John, recv's happen during push too. For example in this traceback it is happening while reading a lockdir's info file.
Looking closely at the traceback, the large size actually comes from this bit of paramiko:
def _read_packet(self):
size = struct.unpack('>I', self._read_all(4))[0]
data = self._read_all(size)
So the SSH server is sending a ridiculously huge packet. Possibly this is because the info file we're retrieving over SFTP is actually that massive, or maybe this is just a bug in that SSH server?
EIther way, my guess is the root problem here isn't bzr's fault, although obviously bzr should be a bit more graceful in how it handles this situation. That's just a guess, though.
cash5, can you tell us what OS and filesystem and sshd is running on the server? Are there any "info" files in the subdirectories of the .bzr/branch/lock directory on the server, and if so what size are they?
John, recv's happen during push too. For example in this traceback it is happening while reading a lockdir's info file.
Looking closely at the traceback, the large size actually comes from this bit of paramiko:
def _read_packet(self): all(4)) [0] all(size)
size = struct.unpack('>I', self._read_
data = self._read_
So the SSH server is sending a ridiculously huge packet. Possibly this is because the info file we're retrieving over SFTP is actually that massive, or maybe this is just a bug in that SSH server?
EIther way, my guess is the root problem here isn't bzr's fault, although obviously bzr should be a bit more graceful in how it handles this situation. That's just a guess, though.
cash5, can you tell us what OS and filesystem and sshd is running on the server? Are there any "info" files in the subdirectories of the .bzr/branch/lock directory on the server, and if so what size are they?