Giving up on download of cufft_xp32_32_16.dll: permanent HTTP error
27 Feb 2015 13:27:13 UTC
Topic 197993
(moderation:
Hello,
when trying to put one of my computers back onto processing Einstein/Nvidia tasks, which it hadn't been doing for a while, I got this:
Quote:
27.02.2015 13:39:44 | Einstein@Home | Requesting new tasks for NVIDIA GPU
27.02.2015 13:39:46 | Einstein@Home | Scheduler request completed: got 3 new tasks
27.02.2015 13:39:46 | Einstein@Home | BOINC will delete file rand_PMPSXT.bank (no longer needed)
27.02.2015 13:39:48 | Einstein@Home | Started download of cufft_xp32_32_16.dll
27.02.2015 13:39:48 | Einstein@Home | Started download of einsteinbinary_BRP4_1.00_graphics_windows_intelx86.exe
27.02.2015 13:39:49 | Einstein@Home | Giving up on download of cufft_xp32_32_16.dll: permanent HTTP error
27.02.2015 13:39:49 | Einstein@Home | Giving up on download of einsteinbinary_BRP4_1.00_graphics_windows_intelx86.exe: permanent HTTP error
27.02.2015 13:39:49 | Einstein@Home | Started download of einsteinbinary_BRP6_1.39_windows_intelx86__BRP5-cuda32-nv301.exe
27.02.2015 13:39:49 | Einstein@Home | Started download of einsteinbinary_BRP6_1.39_windows_intelx86__BRP5-cuda32-nv301.exe-db.dev
27.02.2015 13:39:50 | Einstein@Home | Finished download of einsteinbinary_BRP6_1.39_windows_intelx86__BRP5-cuda32-nv301.exe-db.dev
27.02.2015 13:39:50 | Einstein@Home | Started download of einsteinbinary_BRP6_1.39_windows_intelx86__BRP5-cuda32-nv301.exe-dbhs.dev
27.02.2015 13:39:51 | Einstein@Home | Finished download of einsteinbinary_BRP6_1.39_windows_intelx86__BRP5-cuda32-nv301.exe-dbhs.dev
27.02.2015 13:39:51 | Einstein@Home | Started download of PM0004_010C1_188.bin4
27.02.2015 13:39:54 | Einstein@Home | Finished download of PM0004_010C1_188.bin4
(just an excerpt, the hole list is longer). The other files seem to download ok, however the tasks as such each fail with an error on download. I guess they all miss the fft library. The host venue is set to accept nvidia tasks only, and not to run beta test applications.
Regards
Giving up on download of cufft_xp32_32_16.dll: permanent HTTP er
)
I have the same after clearing BOINC out of my system due to instant bug (here http://einsteinathome.org/node/197977), that I wasn't able to resolve myself. I have made a clear installation in a different BOINC and BOINC DATA folders, not default ones.
Besides this it errors out on BRP4 download with
27.02.2015 21:12:45 | Einstein@Home | Giving up on download of einsteinbinary_BRP4_1.00_graphics_windows_intelx86.exe: permanent HTTP error
Same here: 27.02.2015
)
Same here:
27.02.2015 17:49:23 | Einstein@Home | Requesting new tasks for NVIDIA GPU
27.02.2015 17:49:24 | Einstein@Home | Scheduler request completed: got 1 new tasks
27.02.2015 17:49:26 | Einstein@Home | Started download of cufft_xp32_32_16.dll
27.02.2015 17:49:26 | Einstein@Home | Started download of einsteinbinary_BRP4_1.00_graphics_windows_intelx86.exe
27.02.2015 17:49:27 | Einstein@Home | Giving up on download of cufft_xp32_32_16.dll: permanent HTTP error
27.02.2015 17:49:27 | Einstein@Home | Giving up on download of einsteinbinary_BRP4_1.00_graphics_windows_intelx86.exe: permanent HTTP error
27.02.2015 17:49:27 | Einstein@Home | Started download of PM0004_00831_126.bin4
27.02.2015 17:49:27 | Einstein@Home | Started download of PM0004_00831_127.bin4
This pc:
27.02.2015 15:59:34 | | Starting BOINC client version 7.4.36 for windows_x86_64
27.02.2015 15:59:34 | | log flags: file_xfer, sched_ops, task
27.02.2015 15:59:34 | | Libraries: libcurl/7.39.0 OpenSSL/1.0.1j zlib/1.2.8
27.02.2015 15:59:34 | | Data directory: C:\ProgramData\BOINC
27.02.2015 15:59:34 | | Running under account Alexander
27.02.2015 15:59:34 | | CUDA: NVIDIA GPU 0: GeForce GT 630 (driver version 347.52, CUDA version 7.0, compute capability 3.5, 1024MB, 904MB available, 692 GFLOPS peak)
27.02.2015 15:59:34 | | OpenCL: NVIDIA GPU 0: GeForce GT 630 (driver version 347.52, device version OpenCL 1.1 CUDA, 1024MB, 904MB available, 692 GFLOPS peak)
27.02.2015 15:59:34 | | Host name: WIN-0B7RG7VGK9S
27.02.2015 15:59:34 | | Processor: 4 GenuineIntel Intel(R) Core(TM) i3-2120 CPU @ 3.30GHz [Family 6 Model 42 Stepping 7]
27.02.2015 15:59:34 | | Processor features: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss htt tm pni ssse3 cx16 sse4_1 sse4_2 popcnt syscall nx lm avx vmx tm2 pbe
27.02.2015 15:59:34 | | OS: Microsoft : x64 Edition, (10.00.9926.00)
27.02.2015 15:59:34 | | Memory: 3.94 GB physical, 4.63 GB virtual
Edit: I attached this pc just 30 min before to Einstein
Thanks Stranger, so here is
)
Thanks Stranger, so here is one confirmation post at last. Btw, my host in question is ID: 11701432.
I just did an experiment with another host that was happily crunching einstein nvidia tasks of all sorts so far, ID: 11758367. After deinstalling BOINC, clearing the BOINC data directory entirely, reinstalling and reattaching it is now unable to download the fft library (and the other file I don't know what that is good for). The respective tasks error out as described before.
So I think it is currently impossible to run nvidia crunchers that did not successfully download the failing files successfully before, which means it may not be possible to attach new nvidia hosts to einstein at all.
Regards
Rewording: My guess is, that
)
Rewording: My guess is, that newly attached hosts will not be able to crunch nvidia tasks. That's what I mean. Maybe somebody else can confirm or deny this, to raise the attention level this gets or solve it in any way.
Cheers
I've downloaded these files
)
I've downloaded these files from another host while BOINC was stopped. But BOINC says about HTTP error again and these 2 files become zero sized.
RE: Rewording: My guess is,
)
I can confirm that one of my pc's, attached today, gets these http download errors. http://einsteinathome.org/host/11758363/tasks
Another pc, attached months ago, gets work, but fails immediate with error 1001
http://einsteinathome.org/task/486885252 and empty stderr. Outcome: Client error, Client state: compute error.
Alexander
RE: Another pc, attached
)
About the error 1001 and the empty stderr please see the adjacent thread about trouble with BRP6-beta...
The file named in the thread
)
The file named in the thread title, cufft_xp32_32_16.dll, seems to download OK from
http://einstein-dl2.phys.uwm.edu/download/cufft_xp32_32_16.dll
(the University of Wisconsin at Madison mirror), but fails with a 404 Not Found error when downloaded from the AEI at Hannover.
RE: The file named in the
)
I've downloaded it from there, put it into the Einstein project directory (replacing an existing file), but still get the http error on both cufft and einsteinbinary.
The project directory contains three einsteinbinary ... BRP6-Beta-cuda32-nv301 files, but different extensions: one .exe, one .exe-db.dev and one .exe-dbhs.dev
So maybe the download itself works, but the recognition fails.
RE: I've downloaded these
)
I tried the same and failed the same. I think one cannot just put the file where it belongs because BOINC remembers the download failure and tries to re-download, thus overwriting it again and creating the 0-sized file because download fails again.
Richard: I took the download urls for the cufft-library from client_state.xml and one after the other put them in a web browser to try. I think that is the way BOINC itself is supposed to handle this, but I do not really know.
Hannover is a straight failure, as you said, and UWM took ages to establish a connection to at all. It then downloaded, but I think inside BOINC this will time out before.
I still think the action to be taken is on the server side, unless told better.
Regards