We have servers that deal with power outages (and dirty power in general) and
have ended up with some corrupted durus databases. We've used the normal
"repair" feature to handle a lot of the cases but we have another that's
occurring on occasion that is not handled by repair. It is failing as this
exception:

Traceback (most recent call last):
  File "/usr/local/bin/durus", line 22, in 
    client_main()
  File "/usr/local/lib/python2.6/dist-packages/durus/client.py", line 108, in
client_main
    options.startup)
  File "/usr/local/lib/python2.6/dist-packages/durus/client.py", line 35, in
interactive_client
    storage = FileStorage(file, readonly=readonly, repair=repair)
  File "/usr/local/lib/python2.6/dist-packages/durus/file_storage.py", line 73,
in __init__
    self.shelf = Shelf(filename, readonly=readonly, repair=repair)
  File "/usr/local/lib/python2.6/dist-packages/durus/shelf.py", line 91, in
__init__
    self.file, repair=repair)
  File "/usr/local/lib/python2.6/dist-packages/durus/shelf.py", line 296, in
read_transaction_offsets
    file.seek(position + 8 + record_length)
  File "/usr/local/lib/python2.6/dist-packages/durus/file.py", line 41, in seek
    self.file.seek(n, whence)
IOError: [Errno 22] Invalid argument

We've come up with the following patch:

--- /usr/local/lib/python2.6/dist-packages/durus/shelf.py       2007-04-24
16:10:16.000000000 -0500
+++ durus/shelf.py      2011-08-11 08:18:51.169288642 -0500
@@ -297,18 +297,17 @@
         if file.tell() != transaction_end:
             raise ShortRead
         return transaction_offsets
-    except ShortRead, e:
+    except (ShortRead, IOError), e:
         position = file.tell()
         if position > transaction_start:
             if repair:
                 file.seek(transaction_start)
                 file.truncate()
             else:
-                e.args = repr(dict(
+                raise ShortRead(repr(dict(
                     transaction_start=transaction_start,
                     transaction_end = transaction_end,
-                    position=position))
-                raise
+                    position=position)))
         return None

Or test database started out as:

-rw-r--r--  1 fishfinder fishfinder 49371713 2011-08-10 14:12 db.durus

We ran with this patch and without the repair switch and got the following
(expected) report:

durus.utils.ShortRead: {'position': 47501336L, 'transaction_start': 47501312L,
'transaction_end': 12061693972974895416L}

We then ran again with --repair and the database was able to be opened. The
resulting database looked like this:

-rw-r--r--  1 fishfinder fishfinder 47501312 2011-08-11 08:14 db.durus

Can anybody comment on whether this patch makes the appropriate sense and is
safe enough? We (fortunately!) have a limited number of corrupted databases to
try this on.

Thanks.

Dave