Re: [Duplicity-talk] Problems with "librsync error 103 while in patch cycle"

[Michael Terry]

Heyo!  I'm just dropping a line to say I've seen this email and it looks like I need to do some further work on that area of code in terms of letting the user continue.  But I've been very busy lately and haven't gotten to it.

So I acknowledge the problem, but haven't had time to work on it.  :-/

-mt

On 3 November 2013 21:21, Echo Nolan <address@hidden> wrote:

Hi again. I managed to get most of my data restored :) I'm not sure what the NoneType error means exactly, but the below patch allows restore to continue.

=== modified file 'duplicity/patchdir.py'
--- duplicity/patchdir.py    2013-09-19 00:26:01 +0000
+++ duplicity/patchdir.py    2013-11-03 22:20:35 +0000
@@ -23,6 +23,7 @@
 import types
 import os
 import tempfile
+import pprint
 
 from duplicity import tarfile address@hidden
 from duplicity import librsync address@hidden
@@ -504,11 +505,16 @@
                 # otherwise final patch was delete
                 yield final_ropath
         except Exception, e:
-            filename = patch_seq[-1].get_ropath().get_relative_path()
-            log.Warn(_("Error '%s' patching %s") %
-                     (str(e), filename),
-                     log.WarningCode.cannot_process,
-                     util.escape(filename))
+            if patch_seq[-1] != None:
+                filename = patch_seq[-1].get_ropath().get_relative_path()
+                log.Warn(_("Error '%s' patching %s") %
+                         (str(e), filename),
+                         log.WarningCode.cannot_process,
+                         util.escape(filename))
+            else:
+                log.Warn(_("Error '%s' patching unknown file") %
+                         (str(e)),
+                         log.WarningCode.cannot_process)
 
 def tarfiles2rop_iter( tarfile_list, restrict_index=() ):
     """Integrate tarfiles of diffs into single ROPath iter

That got me a new error, further into the restore process. I was getting a KeyError exception:

Traceback (most recent call last):

  File "bin/duplicity", line 1469, in <module>
    with_tempdir(main)
  File "bin/duplicity", line 1462, in with_tempdir
    fn()
  File "bin/duplicity", line 1396, in main

    restore(col_stats)
  File "bin/duplicity", line 687, in restore
    restore_get_patched_rop_iter(col_stats)):

  File "/home/enolan/duplicity/duplicity/patchdir.py", line 541, in Write_ROPaths
    for ropath in rop_iter:
  File "/home/enolan/duplicity/duplicity/patchdir.py", line 501, in integrate_patch_iters
    for patch_seq in collated:
  File "/home/enolan/duplicity/duplicity/patchdir.py", line 381, in yield_tuples
    setrorps( overflow, elems )
  File "/home/enolan/duplicity/duplicity/patchdir.py", line 370, in setrorps
    elems[i] = iter_list[i].next()
  File "/home/enolan/duplicity/duplicity/patchdir.py", line 140, in difftar2path_iter
    tarinfo_list[0] = tar_iter.next()
  File "/home/enolan/duplicity/duplicity/patchdir.py", line 336, in next
    self.set_tarfile()
  File "/home/enolan/duplicity/duplicity/patchdir.py", line 325, in set_tarfile
    self.current_fp = self.fileobj_iter.next()
  File "bin/duplicity", line 726, in get_fileobj_iter
    backup_set.volume_name_dict[vol_num],
KeyError: 4

To figure out what was going on, I added some debug print lines:

=== modified file 'bin/duplicity'
--- bin/duplicity    2013-09-20 01:32:31 +0000
+++ bin/duplicity    2013-11-04 02:05:07 +0000
@@ -29,7 +29,7 @@
 
 import getpass, gzip, os, sys, time, types
 import traceback, platform, statvfs, resource, re
-import threading
+import threading, pprint
 from datetime import datetime
 
 pwd = os.path.abspath(os.path.dirname(sys.argv[0]))
@@ -719,6 +719,9 @@
         manifest = backup_set.get_manifest()
         volumes = manifest.get_containing_volumes(index)
         for vol_num in volumes:
+            print "vol_num: " + str(vol_num)
+            print "volume_name_dict"
+            pprint.pprint(backup_set.volume_name_dict)
             yield restore_get_enc_fileobj(backup_set.backend,
                                           backup_set.volume_name_dict[vol_num],
                                           manifest.volume_info_dict[vol_num])

Results showed it was looking for a fourth volume in duplicity-inc.20121218T113324Z.to.20130121T051302Z. I decrypted the manifest:

Hostname localhost6.localdomain6
Localdir /home/enolan
Volume 1:
    StartingPath   . 
    EndingPath     ".config/google-chrome/Default/Top\x20Sites" 3
    Hash SHA1 40f6ddd332d82b5197db4bdbf13278b9abfb6e71
Volume 2:
    StartingPath   ".config/google-chrome/Default/Top\x20Sites" 4
    EndingPath     .dropbox-dist/library.zip 112
    Hash SHA1 3ec708b2a6b48e6efa9662dcdb751dcad6c153cf
Volume 3:
    StartingPath   .dropbox-dist/library.zip 113
    EndingPath     .local/share/Steam/SteamApps/common/Uplink/lib/libmikmod.so.2 8
    Hash SHA1 76e5a803315ff9bca2a6de9c7d2a04d81ab77e0f
Volume 4:
    StartingPath   .local/share/Steam/SteamApps/common/Uplink/lib/libmikmod.so.2 9
    EndingPath     ".local/share/Steam/SteamApps/common/World\x20of\x20Goo/WorldOfGoo" 49
    Hash SHA1 f7ea5f82fa34ce159095b46555b3877cd16d5b5a
Volume 3:
    StartingPath   .local/share/Steam/ClientRegistry.blob 
    EndingPath     "mystuff/writing/misc/resume\x20non\x20tech.pdf" 
    Hash SHA1 10eaa0b647bd998c34d7f5775894152a16e45dd1

There are only three volumes in that incremental backup in reality. I deleted the first volume 3 and 4 entries, and made sure that the SHA1 of volume 3 matched the real hash of the volume. After that, restore finished properly, though still with many errors.

Hope this helps somebody. :)

On Sun, Nov 3, 2013 at 1:37 PM, <address@hidden> wrote:

you should definitely rethink your backup strategy for the next time, but i guess you've already came to that conclusion ;)

wrt. the problem. as i wrote to Mike. try catching the error where it is thrown. the restore should continue then with the next file instead of breaking completely.


..ede/duply.net

On 03.11.2013 20:02, Echo Nolan wrote:
> Edgar et al,
>
> Unfortunately, it looks like the initial full backup is corrupt. The restore reaches "Processed volume 13 of 75" before dying, and the initial backup is 25 volumes. I was able to get two important files out with --file-to-restore, I guess they didn't get created until after the first backup? Is there something else I can do to get a partial restore?
>
>

> On Sun, Nov 3, 2013 at 3:30 AM, <address@hidden <mailto:address@hidden>> wrote:
>
>     On 03.11.2013 07:08, Echo Nolan wrote:
>     > Hi All.
>     >
>     > I'm trying to restore a backup. Here's what happens with the version of duplicity shipping with Ubuntu 13.10 (0.6.21):
>     >
>     > address@hidden:~$ rm -r restore && duplicity file://enolan-desktop-backup restore/
>     > Local and Remote metadata are synchronized, no sync needed.
>     > Last full backup date: Sun Jan 15 08:05:07 2012
>     > GnuPG passphrase:
>     > python: ERROR: (rs_tube_catchup) reached end of file while copying literal data through buffers
>     > python: ERROR: (rs_job_complete) patch job failed: unexpected end of input
>     > Traceback (most recent call last):
>     >   File "/usr/bin/duplicity", line 1411, in <module>
>     >     with_tempdir(main)
>     >   File "/usr/bin/duplicity", line 1404, in with_tempdir
>     >     fn()
>     >   File "/usr/bin/duplicity", line 1338, in main
>     >     restore(col_stats)
>     >   File "/usr/bin/duplicity", line 632, in restore
>     >     restore_get_patched_rop_iter(col_stats)):
>     >   File "/usr/lib/python2.7/dist-packages/duplicity/patchdir.py", line 526, in Write_ROPaths
>     >     for ropath in rop_iter:
>     >   File "/usr/lib/python2.7/dist-packages/duplicity/patchdir.py", line 499, in integrate_patch_iters
>     >     final_ropath = patch_seq2ropath( normalize_ps( patch_seq ) )
>     >   File "/usr/lib/python2.7/dist-packages/duplicity/patchdir.py", line 479, in patch_seq2ropath
>     >     misc.copyfileobj( current_file, tempfp )
>     >   File "/usr/lib/python2.7/dist-packages/duplicity/misc.py", line 166, in copyfileobj
>     >     buf = infp.read(blocksize)
>     >   File "/usr/lib/python2.7/dist-packages/duplicity/librsync.py", line 80, in read
>     >     self._add_to_outbuf_once()
>     >   File "/usr/lib/python2.7/dist-packages/duplicity/librsync.py", line 94, in _add_to_outbuf_once
>     >     raise librsyncError(str(e))
>     > librsyncError: librsync error 103 while in patch cycle
>     >
>     > I googled, and found this ticket in Ubuntu's bugtracker: https://bugs.launchpad.net/duplicity/+bug/662442
>     >

>     > It seemed like the same problem, and oh look, someone has already committed a patch allowing me to continue restoring a backup containing broken files (https://code.launchpad.net/~mterry/duplicity/catch-seq-copy-error/+merge/186106 <https://code.launchpad.net/%7Emterry/duplicity/catch-seq-copy-error/+merge/186106> <https://code.launchpad.net/%7Emterry/duplicity/catch-seq-copy-error/+merge/186106>). So I grabbed duplicity from bzr (bzr branch lp:duplicity) and built it. Trying with that version gives a different error however:

>     >
>     > address@hidden:~/duplicity$ bin/duplicity file://../enolan-desktop-backup ../restore
>     > Import of duplicity.backends.dpbxbackend Failed: No module named dropbox
>     > Local and Remote metadata are synchronized, no sync needed.
>     > Last full backup date: Sun Jan 15 08:05:07 2012
>     > GnuPG passphrase:
>     > python: ERROR: (rs_tube_catchup) reached end of file while copying literal data through buffers
>     > python: ERROR: (rs_job_complete) patch job failed: unexpected end of input
>     > Traceback (most recent call last):
>     >   File "bin/duplicity", line 1466, in <module>
>     >     with_tempdir(main)
>     >   File "bin/duplicity", line 1459, in with_tempdir
>     >     fn()
>     >   File "bin/duplicity", line 1393, in main
>     >     restore(col_stats)
>     >   File "bin/duplicity", line 687, in restore
>     >     restore_get_patched_rop_iter(col_stats)):
>     >   File "/home/enolan/duplicity/duplicity/patchdir.py", line 535, in Write_ROPaths
>     >     for ropath in rop_iter:
>     >   File "/home/enolan/duplicity/duplicity/patchdir.py", line 507, in integrate_patch_iters
>     >     filename = patch_seq[-1].get_ropath().get_relative_path()
>     > AttributeError: 'NoneType' object has no attribute 'get_ropath'
>     >
>     > Can any of you help me out? This is my only copy of several important files..
>     >
>
>     Echo,
>
>     currently rsync errors or others restore errors break restoring. the usual workaround is, assuming you have a full/incr backup chain, to restore a time before the broken volumes.
>
>     1. try listing your backup chains and save the output.
>     2. then run your restore as before but with '-v9' maximum verbosity. check which volume breaks the restore. it should be mentioned right above the error stack. interpret it's time value in the file name.
>     3. now try restoring giving a time before the time value you retrieved in step 2.
>
>     Mike: you tackled this before. could you catch his error higher up? we probably need to log a warning that this file will not be restored too.
>
>

>     ..ede/duply.net <http://duply.net>
>
>     _______________________________________________
>     Duplicity-talk mailing list
>     address@hidden <mailto:address@hidden>
>     https://lists.nongnu.org/mailman/listinfo/duplicity-talk

>
>
>
>
> _______________________________________________
> Duplicity-talk mailing list
> address@hidden
> https://lists.nongnu.org/mailman/listinfo/duplicity-talk
>

_______________________________________________
Duplicity-talk mailing list
address@hidden
https://lists.nongnu.org/mailman/listinfo/duplicity-talk

_______________________________________________
Duplicity-talk mailing list
address@hidden
https://lists.nongnu.org/mailman/listinfo/duplicity-talk