Hi,

Is there a sample app_config.xml for Rosetta? If not, I think I need at least the <app_name> and maybe the <plan_class>.

Thanks.

Doug

Hi,

Is there a sample app_config.xml for Rosetta? If not, I think I need at least the <app_name> and maybe the <plan_class>.

Thanks.

Doug

What are you trying to get the app_config file to do for you?

If nothing specific then you don’t *have* to have one. If you have a specific requirement then that drives the content.

Sorry, I should have stated that. I want to limit the number of CPUs Rosetta uses to 2. For some reason, lately BOINC has been running 4 Rosetta tasks at a time, often ones that are due later than tasks from other projects I run (WCG).

Up until 2 or 3 weeks ago, BOINC virtually always apportioned tasks more or less equally. But now it's wanting to run 3 or 4 Rosetta tasks almost all of the time (at least that I check). I consider my WCG project to be at least as important as Rosetta, so I don't want it getting neglected.

I know micro-managing the scheduling is supposed to be counter-productive, which is why I've waited this long to try to do something about it. But this behavior has persisted for a while now.

Thanks for asking for clarification.

Doug

Sorry, I should have stated that. I want to limit the number of CPUs Rosetta uses to 2. For some reason, lately BOINC has been running 4 Rosetta tasks at a time, often ones that are due later than tasks from other projects I run (WCG).

Up until 2 or 3 weeks ago, BOINC virtually always apportioned tasks more or less equally. But now it's wanting to run 3 or 4 Rosetta tasks almost all of the time (at least that I check). I consider my WCG project to be at least as important as Rosetta, so I don't want it getting neglected.

I know micro-managing the scheduling is supposed to be counter-productive, which is why I've waited this long to try to do something about it. But this behavior has persisted for a while now.

Thanks for asking for clarification.

Doug

<project_max_concurrent>2</project_max_concurrent>

Thanks. That looks like that will work.

Now, how did you know that? I haven't seen that element in any of my other app_config files. Is there a schema or other list somewhere of the possible elements in an app_config file?

Thanks again.

Doug

Thanks. That looks like that will work.

Now, how did you know that? I haven't seen that element in any of my other app_config files. Is there a schema or other list somewhere of the possible elements in an app_config file?

Thanks again.

Doug

It will work for any Project should you decide to run several at a time, ie 2 Rosetta 3 NanoHub and 2 Private GFN Number tasks
Just copy and paste that file into each Projects file are in Boinc and change the number as needed. Be sure to go into the Boinc Manager, Options Read config files so it takes affect right away.

I know micro-managing the scheduling is supposed to be counter-productive, which is why I've waited this long to try to do something about it. But this behavior has persisted for a while now.

With your computers hidden i's pretty much impossible to see what's actually going on.

The biggest cause of unexpected behaviour is a large cache setting- that is one that's more than half a day or so.
The more projects you run, the longer it takes for your Resource share settings to be honoured. Also given that the deadlines for Rosetta Tasks are 3 days, and there have been all sorts of configuration issues combined with many Tasks erroring out, BOINC would appear to be doing it's best to meet the requirements you have asked it to. But limiting the number of cores/threads Rosetta can use is just going to make it take even longer for that to occur.

Running more than one project, you would be best served to have no cache at all. If one project has no work, the others will pick up the slack. When that project comes back again, it'll get some extra time compared to the others to make up for it falling behind. Having a cache just results in it take much, much, much longer for your Resource share settings to be met.
Set your cache to 0.01 days and 0.01 additional days (or 0.5 day and 0.01 days additional if you really feel the need for a cache), let all projects have equal access to all available resources, and things will sort themselves out within a week or two.
Having a cache, micromanaging & multiple projects, expect it to take several months- at least- for your Resource share settings to be met (if they ever can be).

---

Grant
Darwin NT

Thanks. That looks like that will work.

Now, how did you know that? I haven't seen that element in any of my other app_config files. Is there a schema or other list somewhere of the possible elements in an app_config file?

Thanks again.

Doug

It will work for any Project should you decide to run several at a time, ie 2 Rosetta 3 NanoHub and 2 Private GFN Number tasks
Just copy and paste that file into each Projects file are in Boinc and change the number as needed. Be sure to go into the Boinc Manager, Options Read config files so it takes affect right away.

And make sure you remember what you have done, or leave a note where you will see it.
Recently someone couldn't figure out why Rosetta wouldn't ask for more work & run more Tasks, then they remembered that they had previously set some values in a configuration file.... Problem fixed.

---

Grant
Darwin NT

Thanks. That looks like that will work.

Now, how did you know that? I haven't seen that element in any of my other app_config files. Is there a schema or other list somewhere of the possible elements in an app_config file?

Thanks again.

Doug

https://boinc.berkeley.edu/wiki/Client_configuration

Grant,

Thanks for your informative reply. My computer should no longer be hidden - I'm not sure if I originally did that on purpose or by mistake.

My Computing Preferences here on the web site are this:
- Store at least: 0.3 days of work
- Store up to an additional 0.3 days of work

However, my local Computing Preferences in the BOINC app are this:
- Store at least: 0.7 days of work
- Store up to an additional 0 days of work

And at the top of that BOINC page, it says "Using local preferences". I believe this means these setting override any setting on any project's web config pages. If I understand correctly, you are saying to set these "local" prefs to .01 and .01.

Thanks.

Doug

My Computing Preferences here on the web site are this:
- Store at least: 0.3 days of work
- Store up to an additional 0.3 days of work

However, my local Computing Preferences in the BOINC app are this:
- Store at least: 0.7 days of work
- Store up to an additional 0 days of work

And at the top of that BOINC page, it says "Using local preferences". I believe this means these setting override any setting on any project's web config pages. If I understand correctly, you are saying to set these "local" prefs to .01 and .01.

Yep.
Local preferences override web based preferences, and looking at the results for that system it appears to using the Local values- it's taking roughly half a day on average to return a Rosetta Task.

The idea is that for those that have a few, or dozens (and some have hundreds) of systems, you set your preferences on the web site- and any system using that account will use those preferences. You also have the option of 3 different web based preferences- under the labels of Home, Work, & School so you can have 3 different settings, easily applied by just selecting the one you want to use for a given system.
If you've got a particular system that needs completely different values for some reason, then you can use the local settings on that computer (as they override the web based settings).

With more than 1 project, there's not really any need for a cache, hence my suggestion of using 0.01 & 0.01. Basically the system will only download work as it needs it. If a project doesn't have any work, then it will get work from one of your other projects, and do extra work for the one that was out of work for a while when it finally gets some more work. If you have a cache, it will download extra work from the project that does have work, meaning the one without work will be owed even more time when it finally does get work- and will take longer for it's debt to be paid back before it will start processing work for the other projects again.

---

Grant
Darwin NT