I am getting the following errors in log.smbd every time a client machine
connects to Samba:
[2003/07/10 15:09:22, 0] lib/util_sock.c:get_socket_addr(895)
getpeername failed. Error was Transport endpoint is not connected
At the same time I get a log.0.0.0.0 file, as well as a log.clientname and
log.ip.addr.of.client file. I did some searching and found some reports saying
that this was a DNS reverse lookup problem. However, I have checked and
nslookup clientname works, as does nslookup ip.addr.of.client (although it
returns "clientname.domainname." because that's what the Windows 2000 DNS
server has defined automatically). WINS is not running on our network. The two
machines in question (the server and the client machine connecting to it) are
on the same domain and the same subnet - the client is running Windows 2000
Server SP3 and the Samba server is running on a fairly clean install of Red Hat
Linux 9.0. There is no firewalling between the two machines. Samba 3.0beta2 was
compiled from source.
This problem does not appear to be adversely affecting Samba's operation, but
it would be nice to figure out why it is occurring.
I have experienced something similar to what Paul just reported, with the
details of my report found at:
I get a 0.0.0.0.log file too (not log.0.0.0.0).
You are likely getting 0.0.0.0.log because your "log file" setting is set to %
m.log rather than log.%m, so this is unrelated.
lowering priority since it appears to be a cosmetic bug
in the log files.
Can you send me a level 10 debug log so I can see where in the
process of things this error is being generated? Please mail
it to me directly.
I have just installed and configured 3.0 rc2 on Redhat 8.0 and am seeing these
getpeername() fails because the client has disconnected. Possible
casues are bad hubs, NICs, drivers, etc.... No response from
questions posted back in July. Closing trhis one out.
originally reported against 3.0.0beta3. CLeaning out
non-production release versions.