---------------------
PatchSet 5089 
Date: 2007/07/20 02:06:32
Author: amosjeffries
Branch: squid3-ipv6
Tag: (none) 
Log:
Add NOTES on netfilter experuiments so far.

Members: 
	NOTES-IPv6:1.1.2.47->1.1.2.48 

Index: squid3/NOTES-IPv6
===================================================================
RCS file: /cvsroot/squid-sf//squid3/Attic/NOTES-IPv6,v
retrieving revision 1.1.2.47
retrieving revision 1.1.2.48
diff -u -r1.1.2.47 -r1.1.2.48
--- squid3/NOTES-IPv6	13 Jul 2007 13:48:03 -0000	1.1.2.47
+++ squid3/NOTES-IPv6	20 Jul 2007 02:06:32 -0000	1.1.2.48
@@ -1,4 +1,4 @@
-$Id: NOTES-IPv6,v 1.1.2.47 2007/07/13 13:48:03 amosjeffries Exp $
+$Id: NOTES-IPv6,v 1.1.2.48 2007/07/20 02:06:32 amosjeffries Exp $
 
 NP: Any problems with this version of squid please contact
 	squid3@treenet.co.nz for assistance.
@@ -21,6 +21,11 @@
 	failover on errors. The default behaviour is now to treat them as sperate
 	unique sources. This solves several other logging and split-stack problems.
 
+PROBLEM: Sockets created with IPv6 family ginve "92 Protocol not supported" errors
+	for getsockopt in clientNatLookup via netfilter (on Debian).
+	this occurs regardless of the actual source and destination IP types.
+	ie IPv4-client to IPv4-server REDIRECT'd to squid on [::]:81
+
 
 PROBLEM: Sockets created with IPv4 'family' cannot be use for IPv6 *anywhere*.
 	Likewise ones opened for IPv6 on split-stack systems cannot be used for IPv4.
@@ -39,7 +44,8 @@
 	believed. The sockets may in fact be being opened from new on both sides of
 	each request. Which is split-stack friendly.
 
-Another intermittent failure:
+
+PROBLEM: Another intermittent failure:
 NP: This one appears to be static and only occuring during destruction of IPA.
 	have reduced the assert to a debugs and printf(stderr) WARNING.
 	Have yet to find the actual place the relevant IPA is being memset? to null.