BRP5 failures
18 Feb 2014 9:19:39 UTC
Topic 197388
(moderation:
I'm getting a lot of errors with the BRP5 app on http://einsteinathome.org/host/7192129.
Machine is Core i7-3820, 2xHD7950, Catalyst 14.1, Win7 x64
Here some examples:
Task 422448756
app_version download error: couldn't get input files:
einsteinbinary_BRP5_1.39_windows_x86_64__BRP5-opencl-ati.exe
-120 (RSA key check failed for file)
signature verification failed
Task 422540286
couldn't start app: Input file einsteinbinary_BRP5_1.39_windows_x86_64__BRP5-opencl-ati.exe missing or invalid: RSA key check failed for file
There's no antivirus nor any other such tool installed.
Any idea why ?
Are you using a firewall that
)
Are you using a firewall that might have some extra security? Comodo perhaps?
If so try to exclude Boinc and it's folders.
Otherwise you could try to download the file manually from one of the following links and save it to Boinc-data-dir\projects\einstein.phys.uwm.edu
http://einstein2.aei.uni-hannover.de/download/einsteinbinary_BRP5_1.39_windows_x86_64__BRP5-opencl-ati.exe
http://einstein.ligo.caltech.edu/download/einsteinbinary_BRP5_1.39_windows_x86_64__BRP5-opencl-ati.exe
it should also be available from http://einstein-dl4.phys.uwm.edu/download/einsteinbinary_BRP5_1.39_windows_x86_64__BRP5-opencl-ati.exe but this mirror seems to be down for the moment.
As to the cause, it could be a corruption of the file maybe due to harddrive error, corruption in the client_state.xml where the signature is stored or maybe but unlikely some hardware error that strikes when calculating the signature...
I'm not using such firewall
)
I'm not using such firewall and have been normally running BRP4 and some BRP5 tasks.
I compared my EXE with the reference files and BINGO ! Indeed, my file is missing 0x1000 bytes in the middle (0's). This is very strange, I don't know how that could have happened, probably a system or drive failure...
I'll do a project reset.
Thanks for the hint !
RE: I'll do a project
)
That should fix it!
Happy to be of service =)
I have just noticed that 4
)
I have just noticed that 4 BRP5-opencl-ati jobs have errored out with times between 162000-195000 "runtime secs". These have just suddenly showed up. I checked "task name detail" and all were "max elapsed time exceeded". Really. I also noticed that one job currently processing was 20 hours into processing so I aborted that job. This is on a Windows 7 Machine running with a Radeon card. These errors just suddenly showd up within the last few days. No hardware, driver changes etc.
Are there known issues with BRP 5 WUs? Or is there some other explanation. This is a new machine with low runtime hours on all hardware. I do need to add that it is a virtual machine utilizing pass-through. But this has not been an issue until now so I doubt that this is a VM issue.
Hi! I have experienced
)
Hi!
I have experienced something like this only once on one of my own machines and it seemed to be driver or OS related: what happened was that the graphics RAM clock speed got stuck at minimum (even tho the GPU clock was correctly fluctuating dynamically , up to the maximum) with the load on the GPU), all this according to the AMD Catalyst Control Center.
I rebooted and everything was fine again.
Cheers
HBE
I just noticed another GPU
)
I just noticed another GPU unit going into hour 20. I rebooted the win box and now that job is showing 4 hours elapsed with 00:25:xx remaining. Think I will download and install the latest drivers from RADEON. Quite strange.
Problem is back again.
)
Problem is back again. Further BRP5 tasks failed, one of them because:
1. I'm wondering why BRP5 checks the BRP4 exe.
2. That file is indeed damaged - a similar hole in the file as the BRP5 before. This happened after a project reset when the files have been downloaded again.
I don't know what's happening there, but I have no other issues - running MW, WCG and T4T (via VirtualBox) there...
Maybe the VirtualBox?
RE: Problem is back again.
)
It checks the file because it's the graphics program and if you select a running BRP5 job in Boinc and click on "Show graphics" that's the file that's used to show it. It's also the screen saver for those using that.
einsteinbinary_BRP4_1.00_graphics_windows_intelx86.exe
As to what's happening I haven't a clue but I would check the disk drive for errors or degradation.
RE: As to what's happening
)
I did both and no errors found...
I am continuing to have
)
I am continuing to have errors on this node running win 7 and an AMD R9 200 series GPU. Extremely long processing times culminating in "Errors while computing. I have upgraded the drivers hoping to fix this problem but does not seem to make a difference. I have removed E@H project using BOINC manager. In the c:\ProgramData\BOINC\projects directory E&G was removed, however, in the "slots" directories 0-3 there are still references to old PA*.bin files. Should these directories also be sanitized to avoid any confusion when re installing E@H? I am going to:
remove E@H project
remove BOINC (this does not clean up the ProgramData\BOINC directory)
and delete c:\ProgramData\BOINC
reboot the win box
reinstall BOINC from a fresh download
and reinstall E@H.
start crunching again.