Right, /sys shouldn't be backed-up. :-) So, why did 1.1.16 pass over this problem, and 1.2.2 choke? In 1.1.16, it failed to read any information about the files in /sys, and thus skipped over it. By 1.2.2, some bugs had been fixed to make rdiff-backup better able to read information about troublesome files (Unicode files, for example, often present the same error codes as the /sys files give when reading the file metadata).
The strange thing is, when you try to get information about files in /sys, you get an "Invalid file" (EINVAL) error code, which rdiff-backup has long been able to handle. But when you try to actually read some of the files in /sys, then you get a "Connection timed out" (ETIMEDOUT) error, which rdiff-backup didn't know about. Based on your bug report, I added handling for the ETIMEDOUT error to CVS the other day. Rdiff- backup 1.2.3 will be able to pass through the /sys directory, just as 1.1.16 did. So, the correct solution is to follow the good practice of not backing- up /sys (just like /proc or /dev/shm, etc.), but rdiff-backup will no longer be tripped-up starting again with 1.2.3. Thanks, Andrew -- rdiff-backup crashed with IOError in read() https://bugs.launchpad.net/bugs/304659 You received this bug notification because you are a member of Ubuntu Bugs, which is subscribed to Ubuntu. -- ubuntu-bugs mailing list ubuntu-bugs@lists.ubuntu.com https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs
