I know there is a way to avoid getting the 'wrong' kind of tasks at a Boinc Project but does anyone have the actual cc_config.xml file, I can figure out the actual name in most cases, but need the file as a starting point. I'm talking about Projects, like Rosetta for me right now as I try to get only Python tasks, but also at other Projects that chose not to employ the task type selection, GpuGrid is another example, and as I try to get more apps over the 100 hour mark not having to delete dozens of the wrong types of tasks would help alot. Micro-managing is a pain in the butt with soo many apps.
mikey

1. catch at least one necessary task, you have to get:
a) a set of binary for the application
b) description of the app_version in client_state

2. create an app_info for the desired application

As far as I know, nope. If project does not provide option to select apps (like Rosetta) and does not allow anonymous platform with custom app_info.xml (like Rosetta), it can't be helped.
Of course it is possible to write a script to monitor current running boinc tasks and cancelling "bad tasks" but it is complicated...

Btw, does wuprop count hours from anonymous platforms? I suppose no, because elsewhere you can easily cheat wuprop with anonymous custom app that does nothing for N hours.

Sergey Kovalchuk wrote:

1. catch at least one necessary task, you have to get:
...

Sometimes it can be even easier, when the project allows to look into the server-side download folder - then you don't need to catch a task and just can zip the used app to the own host and continue with the rest. ;-)

den777 wrote:

Btw, does wuprop count hours from anonymous platforms? I suppose no, because elsewhere you can easily cheat wuprop with anonymous custom app that does nothing for N hours.

Are you referring to use the wuprop-app in an xml-file?
If yes, this is indeed forbidden, because you can use multiple instances like this. There was a separate thread for this somewhere here. You can get banned if you do so.
If you are referring to anonymous platforms on other projects, this is no problem since you mostly use the app given from the project for this. WUProp detects that and does count the hours.

---

Life is Science, and Science rules. To the universe and beyond
Member of BOINC@Heidelberg
My BOINC-Stats

1. catch at least one necessary task, you have to get:
a) a set of binary for the application
b) description of the app_version in client_state

2. create an app_info for the desired application

That could be it but I don't know how to write it yet, I will do some checking and see what I can find.

I think I'm closer but not quite there yet with this from my time on NanoHub:

<app_config>
<app>
<name>boinc2docker</name>
<max_concurrent>8</max_concurrent>
</app>
</app_config>

Maybe this with the names being correct?
<app_config>
<app>
<name>ithena_cnode</name>
<max_concurrent>0</max_concurrent>
</app>
<app>
<name>ithena_perf</name>
<max_concurrent>1</max_concurrent>
</app>
</app_config>

Would that make the project send me zero tasks for the first type of app and send me 1 task at a time for the 2nd app? Assuming I got the names right of course.

no it would not

this controls how many tasks the client will run

it has no impact whatsoever on what gets downloaded, which is a purely server side decision

no it would not

this controls how many tasks the client will run

it has no impact whatsoever on what gets downloaded, which is a purely server side decision

And that's what my research, including the same comment from Michael Goetz an Admin at PrimeGrid, has shown! Oh well back to aborting all the wrong tasks!!

At least Rosetta doesn't seem to have a penalty for failed/aborted tasks. The rosetta app used to always run for 6hr and abort on my ryzen system. Duration was set to 1hr and the mini rosetta tasks ran fine. So the rosetta tasks were aborted. If the project can't add an app selection to project preferences then that's what they get. That seems like a simpler implementation than a task duration.

I'd probably run some python tasks as well for hours but I can't get just those.

At least Rosetta doesn't seem to have a penalty for failed/aborted tasks. The rosetta app used to always run for 6hr and abort on my ryzen system. Duration was set to 1hr and the mini rosetta tasks ran fine. So the rosetta tasks were aborted. If the project can't add an app selection to project preferences then that's what they get. That seems like a simpler implementation than a task duration.

I'd probably run some python tasks as well for hours but I can't get just those.

I just kept aborting tasks until I got some, on my systems they are projected to only last for about 6 hours each but I have 2 running right now that are over 30 hours!! Each task reserves 8gb of ram so on my laptop with only 16gb of ram I only run 1 at a time but on my MacPro with 24gb of ram I run 2 at a time. The laptop is Windows and the tasks are taking between 8k and 16k seconds each, the MacPro is old and the I7 cpu is not nearly as fast.