Message boards :
Number crunching :
Mobile Broadband Compression Corrupting Data
Message board moderation
Author | Message |
---|---|
BMH Send message Joined: 27 May 99 Posts: 419 Credit: 166,294,083 RAC: 125 |
Hi all, I've been participating in SETI@home for a number of years now. I have always had to take my work units into work to upload/download as I didn't have internet access at home. I've recently signed up to mobile broadband (with T-mobile) in the UK and was looking forward to running BOINC without having to mess about with taking work units into work etc. but BOINC fails to successfully download/upload data and reports errors relating to checksum errors and the like. The problem it seems is because T-mobile compress the data. This is obvious when viewing web pages because images are heavily compressed. Their software includes a broadband accelerator which I have disabled with no effect. It is possible to view web pages uncompressed by refreshing the page (F5) whilst holding Ctrl but I can't tell BOINC to do this! I have e-mailed T-mobile and await a reply. Can anyone here offer some suggestions or has anyone else had a similar experience. Thanks! Brian. |
Richard Haselgrove Send message Joined: 4 Jul 99 Posts: 14654 Credit: 200,643,578 RAC: 874 |
Which files are you seeing this on? ISPs should only compress image files, where the loss of quality wouldn't be noticable to the human eye on a small mobile device screen. If SETI is objecting to changes in the small JPEG files used by the screensaver, check the "Skip image file verification?" option on your general computing preferences page. But if T-Mobi;e is messing with data files, report them to Ofcom. |
BMH Send message Joined: 27 May 99 Posts: 419 Credit: 166,294,083 RAC: 125 |
Tried the "Skip image verification" online in my account and within the BOINC application with no change. It's compressing images and any work units it downloads it fails and it 'finishes' them straight away and uploads them. If I leave BOINC running for a few minuites it reaches its daily quota of work units. Brian. |
BMH Send message Joined: 27 May 99 Posts: 419 Credit: 166,294,083 RAC: 125 |
Here's what keeps happening when I start up BOINC: 29-May-2010 08:03:11 [SETI@home] Sending scheduler request: To fetch work. 29-May-2010 08:03:11 [SETI@home] Requesting new tasks 29-May-2010 08:03:16 [SETI@home] Scheduler request completed: got 2 new tasks 29-May-2010 08:03:18 [SETI@home] Started download of seti_603.jpg 29-May-2010 08:03:18 [SETI@home] Started download of 28no06ai.6076.32392.11.10.19 29-May-2010 08:03:20 [SETI@home] Incomplete read of 4250.000000 < 5KB for seti_603.jpg - truncating 29-May-2010 08:03:20 [SETI@home] Finished download of seti_603.jpg 29-May-2010 08:03:20 [SETI@home] Started download of 28no06ai.28608.32392.9.10.48 29-May-2010 08:03:20 [SETI@home] [error] File seti_603.jpg has wrong size: expected 9068, got 0 29-May-2010 08:03:20 [SETI@home] [error] Checksum or signature error for seti_603.jpg 29-May-2010 08:03:23 [SETI@home] Finished download of 28no06ai.6076.32392.11.10.19 29-May-2010 08:03:26 [SETI@home] Finished download of 28no06ai.28608.32392.9.10.48 But under Tasks it states for each work "Download Failed", and the log continues: 29-May-2010 08:04:21 [SETI@home] Sending scheduler request: To fetch work. 29-May-2010 08:04:21 [SETI@home] Reporting 2 completed tasks, requesting new tasks And so it goes on downloading tasks and failing. Brian. |
Richard Haselgrove Send message Joined: 4 Jul 99 Posts: 14654 Credit: 200,643,578 RAC: 874 |
So it is only the JPEG? Download it at work, take it home on a USB stick and copy it into the project folder. |
Josef W. Segur Send message Joined: 30 Oct 99 Posts: 4504 Credit: 1,414,761 RAC: 0 |
Just to clarify what Richard said, BOINC has this concept that all "input" files have to be present before it can start the application to do the work. So the inability to get the .jpg file is enough to make BOINC trash the work rather than running it. The "Skip image file verification" option ought to fix your situation, but IIRC it has failed for others too. There's a copy of the image at http://users.westelcom.com/jsegur/SAHenh/seti_603.jpg in case that's easier than getting it from the project. Joe |
BMH Send message Joined: 27 May 99 Posts: 419 Credit: 166,294,083 RAC: 125 |
Thanks guys. I see also that seti_609.jpg is also present but with a size of 0 bytes so when the server is back up ready to provide work I'll try again with those in place and shall let you know the outcome. Brian. |
BMH Send message Joined: 27 May 99 Posts: 419 Credit: 166,294,083 RAC: 125 |
OK, the server is back up and I've had a chance to try again. I put the following in place (C:\Documents and Settings\All Users\Application Data\BOINC\projects\setiathome.berkeley.edu): seti_603.jpg seti_609.jpg ap405.jpg Now I watched the list of messages as I fired up BOINC. It recognised that seti_603.jpg was already present and 'skipped' the download of that one. For some reason though, as I found out when I checked on the image files, seti_609.jpg had been replaced with a 0KB one and when I looked back to BOINC at the list of downloaded work units it had failed all the CUDA ones and now they're 'ready to report' straight away as before. Success with with the standard work units (and Astropulse ones) though as they're working. So as I've discovered (if I'm correct), the seti_603.jpg graphic is for the standard (CPU) work units, seti_609.jpg is for the (GPU) CUDA work units, and ap405.jpg is, well, for the Astropulse ones. I'm not sure why the seti_609.jpg got downloaded, overwriting the already present file. I've repeated the process again and it did the same, I've also made it read-only and tried quickly aborting that download from the Transfers list. But now I've reached my daily quota again. I'm also getting HTTP errors regarding temporarily failed downloads but I'm not sure if that's my problem or just the backlash from the recent server outage. Brian. |
©2024 University of California
SETI@home and Astropulse are funded by grants from the National Science Foundation, NASA, and donations from SETI@home volunteers. AstroPulse is funded in part by the NSF through grant AST-0307956.