These are the very sorts of things that might get a bit mangled if remote desktop software somehow mishandles things.
Remote Desktop sessions immediately remove the GPU (don't know if all type of GPUs, but surely Intel iGPUs) from the BOINC client.
All GPUs.
It's a Windows driver security 'feature': the computer addressed remotely over RDP switches from using the vendor's user-mode driver to using a Microsoft system-mode driver. BOINC, running in user mode, is not allowed to interact with the system-mode driver.
Other remote access protocols exist and don't have this limitation, but with RDP you're stuck.
(Edit: it's the same reasoning which prevents a service-mode BOINC installation from interacting with the vendor drivers)
Judging tone in text message is notoriously difficult so let me preface by saying that my entire post possibly came across more serious than it was intended. I hoped my mentions of the an omnipotent Validator entity looming over us all would indicate that I do not take this all too seriously. It's a fun little hobby. But if it doesn't work *for me* than can just spend my time and resourced in another way, no harm no foul.
The rig that causes the issues is built specifically to "just run". I don't want to spend too much time troubleshooting it, i.e. checking if the tasks are returned properly or what I was doing/what the ambient temperature was while the errors occur. Why do I insist on checking on it remotely from my main PC? Bc I like it. Plain as that. I do this whole endeavour for fun, and if I can't have fun with it the way I want, well, then that is my problem and I just do smth else.
It is just annoying to discover that your PC ran for 16 hours and the Validator just throws your hard work away, for *seemingly* no reason. I know that there must be some factual reason for the Validator to act this way, I doubt Skynet has spread this far already to make judgements based on other factors, but if it doesn't let me know what I can do to fix it in a clear manner, then, well, that is it for me.
I am not interested in troubleshooting when I'm this much in the dark. Plus I maintain, that IF a simple remote access breaks things, that is stupid. Maybe my fault for insisting to bother the BOINC Client with my BS remotely, but that is the way I like it and want to do it.
I hope I didn't come across as too pissed or angry bc I am not. all good ✌
FWIW, I tried Linux first, bc for a computing rig that is the obvious choice, but BOINC didn't work out of the box and with windows it was just so much simpler, it must be said....
Ah. I only saw this after I posted my other response xD that sounds intriguing and might be the culprit here. I do not use Windows Remote Desktop directly, but it is possible that the BOINC Client on my main rig uses that infrastructure to communicate with my computing rig. BUT, why did it work then for a day or two, when I also kept pestering it remotely? Ah, so many questions....
Well, the solution seems to be to give Linux another shot even if I have to compile the shidd myself :xD
Or I do bite the bullet and meticulously log all result files sent and received to gain insights.... Starfield might interfere with that tho ngl.
[...] I do not use Windows Remote Desktop directly, but it is possible that the BOINC Client on my main rig uses that infrastructure to communicate with my computing rig. [...]
No, BOINC manager definitely doesn't use RDP to connect to remote BOINC clients.
Out of the 441 Tasks I've been assigned, 292 Have resulted in an error. Typically, its been the "All-Sky Gravitational Wave search on O3 v1.07 () windows_x86_64" application that fails. I actually have not once submitted a valid WU under this application. It fails within 2 seconds of starting the WU. No other project or application within those projects constantly fails like this, and I am looking for some better insight as to what is going wrong. Ive looked through this thread and it none of it seems to really apply to my specific situation where it almost immediately fails upon starting the task.
If you could kindly go to your account page and unhide your computer(s) that would help a great deal. Go via the menu path Account->Preferences->Privacy and click on YES for 'Should Einstein@Home show your computers on its website?' That way we can inspect the task logs etc and perhaps advise.
BTW if it fails within a few seconds of starting then it won't be a validate error, those occur with "... an otherwise successfully completed result that the validator daemon is immediately recognising as having something wrong without even needing to compare with any other result in the workunit quorum."
Scrooge McDuck wrote:Gary
)
All GPUs.
It's a Windows driver security 'feature': the computer addressed remotely over RDP switches from using the vendor's user-mode driver to using a Microsoft system-mode driver. BOINC, running in user mode, is not allowed to interact with the system-mode driver.
Other remote access protocols exist and don't have this limitation, but with RDP you're stuck.
(Edit: it's the same reasoning which prevents a service-mode BOINC installation from interacting with the vendor drivers)
Judging tone in text message
)
Judging tone in text message is notoriously difficult so let me preface by saying that my entire post possibly came across more serious than it was intended. I hoped my mentions of the an omnipotent Validator entity looming over us all would indicate that I do not take this all too seriously. It's a fun little hobby. But if it doesn't work *for me* than can just spend my time and resourced in another way, no harm no foul.
The rig that causes the issues is built specifically to "just run". I don't want to spend too much time troubleshooting it, i.e. checking if the tasks are returned properly or what I was doing/what the ambient temperature was while the errors occur. Why do I insist on checking on it remotely from my main PC? Bc I like it. Plain as that. I do this whole endeavour for fun, and if I can't have fun with it the way I want, well, then that is my problem and I just do smth else.
It is just annoying to discover that your PC ran for 16 hours and the Validator just throws your hard work away, for *seemingly* no reason. I know that there must be some factual reason for the Validator to act this way, I doubt Skynet has spread this far already to make judgements based on other factors, but if it doesn't let me know what I can do to fix it in a clear manner, then, well, that is it for me.
I am not interested in troubleshooting when I'm this much in the dark. Plus I maintain, that IF a simple remote access breaks things, that is stupid. Maybe my fault for insisting to bother the BOINC Client with my BS remotely, but that is the way I like it and want to do it.
I hope I didn't come across as too pissed or angry bc I am not. all good ✌
FWIW, I tried Linux first, bc for a computing rig that is the obvious choice, but BOINC didn't work out of the box and with windows it was just so much simpler, it must be said....
Ah. I only saw this after I
)
Ah. I only saw this after I posted my other response xD that sounds intriguing and might be the culprit here. I do not use Windows Remote Desktop directly, but it is possible that the BOINC Client on my main rig uses that infrastructure to communicate with my computing rig. BUT, why did it work then for a day or two, when I also kept pestering it remotely? Ah, so many questions....
Well, the solution seems to be to give Linux another shot even if I have to compile the shidd myself :xD
Or I do bite the bullet and meticulously log all result files sent and received to gain insights.... Starfield might interfere with that tho ngl.
Hello Beercules48, What
)
Hello Beercules48,
What distro of Linux are you trying to run?
What 'version' of BOINC for Linux are you trying to run?
BOINC works very well for me in Linux Ubuntu 22.04 LTS.
Proud member of the Old Farts Association
Beercules48 schrieb:[...] I
)
No, BOINC manager definitely doesn't use RDP to connect to remote BOINC clients.
Out of the 441 Tasks I've
)
Out of the 441 Tasks I've been assigned, 292 Have resulted in an error. Typically, its been the "All-Sky Gravitational Wave search on O3 v1.07 () windows_x86_64" application that fails. I actually have not once submitted a valid WU under this application. It fails within 2 seconds of starting the WU. No other project or application within those projects constantly fails like this, and I am looking for some better insight as to what is going wrong. Ive looked through this thread and it none of it seems to really apply to my specific situation where it almost immediately fails upon starting the task.
If you could kindly go to
)
If you could kindly go to your account page and unhide your computer(s) that would help a great deal. Go via the menu path Account->Preferences->Privacy and click on YES for 'Should Einstein@Home show your computers on its website?' That way we can inspect the task logs etc and perhaps advise.
BTW if it fails within a few seconds of starting then it won't be a validate error, those occur with "... an otherwise successfully completed result that the validator daemon is immediately recognising as having something wrong without even needing to compare with any other result in the workunit quorum."
Also you probably will obtain better insight by reading and posting in the more recent thread NEW IMPROVED GRAVATIONAL WAVE APP - DISCUSSION in Cruncher's Corner.
Cheers, Mike.
I have made this letter longer than usual because I lack the time to make it shorter ...
... and my other CPU is a Ryzen 5950X :-) Blaise Pascal
Ok, unhid PC, and Ill try
)
Ok, unhid PC, and Ill try posting there. Thanks.
Ellesmere have a lot of
)
Ellesmere have a lot of invalid E@H GPU-Tasks.
(AMD Radeon pro WX7100).
Trying to report completed
)
Trying to report completed work but I am getting:
7/9/2024 11:08:14 AM | Einstein@Home | [error] No scheduler URLs found in master file
What does this mean?