OFFICIAL "SERVER PROBLEMS" THREAD

[Uncwilly]

Originally Posted by retina

https://www.mersenne.org/report_classic/

My money is on the classic report not knowing anything about PRP and VDF.

[LaurV]

Originally Posted by chalsall

TL;DR: Deal with it.

We do. For years.

But with the new work types, the web page looks ugly, no assignment list, instead of a list of assignments, you see just a long list of shitty errors.

And moreover, if you lose your assignments somehow (happened to me in the past, lost the worktodo occasionally, and got it back from the assignments page) then you can't get it back anymore, as the table shows the same shitty errors.

So, George/James, do something with it, please!

[Prime95]

Originally Posted by Uncwilly

My money is on the classic report not knowing anything about PRP and VDF.

Bingo. Fixed.

[James Heinrich]

Originally Posted by LaurV

The "unknown worktype 3" error in the /workload page should be fixed, it seems the web page is not aware of the new work types like PM1-S, etc.

I'm also not aware of them. How do I get such an assignment so that I can see something "broken" appear on my workload page, and thereby try to fix it?

[chalsall]

Originally Posted by James Heinrich

I'm also not aware of them. How do I get such an assignment so that I can see something "broken" appear on my workload page, and thereby try to fix it?

Put something like this into a worktodo.txt or worktodo.add file associated with an active mprime instance:

Pminus1=1,2,14540027,-1,1000000,45000000,73

The next time the mprime "talks" with Primenet an AID will be given, which shows up at the top of the GIMPS Assignment Details. But, not in the lower reports (the Per-CPU) on the same page which are used to recover lost assignments (by way of a full block of text, including the AIDs).

Things are moving fast. It isn't surprising this was overlooked. Should be a trivial fix.

[James Heinrich]

Originally Posted by chalsall

Put something like this into a worktodo.txt... Should be a trivial fix.

From one codemonkey to another, thank you for your highly useful example. It should now be fixed.

[chalsall]

Originally Posted by James Heinrich

From one codemonkey to another, thank you for your highly useful example. It should now be fixed.

Thank you. It is.

I call this "high bandwidth" comms. Sometimes Geeks can solve a problem almost instantly, so long as the bug report is clear...

[kriesel]

At the countdown to 60M DC milestone in https://www.mersenne.org/report_milestones/ there is a count of remaining assignments to the next DC milestone which is a link to a list of active assignments to the next DC milestone, but PRP with proof as DC are not listed in active assignments to the next DC milestone in the resulting list: https://www.mersenne.org/assignments...xtf=1&excert=1. This understates the total active count toward achieving the milestone.
It takes https://www.mersenne.org/assignments...xtf=1&excert=1 since PRP & proof generation on a first LL is a first PRP for the same exponent.

[Happy5214]

I just turned in a manual result for a composite P-1 factor using the Prime95 30.8 JSON format, and it appears to have duplicated the result in the database. The exponent in question is 5214289. Here is the JSON if it helps:

{"status":"F", "exponent":5214289, "worktype":"P-1", "factors":["194048402408014112454803526544134231148823"], "b1":2500000, "b2":6555248700, "d":30030, "poly1-size":2880, "poly2-size":9868, "fft-length":327680, "security-code":"xxxxxxxx", "program":{"name":"Prime95", "version":"30.8", "build":8, "port":8}, "timestamp":"2022-01-26 10:23:21"}

My guess is it may be related to the code parsing the new format, but that's just a guess.

[Zhangrc]

https://www.mersenne.org/primenet/

Today's actual TFLOP/sec:
109236293446701972014860550879846174470729121692282379973690158076813251314464816684697703625736479330744125608972500335041881635966799588768542122770717038389522306570951989406651134903437302446865226345693822428553881894640715152530204299987894291227064472301059523122806715725119488.000

Wow!!
https://www.mersenne.org/report_top_500/
Look at the "for you"

[axn]

https://www.mersenne.org/report_expo...6666667&full=1

no factor from 2^80 to 2^128
no factor from 2^127 to 2^1024

Sounds legit

EDIT: Also https://www.mersenne.org/report_expo...7654323&full=1 (no factor from 2^75 to 2^128)