There is very little GPUGrid work at the moment anyway, so it could be much ado about nothing.

Maybe I mentioned this elsewhere, but at the risk of repetition: In comparable situations, I just do Folding on the GPU, and use BOINC for the CPU projects. They get along together fine in most cases, except when a VirtualBox project is involved, in which case Folding can crash.

You are right, GPUGRID seems to have come to a halt until some time in summer, if not even longer (recently they were talking about a new student to come on board at the beginning of June, now it's obviously a month later, and no one can tell, I guess, what it will look like in reality). It's just too bad that such a great and valuable project is down, at least for the time being.

As alternative, I have my two high-end GPUs crunch SETI@home right now.
Folding@home would also sound great, and it's something similar to GPUGRID, dealing with biomedicine. However, I would foresee problems with my settings, since I have ATLAS@home running, too; and this is using a VirtualBox.
However, I might install and try Folding on one of my other PCs with a GTX970 inside. There is no VirtualBox on this PC.
And then, there is another PC with a GTX750 inside. On this one, too, I might try Folding.
It's a pitty that Folding does not run under BOINC.

Ah OK, but if he's already having trouble connecting two BOINC Managers, he may as well have trouble connecting boinccmd to the right directory.

This might well be the case, I am afraid.
By now, my will to further experiment is somewhat exhausted, honestly speaking.

Even more since I'm having second thoughts on coninuing using of the RAMDisk for BOINC.
Twice already the system crashed (once due to electricity failure caused by a defective vacuum cleaner plug, the other time for unknown reasons), so each time I had to install BOINC from scratch.
And all these special settings and what not, in order to maybe finally get 2 BOINCs running, would have to be made over each time. I guess it simply isn't worth it.
Most likely, within the next few days I will simply attach my external USB-3 HDD and have BOINC run there. And no longer have to worry about GPUGRID requesting 7.629MB free disc space before even downloading a new task, and - what I found out the other day - also ATLAS requests 6.199MB free disc space (whereas in reality, it does not need more than about 2GB).

So, at the bottom line, running BOINC with the projects I have chosen, on a RAMDDisk with total 32GB RAM turns out not being the best idea. If I had 64GB RAM, the situation would be a different one.

In any case, many thanks to you people for all the help, hints and suggestions you brought here in this thread, for assisting me with my plan.

It's a pitty that Folding does not run under BOINC.

Yes, I would prefer that too. But they had to develop their own control program because of the way the science works. They keep an essentially zero buffer size, and start downloading a new work unit just before the last one completes (at 99% complete). That is necessary to minimize the delay in using the results of the just-completed work unit in order to generate the next one. So it increases the efficiency of their project. They considered using BOINC a few years ago, but couldn't make it fly. However, it does give us another useful option in situations like the present.

Ah OK, but if he's already having trouble connecting two BOINC Managers, he may as well have trouble connecting boinccmd to the right directory.

This might well be the case, I am afraid.

But the Manager doesn't connect to a directory - and nor does boinccmd. The communication is by TCP/IP (like a network), so it connects to a port. It's the client (the thing that the Manager is connecting to) that reads/writes/manages the directory.

So I suppose we ought to consider which user account boinc.exe is trying to run under, what account permissions have been set on the new data directory, and whether they're allowed to get it together.

I'm almost getting to the stage of being inquisitive enough to try it myself...

[quote] They considered using BOINC a few years ago, but couldn't make it fly. However, it does give us another useful option in situations like the present.

Just too bad that within the range of BOINC projects, at least from what I could see, there is no other biomedical project for GPU like GPUGRID or FoldingHome :-(

[quote] They considered using BOINC a few years ago, but couldn't make it fly. However, it does give us another useful option in situations like the present.

Just too bad that within the range of BOINC projects, at least from what I could see, there is no other biomedical project for GPU like GPUGRID or FoldingHome :-(

[quote]Ah OK, but if he's already having trouble connecting two BOINC Managers, he may as well have trouble connecting boinccmd to the right directory.

But the Manager doesn't connect to a directory - and nor does boinccmd.

I used the wrong name. I was with my head elsewhere -soft pillows, warm covers- when I wrote that. I meant 'the right BOINC instance'.

@Erich, what firewall do you use, and did you add boinc.exe and boincmgr.exe to it? If just the Windows firewall, what are the settings used?

---

@Erich, what firewall do you use, and did you add boinc.exe and boincmgr.exe to it? If just the Windows firewall, what are the settings used?

I am using Symantec Endpoint Protection. In the chapter "Network Thread Protection" I now opened "configure firewall rules" and added a rule to allow all traffic for boincmgr.exe and boinc.exe
However, this did not help :-(

OK, as I suspected, curiosity got the better of me. I created the two batch files BobCat13 suggested in the message I'm replying to.

And it failed. Watching the changes in the second data directory, I saw a file stderrdae.txt being created, with the contents

05-May-2016 15:50:59 Another instance of BOINC is running.
GLE: Cannot create a file when that file already exists.

 (0xb7)

Armed with that information, I edited the cc_config.xml file for the first (originally running) instance, with the

<allow_multiple_clients>1</allow_multiple_clients>

option, and restarted the first client. After that, everything worked as advertised. Two different clients running (two instances of boinc.exe in Task Manager), two different instances of boincmgr.exe in Task Manager as well, two different displays on screen with different project lists, etc. etc.

I currently have a dozen machines running 10 instances of BOINC. I have found that BOINC Manager does not connect to the client when more than one instances of the Manager is open. The second instance of BOINC Manager always gives me the message "Another instance of BOINC Manager is already running on this computer. Please select a client to monitor ". Individually I can connect to each BOINC client instance with no problem.

I start the BOINC clients in a single .bat like this:

boinc.exe --gui_rpc_port 31416 --dir %~dp0Data1\ --detach
boinc.exe --gui_rpc_port 31417 --dir %~dp0Data2\ --detach
boinc.exe --gui_rpc_port 31418 --dir %~dp0Data3\ --detach
boinc.exe --gui_rpc_port 31419 --dir %~dp0Data4\ --detach
boinc.exe --gui_rpc_port 31420 --dir %~dp0Data5\ --detach
boinc.exe --gui_rpc_port 31421 --dir %~dp0Data6\ --detach
boinc.exe --gui_rpc_port 31422 --dir %~dp0Data7\ --detach
boinc.exe --gui_rpc_port 31423 --dir %~dp0Data8\ --detach
boinc.exe --gui_rpc_port 31424 --dir %~dp0Data9\ --detach
boinc.exe --gui_rpc_port 31425 --dir %~dp0Data10\ --detach

Then I have a .bat to launch the Manager for each client I want to connect.
start_mgr_client1.bat

start ..\boincmgr.exe /n localhost /p password /g 31416 /m

start_mgr_client2.bat

start ..\boincmgr.exe /n localhost /p password /g 31417 /m

start_mgr_client3.bat

start ..\boincmgr.exe /n localhost /p password /g 31418 /m

I started with BOINC 7.6.22. Then I loaded 7.6.23 to see if it worked differently. The OS is Windows 7 64-bit & Windows XP 32-bit.

OK, as I suspected, curiosity got the better of me. I created the two batch files BobCat13 suggested in the message I'm replying to.

And it failed. Watching the changes in the second data directory, I saw a file stderrdae.txt being created, with the contents

05-May-2016 15:50:59 Another instance of BOINC is running.
GLE: Cannot create a file when that file already exists.

 (0xb7)

Armed with that information, I edited the cc_config.xml file for the first (originally running) instance, with the

<allow_multiple_clients>1</allow_multiple_clients>

option, and restarted the first client. After that, everything worked as advertised. Two different clients running (two instances of boinc.exe in Task Manager), two different instances of boincmgr.exe in Task Manager as well, two different displays on screen with different project lists, etc. etc.

hm, sounds interesting.
However:
- there is no cc_config.xml in the "first" BOINC (in fact, in the "second" not either). But as far as I understand, I would just need to create one.
- the file "stderrdae.txt" here is two days old, and it's empty. No text at all.
So, the behaviour here seems to be different to what you got while testing.

I guess I now just create the cc_config.xml and put it into the first BOINC, right?

OK, as I suspected, curiosity got the better of me. I created the two batch files BobCat13 suggested in the message I'm replying to.

And it failed. Watching the changes in the second data directory, I saw a file stderrdae.txt being created, with the contents

05-May-2016 15:50:59 Another instance of BOINC is running.
GLE: Cannot create a file when that file already exists.

 (0xb7)

Armed with that information, I edited the cc_config.xml file for the first (originally running) instance, with the

<allow_multiple_clients>1</allow_multiple_clients>

option, and restarted the first client. After that, everything worked as advertised. Two different clients running (two instances of boinc.exe in Task Manager), two different instances of boincmgr.exe in Task Manager as well, two different displays on screen with different project lists, etc. etc.

hm, sounds interesting.
However:
- there is no cc_config.xml in the "first" BOINC (in fact, in the "second" not either). But as far as I understand, I would just need to create one.
- the file "stderrdae.txt" here is two days old, and it's empty. No text at all.
So, the behaviour here seems to be different to what you got while testing.

I guess I now just create the cc_config.xml and put it into the first BOINC, right?

I found everything works better if each BOINC instance has a cc_config.xml with <allow_multiple_clients>. Then you will not have issues if you stop and start the instances in a different order.

I now created this cc_config.xml and put it into both BOINCs. Then I restarted the first one, then the second one.
Same failure as before: on the right-hand lower corner (in kind of a "status bar"), the two notices "connecting to 127.0.01" and "disconnected" change in intervals of a few seconds.

Just to be on the safe side, as I am surely not a specialist in creating xml-files. The one I did looks like this:

<cc_config>
<allow_multiple_clients>1</allow_multiple_clients>
</cc_config>

correct?

One one hand, it seems to me that I am just a tiny step away from succeeding, on the other hand, I suspect that for some unknown reason, it simply cannot made work here :-(

So, with the trial period of my RAMDisk ending in a few days, I'll give up the idea of having two BIONCs (one on the RAMDisc, the other one on the internal SSD). I will install it on an external USB3-SSD.

P.S. Unfortunately, the cc_config-xml text structure is not shown correctly here. In the second line, the arrow is placed exactly unter the underbar of the text in the first line.

I now created this cc_config.xml and put it into both BOINCs. Then I restarted the first one, then the second one.
Same failure as before: on the right-hand lower corner (in kind of a "status bar"), the two notices "connecting to 127.0.01" and "disconnected" change in intervals of a few seconds.

Just to be on the safe side, as I am surely not a specialist in creating xml-files. The one I did looks like this:

<cc_config>
<allow_multiple_clients>1</allow_multiple_clients>
</cc_config>

correct?

One one hand, it seems to me that I am just a tiny step away from succeeding, on the other hand, I suspect that for some unknown reason, it simply cannot made work here :-(

So, with the trial period of my RAMDisk ending in a few days, I'll give up the idea of having two BIONCs (one on the RAMDisc, the other one on the internal SSD). I will install it on an external USB3-SSD.

You are very close, but your cc_config.xml is not formatted correctly. See http://boinc.berkeley.edu/wiki/Client_configuration
You are missing some tags.

You are very close, but your cc_config.xml is not formatted correctly. See http://boinc.berkeley.edu/wiki/Client_configuration
You are missing some tags.

okay, I guess I found out what you talk about. It now looks like this:

<cc_config>
<options>
<allow_multiple_clients>1</allow_multiple_clients>
</options>
</cc_config>

(again, the various lines are moved to the right, for some reason this cannot be reproduced here in the forum).

However, this did not help, either. The failure must be somewhere else, and I doubt that I'll find out.

You are very close, but your cc_config.xml is not formatted correctly. See http://boinc.berkeley.edu/wiki/Client_configuration
You are missing some tags.

okay, I guess I found out what you talk about. It now looks like this:

<cc_config>
<options>
<allow_multiple_clients>1</allow_multiple_clients>
</options>
</cc_config>

(again, the various lines are moved to the right, for some reason this cannot be reproduced here in the forum).

However, this did not help, either. The failure must be somewhere else, and I doubt that I'll find out.

Yes the <options> tags were missing.

I don't think BOINC cares about the layout in the cc_config.xml

I think this
<cc_config>
<options>
<allow_multiple_clients>1</allow_multiple_clients>
</options>
</cc_config>
or this

<cc_config>
	<options>
		<allow_multiple_clients>1</allow_multiple_clients>
	</options>
</cc_config>

would be the same.

If the BOINC Manager isn't connecting to the client you can check the stdoutdae.txt in each data directory to make sure each instances started as expected. Something you might want to give a try is either a remote_hosts.cfg or using the <allow_remote_gui_rpc> option in your cc_config.xml.
The <allow_remote_gui_rpc> options allows any BOINC Manger to connect to the client, if the have the right password, and using a remote_hosts.cfg only allows the IPs of the hosts in the file to connect to the client. The <allow_remote_gui_rpc> would be a quick way to test to see if there was a permission issue.

I got interrupted by a phone call as I was performing my test, and had to go out for a bit, but I should probably amplify a bit.

Machine is Windows 7/64
BOINC was initially installed in user mode (primary instance runs GPUs). Executable files have been updated with self-built versions from head, as at early March this year. I'm pretty sure that doesn't affect the multiple clients issue - no change in that area from 7.6.22/.23/.29/.32 or wherever we're up to now.
Both program and data directories are located on a clean data drive: neither are in the protected C:\Program Files\ or ProgramData areas.

Having tested that the second Manager instance runs properly when the second client is running, I've tried without launching the client first. Nada - tried (and failed) to connect, no error message. In particular, no message about a bad password. Because this is on a secure private network, I have used BOINC Manager itself, and BoincView, to manage machines remotely for many years with an eight-letter password which is an anagram of wssrpoda - makes it easy to avoid errors in that department.

If you're using a modern client, like the recommended v7.6.22, the easiest way to update configurations is to use the Event Log Options tool (Ctrl+Shift+F) - make a trivial change and click 'OK'. That creates a template cc_config.xml file with every possible tag present, and all spelled correctly :-) Then just modify 0 to 1, or whatever, for the tag you need. Did I say I restarted the primary client after making that change? It feels like one which would only take effect at startup, rather than through 'read config files' like some others.

could you please tell me what exactly the cc_config.xml would look like then?
I now have as follows:

<cc_config>
<options>
<allow_multiple_clients>1</allow_multiple_clients>
<allow_remote_gui_rpc>
</options>
</cc_config>

When I open the BOINC manager, I get notices like
Unrecognized tag in cc_config.xml: <allow_remote_gui_rpc>
and
Error in cc_config.xml options
and
Missing end tag in cc_config.xml

If you're using a modern client, like the recommended v7.6.22, the easiest way to update configurations is to use the Event Log Options tool (Ctrl+Shift+F) - make a trivial change and click 'OK'. That creates a template cc_config.xml file with every possible tag present, and all spelled correctly :-) Then just modify 0 to 1, or whatever, for the tag you need. Did I say I restarted the primary client after making that change? It feels like one which would only take effect at startup, rather than through 'read config files' like some others.

sorry, I read your post too late. When I do Ctrl+Shift+F, a window called "BOINC Diagnostic Log Flags" opens, rather than what you were describing.
Concerning making changes: in fact, each time I made these in the cc_config.xml, I had closed down BOINC before and then restarted it.

When I do Ctrl+Shift+F, a window called "BOINC Diagnostic Log Flags" opens.

That's the one. Make a change, 'OK' it. Go and have a look at cc_config.xml

It should contain everything described in the first two sections of Client configuration - Logging flags and Options.