Sztaki Desktop Grid

  • Thread starter Deleted member 88227
  • Start date
let me know when you are done testing and have perfected your setup. Then you should put up a full How-to guide in the guides section for future reference. Docker is something I would love to learn more about but have had less than 0 time to dedicate to learning.

Right now I can see that the tasks are getting scored really low at around 0.2 pts. Digging into this I'm guessing it is because the BOINC CPU benchmarks haven't been ran, and all the CPU stats are reporting as "1000 million ops/sec". I need to figure out how to slowly loop through the instances and run the benchmark. Kind of funny since I see that this makes the wingman's task also score really low. :)
 
Sending the boinccmd --run_benchmarks to all of them isn't working. Still results in 1000 million ops/sec. I think that is the default, and is screwing up the point total I receive.

Edit: I think the key thing is that the Docker container needs to be properly setup and configured from the start. Attempting to modify or configure the instances after creation is not what this docker paradigm is about. They have a service command to scale up and down the instances, so the instance image should be right from the start.
 
Last edited:
Geeze i kind of feel bad now about slacking on this. It seemed like it was drying up and we were not going anywhere but MN Scout has really picked up the pace on this . looking at the charts we are actually gaining again. I thought for sure this was dead lol.

Edit: I still don't think the standings are going to change at this point but with no communication from the devs who knows when this will end...
 
Geeze i kind of feel bad now about slacking on this. It seemed like it was drying up and we were not going anywhere but MN Scout has really picked up the pace on this . looking at the charts we are actually gaining again. I thought for sure this was dead lol.

Edit: I still don't think the standings are going to change at this point but with no communication from the devs who knows when this will end...


Yeah I'm going to ride it out strong till the end. Definitely multiple VMs running the update script is the way to go. Each VM needs to have a different host name (decently certain on this) I saw constant notices of requesting too recently when I didn't change the host name.

Who knows, maybe it'll last long enough for me to get to a million. Probs not, but I can hope.
 
To be clear MN Scout 's issue with the host name is exclusive to VM's and possibly the combination of being cloned. I have ran into it myself a few times. I have even seen it when not cloning but having one set up identical in a very short period of time. However, that does not mean that the host name cannot be identical in general. Here is an example from running multiple clients at Goofy. Though they are the same host but different clients, they all show up with the same name and keep their own stats separate from one another.

hostnames.PNG
 
To be clear MN Scout 's issue with the host name is exclusive to VM's and possibly the combination of being cloned. I have ran into it myself a few times. I have even seen it when not cloning but having one set up identical in a very short period of time. However, that does not mean that the host name cannot be identical in general. Here is an example from running multiple clients at Goofy. Though they are the same host but different clients, they all show up with the same name and keep their own stats separate from one another.

View attachment 79871

Yeah i was curious about this as i was seeing the same problems (for Sztaki) until i started renaming and later i was looking at the goofy grid ones and they were fine and i was scratching my head wondering what the difference was.
 
I'm back in this never ending race. A hundred vms fired up this weekend.

Awesome. Report on what PPD you are seeing. I'm guessing 40 to 50k a day. You may take tasks away from me, but at least it is going towards our team.

How are you managing them? Are they connected to an account manager?

Edit: I'm hoping to fire up 30+ on my second computer Monday evening, it finishes tasks faster than my main computer, so I might be able to fire up double that quantity. Though I have no clue how it will handle the VM instances.
 
They might be getting worried now. This is the second day we've out produced them. XS is even starting to give themselves some points, too.

upload_2018-6-18_17-26-58.png
 
I just finished powering back up what i could so i hope to see some more points tomorrow.

"SZTAKI Desktop Grid is going to close down officially on May 31"

I'm thinking they didn't put 2018 here so could be next year :ROFLMAO:
 
Awesome. Report on what PPD you are seeing. I'm guessing 40 to 50k a day. You may take tasks away from me, but at least it is going towards our team.

How are you managing them? Are they connected to an account manager?

Edit: I'm hoping to fire up 30+ on my second computer Monday evening, it finishes tasks faster than my main computer, so I might be able to fire up double that quantity. Though I have no clue how it will handle the VM instances.

It's looking like just 20-25K per day with the work shortage taking it's toll. For some reason both your and my VMs seem to been benchmarked with a crazy integer speed allowing us to claim 15-20 credits per task. This pays dividends when we end up wingman for ourselves or each other. I'm also trying to pull CAS work with them but I've had limited success.

I'm managing them individually through BAM. I tried using BAM's group function but either it doesn't work properly or I screwed it up somehow.

Hope things have gone well with your additional VMs. It's great to see XS in panic mode splitting their output between themselves and ARS. What a shame it would be if they end up losing both spots after wasting all that effort on trying to push us down a place :LOL: We need to apply more pressure.
 
I got some backup attaching the project from GenMay. God damn I would love to see us claim 1st.
 
The project as a total is only doing between 40 and 70 thousand points per day. 434 active users yesterday.

Yesterday, I wasn't able to get to replicating my VMs onto my second computer. I have a sizable quantity of things to do tonight, and I'm booked solid the rest of the week. I'll see what I can do....
 
Back on the up and up. :cow:
upload_2018-6-19_16-58-51.png


Now to get this Arch linux VM setup further....

Edit: alright, I have a SystemD timer and service that updates Sztaki every 140 seconds. That is working. Just need to make sure that it properly processes tasks with no errors. Just tried installing the Intel-ucode package to hopefully fix a error:
Code:
stderr out
<core_client_version>7.8.4</core_client_version>
<![CDATA[
<message>
process got signal 11</message>
<stderr_txt>
SIGSEGV: segmentation violation

</stderr_txt>
]]>

Edit 2: I just reset the project. Hopefully that works. Now to play the waiting game for a WU to see if it is fixed.
Edit 3: Nope still failing with signal 11 errors. IDK. Something in the OS, or project code is unhappy. Not worth it to troubleshoot this any further.
 
Last edited:
So i wanted to do this and i had got all set and running last night but my landlord is doing a bunch of electrical so i was down all day and tomorrow as well. Managed to get about 800 pts yesterday so not too shabby. I will be back in a few days full bore with all my VM's. The good news is i am getting a dedicated 30 amp added for future DC headroom. I think that will keep me happy for a while.
 
And out

About SZTAKI Desktop Grid
SZTAKI Desktop Grid closed on June 21st, 2018.

SZTAKI Desktop Grid was operated by the Laboratory of Parallel and Distributed Systems at the Hungarian Academy of Sciences, Budapest, Hungary.

The SZTAKI Desktop Grid and its applications were partly supported by the IDGF-SP project. The work leading to these results has received funding from the European Union Seventh Framework Programme (FP7/2007-2013) under grant agreement n° RI-312297. The experts of the International Desktop Grid Federation provided further support for the SZTAKI Desktop Grid infrastructure, its applications, and its integration into the European Grid Infrastructure (EGI).
 
Bugger. We were so close. I'm kicking myself for not finding the time to set the VMs up earlier.

Was a good run. I learned a bunch of things, that will be useful in the future.

Me too. Already putting them to use on CAS and Asteroids.
 
dang, was really hoping you had a bunker cooking for asteroids as well.
 
I still had a number of WU pending for processing (not enough; but still) those are now in nirvana and project removed from clients. RIP
 
Back
Top