Who is rejecting who? / smb.conf reloading

[John Malmberg]

> Yesterday's Topics:
> 
>    1. Who is rejecting who? (john marchant)
>    2. samba smb.conf startup question (john marchant)
>    3. semi-reproducible problem (john marchant)
> 
> --__--__--
> 
> Message: 1
> From: "john marchant" <jmarchant at cobbcounty.organization>
> Subject: Who is rejecting who?
> Who is rejecting who?  Who is denying what?
> 
> [2001/12/17 17:07:01, 0] CMS_ROOT:[SAMBA.SOURCE.LIB]UTIL_SOCK.C;2:(1035)
>   getpeername failed. Error was socket is not connected
>   Denied connection from UNKNOWN (0.0.0.0)
> [2001/12/17 17:07:01, 1] CMS_ROOT:[SAMBA.SOURCE.SMBD]PROCESS.C;2:(607)
> [2001/12/17 17:07:01, 0] CMS_ROOT:[SAMBA.SOURCE.LIB]UTIL_SOCK.C;2:(1035)
>   getpeername failed. Error was socket is not connected
>   Connection denied from 0.0.0.0


I do not know, if you increase the default error number, by using the 
logical name in one of the .txt files, then you will get a much more 
detailed log.

Usually between 5 and 8 will give the most useful information.


> 
> Message: 2
> From: "john marchant" <jmarchant at cobbcounty.organization>
> Subject: samba smb.conf startup question
> Date: Mon, 17 Dec 2001 17:22:27 -0500
> 
> If I modify smb.conf, do I have to restart nmbd also to make it take effect?


No.  The NMBD process periodically checks the smb.conf for a changed 
modification date and re-reads it.

But the NMBD does not really get much from the smb.conf unless you are 
overriding defaults.

 
> 
> Message: 3
> From: "john marchant" <jmarchant at cobbcounty.organization>
> Subject: semi-reproducible problem
> 
> I have been able to "sort of" reproduce a problem with Samba for VMS 2.0.6.
> In response to the:
> 
> [2001/12/17 17:45:46, 0] CMS_ROOT:[SAMBA.SOURCE.LIB]UTIL_SOCK.C;2:(1035)
>   getpeername failed. Error was socket is not connected
>   Denied connection from UNKNOWN (0.0.0.0)
> [2001/12/17 17:45:46, 1] CMS_ROOT:[SAMBA.SOURCE.SMBD]PROCESS.C;2:(607)
> [2001/12/17 17:45:46, 0] CMS_ROOT:[SAMBA.SOURCE.LIB]UTIL_SOCK.C;2:(1035)
>   getpeername failed. Error was socket is not connected
>   Connection denied from 0.0.0.0
> 
> errors, I changed smb.conf to:
> 
> name resolve order = hosts
> 
> from hosts wins
> 
> for one moment in the log files it appeared to authenticate and it used the
> log.machine_name to log the activity.  However, I got a network is busy (as
> I had been getting whether authentication took place or not) and the shares
> did show up.  I just got my usual network is busy.  The second time around
> using the "hosts" alone produced the error above.  I am only able to
> reproduce this every so often and it is very very hard to reproduce.  When
> it did resolve names, and it did authenticate, it did not produce the
> shares, just network is busy.  Thanks again.


The name resolve order means that it first goes to the host file, and I 
am not sure, but on OpenVMS it might also make a DNS query.  If your DNS 
is slow to respond, that could affect things.  The WINS then has it 
trying to contact a wins server.

Generally reading a local host file is the fastest, next a local WINS 
server, and the slowest can be a DNS, if the address is not in it's 
local cache, the DNS might have to query other DNS systems before giving up.

Putting the order " = wins host" might make give better performance, as 
it is most likely that if you have a wins server, the PC will be 
registered with it.

The NMBD process is an emulation of most of the WINS server functionality,
and can be the designated the WINS server for the PC clients.


-John

wb8tyw at qsl.network
Personal Opinion Only