Skip to content

R
rdiff-backup
Switch branch/tag
  1. 26 Jun, 2020 1 commit
    • Jérome Perrin's avatar
      pool: disable parallelism · 176f5f29
      Jérome Perrin authored 
      This seem to have race conditions and not ready for production.

We observed some IndexError caused in

    Traceback (most recent call last):
      File "python2.7/lib/python2.7/threading.py", line 801, in __bootstrap_inner
        self.run()
      File "python2.7/lib/python2.7/threading.py", line 754, in run
        self.__target(*self.__args, **self.__kwargs)
      File "rdiff_backup/pool.py", line 47, in worker
        value = task.func(*task.args)
      File "rdiff_backup/restore.py", line 65, in get_diff
        diff = MirrorStruct.get_diff(mir_rorp, target_rorp)
      File "rdiff_backup/restore.py", line 292, in get_diff
        file_fp = cls.rf_cache.get_fp(expanded_index, mir_rorp)
      File "rdiff_backup/restore.py", line 392, in get_fp
        rf = longname.update_rf(self.get_rf(index, mir_rorp), mir_rorp,
      File "rdiff_backup/restore.py", line 388, in get_rf
        else: del self.rf_list[0]
    
apparently because the underlying code is not thread safe in the first
place.
      176f5f29
  3. 26 Oct, 2018 1 commit
  5. 25 Oct, 2018 1 commit
    • Guillaume Hervier's avatar
      Exit stalled requests on connection error. · 2cfa90a4
      Guillaume Hervier authored 
      If the SSH connection stalled or just existed during a restore, the whole process would stall as it waits for requests which will never finish.

This MR makes the connection propagate the `ConnectionQuit` exception to stalled requests when the connection is cut in the middle of the process, so they will terminate with an exception and raise it, exiting the whole program instead of making it stall

/reviewed-on !6
      2cfa90a4
  7. 30 Aug, 2018 6 commits
  9. 10 Aug, 2018 2 commits
  11. 17 Jul, 2018 1 commit
  13. 10 Jul, 2018 2 commits
  15. 13 Oct, 2014 1 commit
    • Marco Mariani's avatar
      Added --always-snapshot; version bump · c396d831
      Marco Mariani authored 
      The --always-snapshot path/to/file parameter can be repeated
any number of times, and must specify the relative path to each file
(not directory) that results corrupted after running --verify.
These files will be completely transferred again, even if they have
not changed since the last backup.

For instance:

CORRUPTED_MSG="Warning:\ Computed\ SHA1\ digest\ of\ "
CORRUPTED_ARGS=$(./rdiff-backup --verify user@host::/path/to/backup 2>&1 >/dev/null | grep "^$CORRUPTED_MSG" | sed "s/$CORRUPTED_MSG/--always-snapshot\ /g")
echo $CORRUPTED_ARGS
      c396d831
  17. 26 Sep, 2014 4 commits
  19. 25 Sep, 2014 1 commit
  21. 19 Sep, 2014 1 commit
  23. 16 Mar, 2009 4 commits
  25. 09 Mar, 2009 1 commit
  27. 08 Mar, 2009 7 commits
  29. 04 Mar, 2009 1 commit
  31. 03 Mar, 2009 5 commits
  33. 02 Mar, 2009 1 commit
GitLab Nexedi Edition | About GitLab | About Nexedi | 沪ICP备2021021310号-2 | 沪ICP备2021021310号-7