Sztaki Desktop Grid

  • Thread starter Deleted member 88227
  • Start date
Last night:


Capture3.PNG



This morning:

Capture4.PNG



Need more power!!!
 
ChristianVirtual said:
I don’t understand why those are so wired distributed ... I really not getting that many with all crontab/script tricks in place ...
Keep fighting XS !
Click to expand...

cause my windows batch file :punch: your linux script.
tbh i don't know really, it's just luck of the draw and how many systems you can throw at it. maybe try to make sure your scripts are staggered so they are hitting the servers at different times. I power up my VB one at a time so they are all slightly staggered that way. I like the progress i am seeing in the results!
 
I only have 5 systems that are hitting the project up and I seem to be doing pretty well.
 
Gilthanis said:
The progress bar doesn't work but the work units run for 1 hour each. If you really want to see some points boosting or get the badges quick.... https://hardforum.com/threads/goofyxgrid-multi-client-setup.1924298/

View attachment 77300
Click to expand...

I just did this on my main rig which was already doing a ton of stuff. First time i have ever made this Ryzen 7 really stutter . Looks like i need more ram but shit is expensive. I am guessing i could do this procedure inside each of my VB if i really wanted to push things?
 
Why did you post that here?

It's an NCI project which means non-cpu intensive. It basically just keeps track/logs what you do on BOINC projects IIRC.
 
Skillz is thinking of WUProp.

motqalden, it only uses a couple cycles of the CPU every so often to do the processing. Each work unit really doesn't need that much CPU by design. They could theoretically make a CPU intensive version if they wanted to and have dabbled with. However, most users have stated that they only support the project because it is NCI and that most would not support the "science" if it took away from other projects. However, by allowing people to milk the gross amounts of points you can earn running multiple clients, some of those people will go ahead and run their system rough anyways. https://en.wikipedia.org/wiki/Infinite_monkey_theorem However, Skillz is right that we should really discuss that in a thread not about SZTAKI. Perhaps the public forum for GoofyxGrid. https://hardforum.com/threads/goofyxgrid-home.1904588/

And yes you can do the same process within VM's if you have the RAM. Also, the IO's may tax the HDD, so don't run too many VM's on the same disk. Try spreading them out across multiple disks. SSD's obviously work better. There is also a 1,000 tasks in progress limitation built into BOINC. So, to get beyond that limitation on a single box, VM's are required. Goofy has 4 NCI work units which means 250 clients if all four are running.
 
motqalden said:
cause my windows batch file :punch: your linux script.
tbh i don't know really, it's just luck of the draw and how many systems you can throw at it. maybe try to make sure your scripts are staggered so they are hitting the servers at different times. I power up my VB one at a time so they are all slightly staggered that way. I like the progress i am seeing in the results!
Click to expand...
Build in a delay between the multiple instances to do that, but no effect ...
Run every 2 min and added some sleeps to get above 140seconds. Remain a mystery to me
 
By the way: keep running over May 31st as to my understanding the project wrote they keep the server open for results ... not sure if they also fully stop sending. Not that there is a bunker somewhere laying around or we miss to catch some resented WU due to validation issue etc.

As the system load is not that high suggest to at least until end of June to keep up or we see really nothing anymore moves and server status of the project reflect that.
 
That is totally the plan. I think this is going to come down to the wire and depend alot on pending validation work. I have 450 points in pending and i bet that mmmbeer guy has alot more.
last hour we gained 250 points or something like that and alot of our new dudes just came online today, but its gonna be really close! if this continues i wouldn't be surprised if they stop pushing Meisterkuehler and move to Ars to stop us overtaking, but we have at least done a good job of saving the loss of 3 points here which would have totally happened.
 
motqalden How many VMs/computers do you have running? I see you did over 2000 points in a day. Just curious.

I setup 15 Ubuntu virtual machines.

Does anyone know if the computer names have to be different? Is it good enough to just have one image, and "clone" it in Virtual box many times? I couldn't figure out how to automatically run a script on boot. I was trying to add a crontab entry, and point it to a script file, but it would never run it. Even after I made the file executable. I'm not sure what I was doing wrong there. Right now I have to past the script into the cmd prompt every time.
 
MN Scout said:
motqalden How many VMs/computers do you have running? I see you did over 2000 points in a day. Just curious.

I setup 15 Ubuntu virtual machines.

Does anyone know if the computer names have to be different? Is it good enough to just have one image, and "clone" it in Virtual box many times? I couldn't figure out how to automatically run a script on boot. I was trying to add a crontab entry, and point it to a script file, but it would never run it. Even after I made the file executable. I'm not sure what I was doing wrong there. Right now I have to past the script into the cmd prompt every time.
Click to expand...


4 total real computers, one is running 5 VM, 1 running 4 VM, 1 running 2 VM. 15 total .. When i was cloning i made sure to reset mac / network , but i was seeing error "client disconnected" and i started renaming them as i continued to clone and stopped seeing that. It is possible that i made a mistake and didn't reset the mac when cloning but i did the renaming to be sure. I also don't know how you would tell them apart in your stats if they had the same name. biggest thing to look for is check to make sure you don't have tasks with "client disconnected" error this would indicate a conflict.

Make sure you don't install boinc before you clone (like i was doing) or you will have to go back like i did and uninstall with revo-uninstaller and fully delete out all boinc folders / registry entries. using this method > then resetting network again and boinc reinstall i was able to recover clones that were problematic.

Edit: looking at your computers it seems yours are all ok. most of them have completed work i can see. some still have 0 but also no pending so they just havn't gotten any work yet. seems you only need to rename if you want to tell them apart
 
I ended up doing the VM thing two ways. 1st was cloning before installing BOINC. Second way was cloning a VM with BOINC.

My last 10 were done with cloning an already installed Ubuntu 18 BOINC setup (one of the first 5). I used the clone command in Virtual Box, on a machine that was shutdown without a Sztaki task running.

The first time the VM started I went into settings and immediately renamed the computer name with a new UbuntuVM##.
Then restarted right away.
It was happy after that. My BOINC came from the Ubuntu repository, so it automatically starts. I just had to paste the polling update script into a cmdline, and start that.

Looking at the points this morning, it appears that it was worth doing.
 
MixManSC has been running since last week and has had around 150 tasks already, but every single one of them "Error while computing" errors.

Any ideas what might be causing that?
 
If my calculations are correct we will have an ~11k point update today.

And Ars Technica will have around a 7k+ point update as well.
 
My progress is definitly slowing down since last night. I went from 15-20 in progress to only 5-8 at most all night from what i can see. I think alot of people have been jumping on this so who knows where we will land but you can't say we have not tried hard in the last push here and we stand a really good shot of overtaking. less than 5k points to go since i went to bed last night!!!
 
Skillz said:
MixManSC has been running since last week and has had around 150 tasks already, but every single one of them "Error while computing" errors.

Any ideas what might be causing that?
Click to expand...

My friend from work Darkeagle has the same issue for a week. I have been troubleshooting it to no avail. he says they progress for a few minutes and then just stop...
 
Pretty sure he's running on his servers so no to both.
 
Is their anything special I need to setup or config for this project? I was just looking at MN Scout computers and his completed tasks. He's getting 20+ points per task he completes while none of my tasks grant more than like 3 or 4 points per task.

If I am doing ~900 PPD with tasks that are only generating 3 - 4 points each imagine what my PPD would be if they granted 20+ points each.
 
Also we are only 4,373 points behind them with this latest update.
 
Might be different work units? I know this project has a few different things it focuses on so different wu might give different points
 
So I never get the higher paying WUs? ALL my tasks are low scoring.
 
motqalden said:
Well it looks like they are pulling away... at least we accomplished half the objective of not bleeding 3 points.
Click to expand...

I guess we'll call it a draw. I span up 40 EC2 single core free tier instances last night. I'd better shut them down before they start costing me money!

Great call to arms on this motqalden.
 
You must log in or register to reply here.
Back
Top