I have several machines running BOINC projects. They are all at my home, but they have different purposes, OSs and hardware configurations and I wish to use different computing profiles to them. I see that there are default, home, school and work profiles available (rather quaintly constrained and unimaginative thinking, but that's a conversation for another day), but I cannot see where to assign a particular profile to a particular machine. Where do I do that, please?

Through project web sites.

Your 'account' page at any given project website has an entry for "Computers on this account", with a link to 'view' them. For each computer linked to the project, you can see the host name, and a link for that computer's details. [Other users can't see the name - just an ID number - and a limited amount of detail]

At the foot of the page of details, you can see the 'Location' that computer is currently assigned to: this is a drop-down list, and you can change it to another profile. Click the 'Update location' to confirm a change.

Thanks for that, Richard. I found the information of which you speak for World Community Grid, but for Climate Prediction and Milky Way all machines on the same account use the same profile. I guess that's just the way they do things? I notice the Climate Predictions project has some other recommendations regarding processor usage, conditions for suspending computing.

Thanks again.

I have several machines running BOINC projects. They are all at my home, but they have different purposes, OSs and hardware configurations and I wish to use different computing profiles to them. I see that there are default, home, school and work profiles available (rather quaintly constrained and unimaginative thinking, but that's a conversation for another day), but I cannot see where to assign a particular profile to a particular machine. Where do I do that, please?

Using the location would restrict you to 4 sets of computers. If you have more configurations than this you can set the parameters at local level and vary it by computer.

MilkyWay certainly has the full set of locations available: Preferences for this project MilkyWay@home preferences.

CPDN is going through a minimalist phase, but the framework is there, too.

Climate Prediction is the same as all other projects. If you want to use a different Location, then just change it from the default.

Thanks, Les. My request wasn't simply about locations; in fact it wasn't about location at all: it was about executing different profiles for different machines and projects on those machines. The concept of profiles in BOINC is implemented as locations, but that's only one criterion: different machines and different projects are also valid reasons for wanting different configurations IMO.

Climate Prediction expressly isn't the same as all other projects. It explicitly states that all machines on that account use the same profile. MilkyWay is the same - in fact it uses the same page layout for editing "locations" (profiles).

I can - and it looks like I will have to - configure each machine separately, but that still doesn't solve the unique recommendations of the CP project.

Thanks, Bryn Mawr. Yes, location-based profiles would indeed limit me that way, so I'll have to have per-machine profiling. The only issue there is that I can't customize per computer AND per project, but that's OK: Climate Prediction is the only project I run that has different recommendations to the others, so I'll probably run that on a machine that doesn't get disturbed much; that'll resolve the suspend issue.

Thanks for your response.

I've been running Climate Prediction for 15 years or so, and if I want a different setup for different computers, I just change the setting for that machine.

Yes, Les, that's what I think I'll have to do, and that's no problem.
As you have a lot of experience with this project, maybe you can answer another question. I'm having trouble running it now - I don't get work units. I've made various changes to my computers and their respective project allocations over the years - nothing surprising there - but I haven't got a CP work unit on any machine for a couple of months now. Any clues?

Additionally, I just tried subscribing to it ona 2013 iMac, and received the warning that "this project may not have work for your type of computer. Do you still want to add it anyway?" The Mac icon shows up on the list of supported systems, however. Any clues there?

EDIT: Oh never mind on that last question. I got the notice that the x86_64-apple-darwin architecture isn't supported.

Thanks for your advice.

Ah, you're running Windows.
There hasn't been much work from the 3 research groups that use that, since soon after the pandemic started.
It's all Linux at present.

There may be some Mac work in a few weeks, BUT as with all the work here, it will be 32 bit work.

********************

It occurred to me that by "preference" you mean the ability to select which area of the planet to run models.
This option was removed a few years ago, as people were ignoring some, in preference to running work from areas closer to where they lived.
And with the newer programs running at higher resolutions, it's more about the modelling, than the area that's being modeled.
So, yes, that page was removed.

Another change, is that each modelling program is only being written for a single OS, to save on programmer person power.
With a couple of exceptions, just to muddy things a bit.

You can see from the Applications page what is on which OS.

I have two Linux boxes, two Windows boxes and a 2013 iMac running or logistically capable of running BOINC projects. At present I can't get work units for CP on either the Windows or the Linux boxes; I don't know why. The messages in the event log tell me that nothing was downloaded, but they offer no explanation - and certainly nothing related to configuration issues.

By "preference" I'm only referring to machine configuration; I am totally happy with the organizations running whatever projects they wish on my machines; and that goes for CP and WCG. Thanks for guessing about how my deranged mind might work, though :-)

So only Linux projects for CP? That explains the empty WU list on my Windows boxes. Thanks for that info :-)

So... on to the 32-bit Linux issue: in my investigation I saw this morning that I needed to install 32-bit libraries on my Linux boxes (the upgrade from 32 to 64 was one of the changes I was talking about having made relatively recently). I've just done that earlier this afternoon, so it may be that I simply need to wait for the server to notice.

Thanks again for your help.

Thanks, Bryn Mawr. Yes, location-based profiles would indeed limit me that way, so I'll have to have per-machine profiling. The only issue there is that I can't customize per computer AND per project, but that's OK: Climate Prediction is the only project I run that has different recommendations to the others, so I'll probably run that on a machine that doesn't get disturbed much; that'll resolve the suspend issue.

Thanks for your response.

But you cannot have different profiles per project anyway. Things like %cpu, memory allocation, etc are machine based, not project based.

I have two Linux boxes, two Windows boxes and a 2013 iMac running or logistically capable of running BOINC projects. At present I can't get work units for CP on either the Windows or the Linux boxes; I don't know why. The messages in the event log tell me that nothing was downloaded, but they offer no explanation - and certainly nothing related to configuration issues.

By "preference" I'm only referring to machine configuration; I am totally happy with the organizations running whatever projects they wish on my machines; and that goes for CP and WCG. Thanks for guessing about how my deranged mind might work, though :-)

So only Linux projects for CP? That explains the empty WU list on my Windows boxes. Thanks for that info :-)

So... on to the 32-bit Linux issue: in my investigation I saw this morning that I needed to install 32-bit libraries on my Linux boxes (the upgrade from 32 to 64 was one of the changes I was talking about having made relatively recently). I've just done that earlier this afternoon, so it may be that I simply need to wait for the server to notice.

Thanks again for your help.

With the 32bit libraries CP blacklist hosts where they notice that the libraries are not installed and you may have to post a request on the forum for them to clear that.

The other possibility is one that stopped me from processing CP for 8 months, check you cc_config to make sure no_alt_platform is not set.

I have two Linux boxes, two Windows boxes and a 2013 iMac running or logistically capable of running BOINC projects. At present I can't get work units for CP on either the Windows or the Linux boxes; I don't know why. The messages in the event log tell me that nothing was downloaded, but they offer no explanation - and certainly nothing related to configuration issues.

By "preference" I'm only referring to machine configuration; I am totally happy with the organizations running whatever projects they wish on my machines; and that goes for CP and WCG. Thanks for guessing about how my deranged mind might work, though :-)

So only Linux projects for CP? That explains the empty WU list on my Windows boxes. Thanks for that info :-)

So... on to the 32-bit Linux issue: in my investigation I saw this morning that I needed to install 32-bit libraries on my Linux boxes (the upgrade from 32 to 64 was one of the changes I was talking about having made relatively recently). I've just done that earlier this afternoon, so it may be that I simply need to wait for the server to notice.

Thanks again for your help.

With the 32bit libraries CP blacklist hosts where they notice that the libraries are not installed and you may have to post a request on the forum for them to clear that.

The other possibility is one that stopped me from processing CP for 8 months, check you cc_config to make sure no_alt_platform is not set.

So, "no_alt_platform" is not in /etc/boinc-client/cc_config.xml; I take it that it shouldn't be. (The file only contains settings for log flags). In fact, the file is date-stamped Feb 23 2016, along with gui_rpc_auth.cfg and remote-hosts.cfg. Only global_prefs_override.xml is recently changed, and that's because I made the configuration local this morning.

I'll post a message on the forum, as you suggest.

Thanks.

But you cannot have different profiles per project anyway. Things like %cpu, memory allocation, etc are machine based, not project based.

Yeah, that should have been blindingly obvious to me :-)

With the 32bit libraries CP blacklist hosts where they notice that the libraries are not installed and you may have to post a request on the forum for them to clear that.

The other possibility is one that stopped me from processing CP for 8 months, check you cc_config to make sure no_alt_platform is not set.

I removed the project and reconnected. cpdn immediately downloaded and app, aa work unit, and started crunching. As of this moment I'm 20 minutes in, and only 24 days, 9 hours and 30 minutes to go :-)

Thanks, once again for your help; you too, Les :-)

Julian.

With the 32bit libraries CP blacklist hosts where they notice that the libraries are not installed and you may have to post a request on the forum for them to clear that.

For completeness, if a host is blacklisted it will when you look at the page for that computer show

Maximum daily WU quota per CPU -1/day
Instead of

Maximum daily WU quota per CPU 0/day

When he has time, Andy is looking to write a script to automate this for machines crashing tasks with the missing 32bit libraries message though given the lack of posts on the CPDN boards from those crashing everything, I don't expect a flood of complaints about no tasks being sent when it happens.

For completeness, if a host is blacklisted it will when you look at the page for that computer show

Maximum daily WU quota per CPU -1/day
Instead of

Maximum daily WU quota per CPU 0/day

When he has time, Andy is looking to write a script to automate this for machines crashing tasks with the missing 32bit libraries message though given the lack of posts on the CPDN boards from those crashing everything, I don't expect a flood of complaints about no tasks being sent when it happens.

"-1/day" a bit cryptic, don't you think??? I wouldn't have guessed that that meant I was blacklisted.

Anyway, thank to you, to Andy, and all of those who help make the world a better place.

Julian.

"-1/day" a bit cryptic, don't you think??? I wouldn't have guessed that that meant I was blacklisted.

I am sure Richard or Jord among others could say why it is -1 rather than 0 which to my mind would make more sense. Also worth noting when it says /cpu it really means /core so a machine that has the number set to 0 could in some instances be trashing 128 tasks/day. When the majority of machines had only one or two cores it was in some ways less of an issue.