mersenneforum.org - OFFICIAL "SERVER PROBLEMS" THREAD

mersenneforum.org (https://www.mersenneforum.org/index.php)

- PrimeNet (https://www.mersenneforum.org/forumdisplay.php?f=11)

- - OFFICIAL "SERVER PROBLEMS" THREAD (https://www.mersenneforum.org/showthread.php?t=5758)

I have not had any problems the last several hours.

The server error log is reporting many 300 second timeouts, but it happens in bursts. No problems for an hour, then a few dozen timeouts.

All I can think of, if that my blasting TF assignments has caused Chris' scripts to go a little crazy. I have no evidence to back that up.

[QUOTE=Prime95;550411]OK, not as bad as first thought... TF assignments below 99xxxxxx that had an unverified first-time LL/PRP test got axed (in other words exponents that need a DC).[/QUOTE]And as at the time of writing this our friend lolapus has taken back 7 TF assignments. But strangely it says assigned "2020-06-23".

[url]https://www.mersenne.org/assignments/?exp_lo=50845813&exp_hi=53000000&execm=1&exdchk=1&exfirst=1&exp1=1[/url]

I have two machines running from the very same network. I tried manual communication on both of them "simultaneously". Result: One of them communicated flawlessly, the other one is timing out since then (already multiple tries).

The only difference between them (aside of hardware): The one which functioned reported completion dates for ECM, the one which could not connect tried to report completion dates for LL-DC.

[QUOTE=kruoli;550428]The only difference between them (aside of hardware): The one which functioned reported completion dates for ECM, the one which could not connect tried to report completion dates for LL-DC.[/QUOTE]

Ah, good catch. Both of my prime95/mprime machines are running LL-DC and can't communicate. One still gets the CURL timeout error, the other seems to get a server response of 500 - Internal server error.

[QUOTE=Prime95;550411]OK, not as bad...[/QUOTE]
[B][SIZE="2"]The connection to the server takes ages. With manual submission I was able to enter a p-1 result twice into the database.[/SIZE][/B]
[url]https://www.mersenne.org/report_exponent/?exp_lo=107753341&full=1[/url]

Output of communication Thread Windows64,Prime95,v29.8,build6

[SIZE="1"]][B][Mon Jul 13 10:02:18 2020 - ver 29.8]
Updating computer information on the server
PnErrorResult value missing. Full response was:
<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">

<html xmlns="http://www.w3.org/1999/xhtml">

<head>

<meta http-equiv="Content-Type" content="text/html; charset=iso-8859-1"/>

<title>500 - Internal server error.</title>

<style type="text/css">



</style>

</head>

<body>

<div id="header"><h1>Server Error</h1></div>

<div id="content">

<div class="content-container"><fieldset>

<h2>500 - Internal server error.</h2>

<h3>There is a problem with the resource you are looking for, and it cannot be displayed.</h3>

</fieldset></div>

</div>

</body>

</html>[/B][/SIZE]

Strange indeed: within the last 30 minutes, most of my machines (win+linux, including some Colab-based ones) had a successful update. OTOH my laptop, which shares the same net access, had a response similar to moebius' one.

Forced communication: no way. Renamed prime.spl, so as to start with an empty one: again, no way.

If it's confirmed elsewhere, guess this is a tricky one...

[QUOTE=Prime95;550421]All I can think of, if that my blasting TF assignments has caused Chris' scripts to go a little crazy. I have no evidence to back that up.[/QUOTE]

Not guilty Gov!

When you alerted us to this I had GPU72 unreserve the DCTF assignments it held (so they weren't listed as expired), and then brought it a couple of hundred for Mark to work.

Beyond that, no differences. It ain't me, but one of my P-1'ing machines just had another timeout experience.

Something sure is out of place. I am getting HTML in [I]mprime's[/I] console when it tries to send a result. I think I will stop running for now.

I managed to report 1 manual result as a test, but then when I tried with the rest it got stuck again.

Are there any policy changes related to Fermat ECM?
I had ten F26 curves almost ready to finish, (supposed to finish on 14th and 17th and 22), and an eleventh not yet started (supposed to finish on 27th) and all 11 magically disappeared from my assignments. Lots of wasted time and resources... In the log I am getting "invalid key", and any try to get ECMF work with P95v29.8 fails. I have also seen that the assignment type disappeared from the manual page. If there is any policy change, like "hey guys, use your resources better than doing ECMF!", then it should be announced on the web page, news section, or something, and started assignments that are still reporting daily progress should not be killed! :rant:

I understand that new assignments are not handled out anymore, but making me waste about a month of 10 cores worth, just few days before finish, I don't understand...