NPLB Database

[kar_bon]

questions:

1.
why was the dat-file "results_20080928_0658_IB_nplb_7000.txt" updated on the server?
the file is totally different to that one i got for same date!

2.
why was "results_20081114_0657_IB_nplb_400.txt" updated? there's no difference from before!

3.
some GB4000-port files missing on http://nplb.ironbits.net/results/?C=M;O=D (only on http://nplb-gb1.no-ip.org/llrnet/results/) as used before!

[AMDave]

1 and 2 were reloaded to fix an old date format issue in the stats db, for completeness basically. The files were not updated as such, they just got a new timestamp in the web folder. I would like to do a diff on the file that you have of 1 as there should be no difference.

3, I will look into in about 10 hours, but I expect that is 1 or 2 files that Max and I had a problem with when the new GB port started up the other day. We are confident that the data is in the stats db [due to the hourly progress updates], but it looks like I need to push those 2 files through again to line things up after all.

[AMDave]

There does not appear to be any files missing.

The files are in order of filesystem timestamp which may have shown them out of sequence.

I'll see if we can make that order by name.

/ed- done, link from stats menu now sorts by name descending by default -ed/

[kar_bon]

Originally Posted by AMDave

There does not appear to be any files missing.

The files are in order of filesystem timestamp which may have shown them out of sequence.

I'll see if we can make that order by name.

/ed- done, link from stats menu now sorts by name descending by default -ed/

look at gary's page for his GB4000 and GB8000 server:
there're results for GB8000 from 2009-02-13 on:

and

results_20090214_0001_GB_nplb_4000.txt is from GB8000!!!!


i tried to progress the new results but yesterday it was the horror:

1.
results_20090212_0000_IB_nplb_4000.txt contains this:

user=gd_barnes
[2009-02-11 01:23:03]
511*2^630637-1 is not prime.  Res64: 27C0C814EE688F0D  Time : 2465.0 sec.
user=gd_barnes
[2009-02-11 00:40:34]
511*2^630637-1 is not prime.  Res64: 27C0C814EE688F0D  Time : 8458.0 sec.

2x the same result!!! how did this happen?

2.
results_20090214_0001_GB_nplb_4000.txt contains this:

user=MyDogBuster
[2009-02-13 00:00:22]
315*^600282-1 is not prime.  Res64: 71F02131A1C1E24C  Time : 467.0 sec.

this is a result from the server GB8000!!! again: how did this happen?

3.
results_20090213_0001_GB_nplb_4000.txt
results_20090214_0001_GB_nplb_4000.txt
are totally unuseable!!!

with "*^" instead of "*2^" and results not time-sorted!

so please to all:

check everthing first before setup a new server and test this!

to Gary:

please, 4 servers (3 for the Drives #5, #6 and #7 and one for other things) will be enough.
now we are at 8 servers and with such sh*t resultfiles it cost me more time to correct those data.
NPLB -> No Prime Left Behind implies to be correct but please in all purposes!

hope this will be considered!

PS: i've uploaded the 3 above files corrected here: www.rieselprime.de/dl/new_res200902.zip

[AMDave]

kar_bon

Ah I see - the mislabeled file due to the port re-direct hiccup.
please x-ref this thread:
http://www.mersenneforum.org/showthr...105#post163105

Max,

can you help out kar_bon here?

Please let me know if there are any files that need to be re-sent to the stats db.

[kar_bon]

Originally Posted by AMDave

kar_bon

Ah I see - the mislabeled file due to the port re-direct hiccup.
please x-ref this thread:
http://www.mersenneforum.org/showthr...105#post163105

Max,

can you help out kar_bon here?

Please let me know if there are any files that need to be re-sent to the stats db.

no need, see previous post!

[AMDave]

Many thanks kar_bon.

I will re-load these files in the stats db.

Max,

please replace the files in the GB web folder with the corrected GB files from kar_bon. (link in PS above)
Skip the IB file obviously.

I have made the replacements in the IB web-folder.
The stats db re-load will be completed sometime tomorrow (UTC+10)

[kar_bon]

Originally Posted by AMDave

Many thanks kar_bon.

I will re-load these files in the stats db.

Max,

please replace the files in the GB web folder with the corrected GB files from kar_bon. (link in PS above)
Skip the IB file obviously.

I have made the replacements in the IB web-folder.
The stats db re-load will be completed sometime tomorrow (UTC+10)

oh, i forgot the 4th issue:

4.

user=MyDogBuster
[2009-02-12 00:00:11]
987*2^632124-1 is not prime.  Res64: 332089FB79D05A75  Time : 543.0 sec.
user=MyDogBuster
[2009-02-12 00:00:15]
875*2^632120-1 is not prime.  Res64: BCAD6B398C35C1D6  Time : 1251.0 sec.
user=MyDogBuster
[2009-02-12 00:00:23]
847*2^632125-1 is not prime.  Res64: 1AC8863DA0ABFD69  Time : 445.0 sec.

these 3 results were in the files
results_20090212_0001_GB_nplb_4000.txt and results_20090213_0001_GB_nplb_4000.txt.

but it's cleared with above new result-files!

[AMDave]

Lovely.

We know there was a problem with the files and this was pointed out.

Unfortunately, I myself in fact had nothing to do with it or I might have written to you personally to ask for your advice on the corrections required.

However, you have taken it upon yourself to enact the corrections without providing the opportunity for anyone else to assist you or to learn how to do so.

I have thanked you for the corrections and taken them on board.

If you feel that I have personally offended you, then please PM me directly.

Otherwise, I would be most happy if you desist repeating my name in your posts about this issue as my interpretation of your written tone does not represent you well at this time.

I am moving on past this. I hope that you will join me.

[kar_bon]

i'm at NPLB from the beginning, so i know how to handle those resultfiles.

there were other issues in the past and it will be happen again, i'm sure!

David, i appreciate your enormours work for the NPLB Database and all the stats and layout for our project! Great work!

what i'm criticise is the amount of different servers and therefore more chances to make a mistake:
- a new user had to read first precisely which server to choose
- a server can break down
- the scripts had to be adjust for every server

many points to fail, so everything has to be tested before going online!

and another: these servers do their work automatically, so no need to do work by hand.
if there are pairs missed (or perhaps doubled or not the correct server like the last time) i will get aware of them by my spreadsheet/macros i had wrote since the first LLRnet results to get an overview and, what's most important, verify all results!
i easily can add a new part for a new server but's it becomes more timeconsuming with every file i have to process, and things like the last cost me hours of work that were unnecessary!
if i got salary for that, no way, make errors, i got them corrected...
but i'm got my work and this is just fun (should be!).

that's all

(*** getting up from the couch and squeeze the hankie ***)


oh,
PS:
yes, we shall move on! with some considerable thoughts!

[AMDave]

Thank you for your reply.

I have reviewed your webpages and can empathise with the stress induced by the workload that you have, and indeed the non-systematic method in these forums of reserving by post wherever people feel like it. Tracking those down cannot be easy.

I hope that you can gather some heart from the recent work to clean out port IB-2000.
IB-3000 is almost finished as well.
A number of us are working to close-out port IB-9000.

That should reduce your workload considerably in terms of this project.

Indeed we are doing this for fun.
And humans are humans and mistakes happen.
I think that the lesson about testing has been learned.

If your workload is too great and the work caqnnot easily be simplified, then I hope you will consider the possibility of sharing it and work out what is required to achieve that.

There are plenty of people here that are capable and probably willing, if asked.

That is how I came to be working on the stats :)

In the mean time, we are giving IB-3000 and IB-9000 a thorough flogging for you ;)