Intermittent problems connecting to Samba share-Domain Ctlr issue

[GSheridan]

Hi everyone,

We've been having intermittent problems connecting to Samba shares from Windows machines. I wrote a test program that tries to connect to the share and get a list of all of the files in it. It tries to do this every second. On some occasions this test program succeeds 14 times in a row before failing. It then fails once, and then succeeds 14 or so more times. It keeps repeating this pattern. Other times, this test program can consistently connect to that same share every second for over an hour.

The error in the log seems to indicate that the problem happens when the Samba machine is talking to the Domain Controller to authenticate the user.

Here's the error from from the Linux machine's syslog(PGIDomainController is our Windows Domain Controller):
-----------------------------------------------------------
Sep 18 20:50:12 scstor003 winbindd[1226]: cli_rpc_pipe_close: cli_close failed on pipe \NETLOGON, fnum 0x4006 to machine PGIDomainController. Error was Write error: Connection reset by peer
-----------------------------------------------------------

We've also seen this block of error msgs:
-----------------------------------------------------------
Sep 18 15:45:23 scstor008 winbindd[5195]: [2006/09/18 15:45:23, 0] lib/util_sock.c:write_data(559)

Sep 18 15:45:23 scstor008 winbindd[5195]: write_data: write failure. Error = Connection reset by peer

Sep 18 15:45:23 scstor008 winbindd[5195]: [2006/09/18 15:45:23, 0] libsmb/clientgen.c:write_socket(138)

Sep 18 15:45:23 scstor008 winbindd[5195]: write_socket: Error writing 446 bytes to socket 15: ERRNO = Connection reset by peer

Sep 18 15:45:23 scstor008 winbindd[5195]: [2006/09/18 15:45:23, 0] libsmb/clientgen.c:cli_send_smb(168)

Sep 18 15:45:23 scstor008 winbindd[5195]: Error writing 446 bytes to client. -1 (Connection reset by peer)

Sep 18 15:45:23 scstor008 winbindd[5195]: [2006/09/18 15:45:23, 0] rpc_client/cli_pipe.c:rpc_api_pipe(790)

Sep 18 15:45:23 scstor008 winbindd[5195]: rpc_api_pipe: Remote machine PGIDomainController pipe \NETLOGON fnum 0x6returned critical error. Error was Write error: Connection reset by peer
-----------------------------------------------------------

Has anyone ever seen anything like this?
Any feedback on this would be greatly appreciated.

Thanks,
Greg Sheridan

 

[PHV]

And what are saying the PGIDomainController's logs ?

Hope This Helps, PH.
Want to get great answers to your Tek-Tips questions? Have a look at FAQ219-2884 or FAQ181-2886

 

[GSheridan]

Hi PH,

Thanks for the reply.

Our sysadmin looked in the Domain Controller's Event Logs
and didn't see anything at the time we had the error.

Anyone know of anywhere else on the PDC or the Linux machine that we could look for info? Has anyone heard of
any kind of intermittent problems like this with Samba?

I'd like to solve this problem, but people here are starting to give up. They want to scrap Samba, and
convert all the machines that are currently connecting
to Samba shares to Linux.

Thanks,
Greg Sheridan

 

[John Lewis]

This could well be a hardware issue. I would suspect a cable first, although a network card or switch port or event the switch could be causing the problem. The "connection reset by peer" does not specifically indicate a problem with Samba or the linux box.

I did a bit of searching and did not turn up anything that resembles this. It would be helpful to have your Samba version ('smbd -V' -- capital V) to filter things down a bit.

 

[GSheridan]

Thanks for the reply. It does seem like it could be a hardware problem given it's intermittent nature, but I forgot to mention that we have other servers that are Windows that we can connect to reliably. The Windows and Samba severs are all on the same hardware, including switches, network segments, etc.

So it seems like it is something to do with Samba.

Thanks again,

Greg Sheridan