Pulling pretty big revision...

[Asmodehn Shade]

HI all,

Recently I was having trouble pulling revisions around because of their
size, and because I was hitting a problem in bzr 1.8 using bzr+ssh://

Since then I switched to sftp:// which seems to work better than bzr+ssh://
for very big transfers ( splitting them in chunks ), even if it is now
slower.

However I also had a weird problem recently... after a long transfer ( many
hours ), I ended up with this :

 lock_write
    token_from_lock = self._lock.lock_write(token=token)
  File "/usr/lib/python2.5/site-packages/bzrlib/lockdir.py", line 576, in
lock_write
    return self.wait_lock()
  File "/usr/lib/python2.5/site-packages/bzrlib/lockdir.py", line 500, in
wait_lock
    return self.attempt_lock()
  File "/usr/lib/python2.5/site-packages/bzrlib/lockdir.py", line 461, in
attempt_lock
    result = self._attempt_lock()
  File "/usr/lib/python2.5/site-packages/bzrlib/lockdir.py", line 223, in
_attempt_lock
    tmpname = self._create_pending_dir()
  File "/usr/lib/python2.5/site-packages/bzrlib/lockdir.py", line 281, in
_create_pending_dir
    info_bytes = self._prepare_info()
  File "/usr/lib/python2.5/site-packages/bzrlib/lockdir.py", line 436, in
_prepare_info
    user = config.user_email()
  File "/usr/lib/python2.5/site-packages/bzrlib/config.py", line 214, in
user_email
    return extract_email_address(self.username())
  File "/usr/lib/python2.5/site-packages/bzrlib/config.py", line 242, in
username
    name, email = _auto_user_id()
  File "/usr/lib/python2.5/site-packages/bzrlib/config.py", line 842, in
_auto_user_id
    w = pwd.getpwuid(uid)
KeyError: 'getpwuid(): uid not found: 2050'

bzr 1.8 on python 2.5 (linux2)
arguments: ['/usr/bin/bzr', 'pull', '-r33',
'sftp://[...]/home/autobzr/deployBZR/GameBZR']
encoding: 'UTF-8', fsenc: 'UTF-8', lang: 'en_US.UTF-8'
plugins:
  bzrtools
/usr/lib/python2.5/site-packages/bzrlib/plugins/bzrtools [1.8.0]
  launchpad
/usr/lib/python2.5/site-packages/bzrlib/plugins/launchpad [unknown]


I am curious to investigate it a bit further if someone can have time for
that...

uid 2050 is a user that I use on both machines. It s actually a LDAP user (
remote LDAP server ). This user is usually well know without problem.
since I was using sftp, the transfer was way under the bandwith available so
I dont think it s due to a network pipe overloaded with dataflow...

It's not the first time it happens... it seems quite reproduceable...
I hit this problem on sftp:// transfer when it takes "too long" but not on
bzr+ssh://, for the same revision of the same branch...

So I would like to know
- if it sounds like a known issue to someone,
- if so, is it going to be fixed in 1.9 ?
- if someone has an idea about what may cause this
- if someone is interested to debug it let me know...

Thanks a lot !

--
Alex
-------------- next part --------------
An HTML attachment was scrubbed...
URL: https://lists.ubuntu.com/archives/bazaar/attachments/20081030/e0ae4ecb/attachment.htm