Hello again, The fix doesn't work. Probably because write() ignores any seek done, if the file was opened with "append" flag. My temporary solution: Before seeking, I close the file and open it again without append, I think that it works (well, at least I haven't seen any of the corrupted results yet).
TJM http://www.enigmaathome.net On Tue, 15 Feb 2011 15:44:00 -0800 David Anderson <[email protected]> wrote: >I checked in a possible fix. >-- David > >On 15-Feb-2011 9:44 AM, Slawomir Rzeznicki wrote: >> Hello, >> >> I have a reason to believe, that there might be a bug in >>the file upload >> handler, perhaps related to partial uploads / resuming >>uploads. >> >> Recently I got a few reports about workunits thrown out >>randomly by my >> validator. Yesterday I started tracking the problem and >>I quickly found out, >> that something is wrong with the uploaded files. Every >>now and then, the file >> is "received" with it's contents multiplied inside, >>usually it looks like the >> same content is appended 2 to 6 times at the end of the >>file. >> >> At first I thought that these files come from faulty >>app, until I checked >> xml_doc_out for affected results. It clearly says, that >>the file uploaded had >> a certain length, while the same file in the upload dir >>is 2-6 times bigger. >> >> <file_info> <name>pldrv59-1_0_13699516_570_1_0</name> >> <nbytes>104.000000</nbytes> >><max_nbytes>1000.000000</max_nbytes> >> <md5_cksum>b5f823edbc321d27020d0a73745b40a4</md5_cksum> >> <url>http://www.enigmaathome.net/file_upload_handler</url> >></file_info> >> >> >> So then I checked the FUH's log just to find something >>like that: >> >> 2011-02-15 18:25:56.7653 [PID=22275] [debug] >> handle_request:<data_server_request> 2011-02-15 >>18:25:56.7654 [PID=22275] >> [debug] >> handle_request:<core_client_major_version>6</core_client_major_version> >> 2011-02-15 18:25:56.7654 [PID=22275] [debug] >> handle_request:<core_client_minor_version>10</core_client_minor_version> >> 2011-02-15 18:25:56.7654 [PID=22275] [debug] >> handle_request:<core_client_release>58</core_client_release> >>2011-02-15 >> 18:25:56.7654 [PID=22275] [debug] >>handle_request:<file_upload> 2011-02-15 >> 18:25:56.7657 [PID=22275] Starting upload of >>pldrv59-1_0_13699516_570_1_0 >> from 69.12.216.209 [offset=0, nbytes=104] 2011-02-15 >>18:25:56.7659 >> [PID=22275] Ended upload of >>pldrv59-1_0_13699516_570_1_0 from 69.12.216.209; >> retval 0 2011-02-15 18:25:56.7659 [PID=22275] [debug] >>elapsed time 0.000755 >> seconds >> >> and a dozen of megabytes below, again: >> >> 2011-02-15 18:32:03.5711 [PID=24070] [debug] >> handle_request:<data_server_request> 2011-02-15 >>18:32:03.5711 [PID=24070] >> [debug] >> handle_request:<core_client_major_version>6</core_client_major_version> >> 2011-02-15 18:32:03.5711 [PID=24070] [debug] >> handle_request:<core_client_minor_version>10</core_client_minor_version> >> 2011-02-15 18:32:03.5711 [PID=24070] [debug] >> handle_request:<core_client_release>58</core_client_release> >>2011-02-15 >> 18:32:03.5712 [PID=24070] [debug] >>handle_request:<file_upload> 2011-02-15 >> 18:32:03.5715 [PID=24070] Starting upload of >>pldrv59-1_0_13699516_570_1_0 >> from 69.12.216.209 [offset=0, nbytes=104] 2011-02-15 >>18:32:03.5715 >> [PID=24070] [CRITICAL] file pldrv59-1_0_13699516_570_1_0 >>length on disk 104 >> bytes; host upload starting at 0 bytes. 2011-02-15 >>18:32:03.5715 [PID=24070] >> Ended upload of pldrv59-1_0_13699516_570_1_0 from >>69.12.216.209; retval 0 >> 2011-02-15 18:32:03.5716 [PID=24070] [debug] elapsed >>time 0.000622 seconds >> >> Why was the file uploaded again and attached to the end >>of already existing >> file ? Is it a bug ? I'm running revision 22566. >>Dnetc.net's project admin >> also confirmed the same problem on his server, v22488. >> >> The problem is not related to any specific client >>version. >> >> >> >> TJM >> >> http://www.enigmaathome.net >> >> >> >> >> >> >> >> >> _______________________________________________ >>boinc_dev mailing list >> [email protected] >> http://lists.ssl.berkeley.edu/mailman/listinfo/boinc_dev >>To unsubscribe, >> visit the above URL and (near bottom of page) enter your >>email address. >_______________________________________________ >boinc_dev mailing list >[email protected] >http://lists.ssl.berkeley.edu/mailman/listinfo/boinc_dev >To unsubscribe, visit the above URL and >(near bottom of page) enter your email address. _______________________________________________ boinc_dev mailing list [email protected] http://lists.ssl.berkeley.edu/mailman/listinfo/boinc_dev To unsubscribe, visit the above URL and (near bottom of page) enter your email address.
