Geek View Post
If you'd prefer it sorted by length, (which is practically by n for NPLB's purposes) this is a simple change in PRPnet's source: prpserver.cpp line 1446, change "theCandidate = g_CandidatesByBKCN[i];" to "theCandidate = g_CandidatesByLength[i];". I've tried this, and AFAICT it works perfectly.
Not sure if it's worth making the change each version, but there it is.
Outstanding!! Thanks for the great tip Tim.

Max, can you please make this change to the various PRPnet servers, recompile them, stop them, and restart them. Perhaps you can call the changed binary "prpservern", which indicates that the prpserver.candidates file would be sorted by n. The sorting of the candidates file has always annoyed me greatly. You can't quickly tell what candidates are handed out to people. You also can't quickly tell what the "true" lowest and highest n-value that are remaining.

You'd also have to do this with each new version of the PRPnet server that comes out unless you can talk Mark into making it the default sorting method in the code. For that reason, I would not suggest putting the version # in the name of the changed binary. We'd have too many of them very quickly.

Another thing, can we delete all of the previous temporary prpserver244 binaries from the various prpnet folders? I also see that one of them has a prpserver230, which I assume can be deleted. That way I'll always know to restart them with the prpserver binary as needed.


Thanks,
Gary

Last fiddled with by gd_barnes on 2009-11-30 at 21:39
gd_barnes is offline   Reply With Quote
Reply

Thread Tools


Similar Threads
Thread Thread Starter Forum Replies Last Post
PSP goes prpnet ltd Prime Sierpinski Project 86 2012-06-06 02:30
Setting up PRPnet Mattyp101 Conjectures 'R Us 2 2011-02-07 13:53
PRPNet 4.0.1 Released Joe O Sierpinski/Riesel Base 5 1 2010-10-22 20:11
PRPNet 3.0.0 Released rogue Conjectures 'R Us 220 2010-10-12 20:48
PRPNet released! rogue Conjectures 'R Us 250 2009-12-27 21:29

All times are UTC. The time now is 01:19.

Fri Jul 10 01:19:37 UTC 2020 up 106 days, 22:52, 0 users, load averages: 1.25, 1.34, 1.35

Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2020, Jelsoft Enterprises Ltd.

This forum has received and complied with 0 (zero) government requests for information.

Permission is granted to copy, distribute and/or modify this document under the terms of the GNU Free Documentation License, Version 1.2 or any later version published by the Free Software Foundation.
A copy of the license is included in the FAQ.