Bug 2439 - [PATCH] query_name_response: Multiple responses received for a query
[PATCH] query_name_response: Multiple responses received for a query
Status: NEW
Product: Samba 3.0
Classification: Unclassified
Component: nmbd
3.0.11
x86 Linux
: P3 normal
: none
Assigned To: Samba Bugzilla Account
Samba QA Contact
:
Depends on:
Blocks:
  Show dependency treegraph
 
Reported: 2005-03-11 04:35 UTC by Filipe Henriques
Modified: 2005-03-11 04:39 UTC (History)
0 users

See Also:


Attachments
Here is the patch (1.88 KB, patch)
2005-03-11 04:38 UTC, Filipe Henriques
no flags Details

Note You need to log in before you can comment on or make changes to this bug.
Description Filipe Henriques 2005-03-11 04:35:03 UTC
Two nmblookup responses with the same IP address shouldn't log an error.

I got a Linux box, running samba-3.0.11 on a Windows network, controlled by 
Windows Servers (I'm not the system administrator :-)).
My team make a lot of developments in this Linux box and I configured it to 
provide file-sharing to some windows clients.

But I've noticed a lot of log messages like this:

[2005/02/20 04:14:56, 0] nmbd/nmbd_namequery.c:query_name_response(101) 
  query_name_response: Multiple (2) responses received for a query on subnet 
172.16.67.176 for name MYDOMAIN<1d>. 


When I do a nmblookup, I got this:

$> nmblookup -M MYDOMAIN
querying MYDOMAIN on 172.16.127.255
172.16.127.90 MYDOMAIN<1d>
172.16.127.90 MYDOMAIN<1d>


I know that MYDOMAIN got two name servers for reduncy, so when a query is 
made, I got two responses with the same IP address.
But I think this is not an erroneous behaviour. So, when I got two (or more) 
responses with the same IP, nmbd shouldn't log an error.

With this on my mind, I patched nmbd_namequery.c so that don't log if got two 
(or more) responses with the same IP address.
I submit the patch to you, for your analysis, approval and possibly include it 
on future samba releases.
I need to keep the last IP received, so I put it on a global variable 
(last_answer_ip). You will probably need to put it in a better place.


Here is the patch:

diff -Nur samba-3.0.11.org/source/nmbd/nmbd_namequery.c
samba-3.0.11.new/source/nmbd/nmbd_namequery.c
--- samba-3.0.11.org/source/nmbd/nmbd_namequery.c	Mon Oct 25 22:05:05 2004
+++ samba-3.0.11.new/source/nmbd/nmbd_namequery.c	Mon Mar  7 12:32:11 2005
@@ -23,6 +23,10 @@
 
 #include "includes.h"
 
+/* -ffh 2005-2-23 (Filipe Henriques):
+   Keep record of last response. */
+struct in_addr last_answer_ip;
+
 /****************************************************************************
  Deal with a response packet when querying a name.
 ****************************************************************************/
@@ -83,7 +87,11 @@
 			success = True;
 
 			putip((char *)&answer_ip,&nmb->answers->rdata[2]);
-	
+
+			/* -ffh 2005-2-23 (Filipe Henriques):
+			   Keep record of this response. */
+			putip((char *)&last_answer_ip,&nmb->answers->rdata[2]);
+
 			if( DEBUGLVL( 5 ) ) {
 				dbgtext( "query_name_response: On subnet %s ", subrec->subnet_name );
 				dbgtext( "- positive response from IP %s ", inet_ntoa(p->ip) );
@@ -97,8 +105,14 @@
 			*/
 		}
 	} else if( rrec->num_msgs > 1) {
+		/* More than one response. */
 
-		if( DEBUGLVL( 0 ) ) {
+		/* -ffh 2005-2-23 (Filipe Henriques):
+		   If this response answered the same IP address,
+		   then that's because the network got reduncy servers
+		   responding to this query.
+		   Ignore, unless high debug level or diferent IP address. */
+		if( DEBUGLVL( 5 ) || ( (memcmp (&last_answer_ip, &nmb->answers->rdata[2], 4)
!= 0) && DEBUGLVL( 0 ) ) ) {
 			if (nmb->answers)
 				putip( (char *)&answer_ip, &nmb->answers->rdata[2] );
 			dbgtext( "query_name_response: " );
@@ -107,6 +121,7 @@
 			dbgtext( "for name %s.\nThis response ", nmb_namestr(question_name) );
 			dbgtext( "was from IP %s, reporting ", inet_ntoa(p->ip) );
 			dbgtext( "an IP address of %s.\n", inet_ntoa(answer_ip) );
+			dbgtext( "Previous response reported an IP address of %s.\n",
inet_ntoa(last_answer_ip) );
 		}
 
 		/* We have already called the success or fail function, so we
Comment 1 Filipe Henriques 2005-03-11 04:38:22 UTC
Created attachment 1026 [details]
Here is the patch