bzrlib.errors.ShortReadvError on commit

[Chadarius]

I get a ShortReadvError when I try to commit. I'm not sure exactly what is wrong yet, but I figured I would at least get a bug submitted.
The full error is :
bzr: ERROR: bzrlib.errors.ShortReadvError: readv() read 0 bytes rather than 183 bytes at 0 for "bade61d9c4160e99eb362eb4b5fdb2d1.rix"

Just for fun I tried running the commit with sudo to see if having root access would make a difference. It did not.

[Chadarius (csutton-chadarius)]

I found the following discussion http://web.archiveorange.com/archive/v/mvjb4uGsyRmfCHjnSdvG and used the script suggested to fix the issue. Well at least get me to a point where I can run bzr reconcile. Now I just get an out of memory error when I run reconcile. :)

#!/usr/bin/python

from bzrlib import repository, btree_index, transport

t = transport.get_transport('/home/repo/win7/.bzr/repository')
pn_index = btree_index.BTreeGraphIndex(t, 'pack-names', None)
nodes = pn_index.iter_all_entries()

new_index = btree_index.BTreeBuilder(0, 1)

for _, key, value in nodes:
  if key == ('bade61d9c4160e99eb362eb4b5fdb2d1',):
    continue
  new_index.add_node(key, value)

new_index_content = new_index.finish()
t.put_file('pack-names', new_index_content)

[John A Meinel (jameinel)]

There are a few things we still need to work on here

1) Does your repository seem to work 'in general' once you removed the empty pack file?
2) It would be good to understand how you got to be in this situation, though that is difficult to determine after the fact. But somehow, a file we thought we put data into, came out as being empty. Which is pretty serious data loss that would appear to be out of our control.
3) If you want us to poke into the memory issues, then a traceback of what was going on when it happened would be good. If you don't already have one then running "bzr reconcile -Derror" should be a good way to get one.