Formula Boinc Sprints 2018

[Ken g6]

So, ODLK1. I don't even know the right URL for this one?

 

[StefanR5R]

It's in the sticky thread,

ODLK1
BOINC-project ODLK1 continues to solve the problem of BOINC-project ODLK.
The project generates a database of canonical forms (CF) of diagonal Latin squares (DLS) of order 10 having orthogonal diagonal Latin squares (ODLS).
Jan. 2018, Working Project. CPU Only. Windows, Linux and FreeBSD.

 

[zzuupp]

aw p00p! I haven't been able to connect to that in two months. It just times out everytime.

 

[ao_ika_red]

aw p00p! I haven't been able to connect to that in two months. It just times out everytime.

Have you tried different isp? I once had similar problem across different projects and switching the isp (land cable isp to mobile phone) would do the trick.

 

[StefanR5R]

Half an hour ago the web site was slow, now it is completely DDoS'd.

 

[Ken g6]

Half an hour ago the web site was slow, now it is completely DDoS'd.

This may wind up being as bad as the MOD_PYTHON project.

 

[StefanR5R]

The end of this sprint (Monday 03:59 AM UTC) is already drawing near, rapidly.

Until then, not everybody will manage to join the project and to complete the ~1.5 MB download of the ODLK1 application.

;-/

 

[biodoc]

I can't get to the web site either. Adding project in boinc manager times out.

 

[zzuupp]

I can't get to the web site either. Adding project in boinc manager times out.

Same. I can't get a browser to open their homepage.

 

[zzuupp]

Have you tried different isp? I once had similar problem across different projects and switching the isp (land cable isp to mobile phone) would do the trick.

Alas, that's not an option.

 

[ao_ika_red]

Alas, that's not an option.

I try using my mobile browser and it's still down. So, as Stefan said, it's DDoS'd.

 

[StefanR5R]

Adding project in boinc manager times out.

It took me about two hours to create my own account (i.e.: add it in the first host via boincmgr, set nickname + team affiliation + basic preferences in the web interface).

After that, I added the project to additional hosts with

C:\Program Files\BOINC\boinccmd --host ABC123 --passwd XYZ987 --project_attach 'https://boinc.multi-pool.info/latinsquares/' 0123456789abcdef0123456789abcdef​

Replace the red&italic parts with your own data. I took the 32 characters long hex code from C:\ProgramData\BOINC\account_boinc.multi-pool.info_latinsquares.xml of the first host. It's in the <authenticator> tag at the top of the file.

Doing this via boinccmd instead of the boincmgr "Add project..." dialogue isn't really faster or any more successful during project server overload. But it spares you to sit there and wait to be able to press [OK] or whatever.

--------

So, this project is whitelisted by Gridcoin, and Gridcoin is doing 89.7 % of the project's daily production.

This means that the DDoS is probably not due to us who want to do a Formula Boinc Sprint there, but due to the miners who are loving this project to death currently.

 

[StefanR5R]

I completed a number of tasks by now. And while all looks well on a Linux Xeon E3v3 so far, I get a high fraction of computation errors on Linux single-socket and dual-socket Xeon E5v4.

Though without uploads, I have nothing to go on what's happening. Except two things: (a) The error'ed tasks have the same elapsed time as successfully completed tasks. (b) Task state goes from running to 100 % -> aborted -> computation error. So perhaps this is the type of boinc bug with a race condition between execution finish and writing of the output files.

Edit, one difference between the good E3v3 and the bad E5v4's was that the former was set to 99 % CPUs, the latter to 100 % CPUs. I reduced the later to less than 100 % CPUs as a precaution. Either due to this or by pure chance, I haven't had errors in the last half hour.

 

[StefanR5R]

I hate this MOD_PYTHON project. When it's selected it's always ERRORing out.

Le tirage au sort du projet n'est pas totalement aléatoire. Cela nécessite une action manuelle pour la validation du projet.
Ces temps-ci, j'ai tendance à oublier de valider.

via Google translate:
"The draw of the project is not totally random. This requires a manual action for project validation.
These days, I tend to forget to validate."
https://forum.boinc-af.org/index.php/topic,7660.150.html

 

[TennesseeTony]

If anyone (such as myself) is STILL trying to get an account created, much less attaching a client, this link might save you a few pages of navigation, I haven't made it that far yet, lol, so not for sure....

https://boinc.multi-pool.info/latinsquares/create_account_form.php


EDIT: Nevermind, I got this message twice:

Can't create account
Invalid email address: you must enter a valid address of the form name@domain
Click your browser's Back button to try again.

 

[Ken g6]

I finally got an account created and hooked to the TeAm. Now to get stuff downloaded and hook up more computers. I don't suppose it works to copy the bigger files from one machine to another?

 

[biodoc]

It takes a lot of effort but I'm getting there. One boinc instance on each machine running and second instances are downloading. One machine had a lot of download errors. This is frustrating.

 

[StefanR5R]

I don't suppose it works to copy the bigger files from one machine to another?

It works, if you shut down the client on the target machine and edit all relevant lines into client_state.xml, then restart the client. I haven't done this with ODLK1, but with other projects on two or three occasions. (This requires same machine architecture of course.)

It takes a lot of effort but I'm getting there.

One PC which I added maybe 5 hours ago is still struggling to download the large (not) exe files. This PC is away from the others, and it's late night here now, otherwise I would perform the sneakernet transaction now which Ken asked about.

 

[TennesseeTony]

All but one intended race host is attached to the project, and of course struggling with downloads and uploads. It appears the server for this project is connected to the rest of the world via dial-up internet, with perhaps a connection speed of 28.8 kbit/s.

 

[TennesseeTony]

Formula BOINC's intended purpose (#3) is to "..... also help small or new projects....."

I think instead they are putting nails in the coffin for these poorly implemented projects.

 

[Kiska]

Though without uploads, I have nothing to go on what's happening. Except two things: (a) The error'ed tasks have the same elapsed time as successfully completed tasks. (b) Task state goes from running to 100 % -> aborted -> computation error. So perhaps this is the type of boinc bug with a race condition between execution finish and writing of the output files.

You can open client_state.xml in notepad++ and scroll through that file to find those WUs, can you search by WU name to see status, in the xml it will have the stdout and that should tell you whats going on

 

[TennesseeTony]

I'm not bunkering, as I had such a late start, and the connectivity issues with the server, etc, but so far I have a 10:1 ratio of pending versus validated. So no real loss I guess.

 

[zzuupp]

I wish all good luck. It doesn't look like all be any help at all.

I've got one box connected, but it can't/won't receive tasks. The other is still SOL.

 

[StefanR5R]

I completed a number of tasks by now. And while all looks well on a Linux Xeon E3v3 so far, I get a high fraction of computation errors on Linux single-socket and dual-socket Xeon E5v4. [...] perhaps this is the type of boinc bug with a race condition between execution finish and writing of the output files.

You can open client_state.xml in notepad++ and scroll through that file to find those WUs, can you search by WU name to see status, in the xml it will have the stdout and that should tell you whats going on

Thanks. It's the familiar error:

<stderr_out>
<![CDATA[
<message>
finish file present too long
</message>
<stderr_txt>
21:02:20 (6393): called boinc_finish(0)

</stderr_txt>
]]>
</stderr_out>

 

[StefanR5R]

Uploads work better for me than downloads. The server appears to be located behind a consumer-grade heavily asymmetric connection.