--- Version 21
+++ Version 22
@@ -485,25 +485,37 @@
'''03/02/06'''
'''Dan's Hastily Scribbled Testing Notes'''
+
* Thanks for the hostnames in the HTML files - makes it easier to track what's what
+
* Thanks for the pre_sript paramater passing - makes scripting for iptables a lot easier.
** Can you please do the same for post_script? That way when frottle shuts down it can delete the iptables rules it set up in pre_script. If you leave the rules in place when frottle shuts down you can get locked out of the system!
'' yep - I assume you want the same parameters passed in?''
+** Yep exactly
+
* Can you make a note in the config file that the single-mode options override the multi-mode options?
'' It shuldn't really. I have had a lot of problems with adding the multi interface and not breaking the original definitions. There are some places where I havent fully dual pathed the code so it falls back to a parameter that is defined using the original single-mode parameters. I'm going to change this so the single mode parameters are put in a domain "default" and then the code gets cleaner again. ''
+
* Can you release your code revisions as a source patch so I can test on my i386 system?
'' I build on one platform and when it is mostly right I generate the patch , I don't want to do that untilthe major bugs are squished. But as soon as possible I'll generate the patch. ''
+** Cool
+
* In Single Master mode, Frottle passes eth1 to the master_pre script no matter what interface it is running on - it'd be nice to be able to specifiy this in the config file so it can be passed to the pre and post master scripts
'' You should be able to specify this with winterface in the config file. I'll remove the default and generate an error if the interface is not specified''
+** I did try changing winterface - this works in client mode, but not master mode - winterface shows up in the "client only settings"
+
* How do you set hiports for a Master running in selfclient mode? Is this another disadvantage of selfclient mode - that you can't set hiports? I ask because hiports show up on the Master html page.
'' not sure will check the code ''
* In Multi-mode it would be very nice to be able to run on port 999 on all interfaces - can't frottle discriminate between Master threads based on IP address rather than port number? Having different port numbers makes frottle more difficult to configure for users - one more bit of info they have to co-ordinate.
'' I knew you would ask for this. It depends on how the incoming messages are presented. If the thread only receives messages on it's address or if it receives all mesages that come in on that port. Usually in userspace listening to a port will get messages from all interfaces, that would be a real problem for frottle control messages - the extra interrupts to process and discard the unwanted ones would introduce too much overhead''
+** OK, overhead is something we can't really afford on WRT's. I'm happpy to stick with separate port numbers. Just something to keep in mind - if ever there comes an easy way to use port 999 for all Masters we should go for it.
+
* Can you make the frottle package depend on the kmod-ipt-queue package from the WhiteRussian repository?
'' I can add the dependancy in the ipkg description. That way it will bring in the kmod-ipt-queue that it finds in the installed ipkg list ''
This is the URL
http://downloads.openwrt.org/whiterussian/packages/kmod-ipt-queue_2.4.30-brcm-2_mipsel.ipk
I don't think ipkg checks dependencies from packages installed directly when the full URL is specified, but it's a nice thing to include anyway.
+
* There's a spelling mistake in comments in frottle.conf - Maser only settings. :)
''There's probably more than one !!''
@@ -540,11 +552,14 @@
'' This error indiates a return code of -1 from ipq_read. I suspect this is a version squew between what I am compiling with and the shared lib on your machine. In the past a timeout would have a return code of 0i and was used to do background processing''
-For what it's worth, here is my master_pre script:
+I'm using http://downloads.openwrt.org/whiterussian/packages/kmod-ipt-queue_2.4.30-brcm-2_mipsel.ipk from OpenWRT
+
+
+* Do you think you could put these scripts into the OpenWRT package?
+/usr/local/frottle_master_pre.sh
#!/bin/sh
- echo 1 $1 2 $2 3 $3 4 $4
- iface=vlan1
+ iface=$1
port=$2
echo iface $iface port $port
@@ -559,16 +574,32 @@
iptables -A OUTPUT -p ALL -o $iface -j QUEUE
iptables -A FORWARD -p ALL -o $iface -j QUEUE
-And here's my client_pre script:
+/usr/local/frottle_master_post.sh
#!/bin/sh
- #/bin/echo "0" > /proc/sys/net/ipv4/conf/all/accept_redirects
- insmod ip_queue
+ iface=$1
+ port=$2
+ echo iface $iface port $port
+
+ #/bin/echo "1" > /proc/sys/net/ipv4/conf/all/accept_redirects
+ #rmmod ip_queue
+
+ iptables -D INPUT -p UDP -i $iface --dport $port -j ACCEPT
+ iptables -D OUTPUT -p UDP -o $iface --sport $port -j ACCEPT
+
+ iptables -D OUTPUT -p ALL -o $iface -j QUEUE
+ iptables -D FORWARD -p ALL -o $iface -j QUEUE
+
+/usr/local/frottle_client_pre.sh
+ #!/bin/sh
iface=$1
port=$2
echo iface $iface port $port
+ #/bin/echo "0" > /proc/sys/net/ipv4/conf/all/accept_redirects
+ insmod ip_queue
+
#allow control packets
iptables -A INPUT -p UDP -i $iface --sport $port -j ACCEPT
iptables -A OUTPUT -p UDP -o $iface --dport $port -j ACCEPT
@@ -576,5 +607,24 @@
#queue packets on these interfaces (that clients are running on)
iptables -A OUTPUT -p ALL -o $iface -j QUEUE
iptables -A FORWARD -p ALL -o $iface -j QUEUE
+
+/usr/local/frottle_client_post.sh
+ #!/bin/sh
+
+ iface=$1
+ port=$2
+ echo iface $iface port $port
+
+ #/bin/echo "1" > /proc/sys/net/ipv4/conf/all/accept_redirects
+ #rmmod ip_queue
+
+ iptables -D INPUT -p UDP -i $iface --sport $port -j ACCEPT
+ iptables -D OUTPUT -p UDP -o $iface --dport $port -j ACCEPT
+
+ iptables -D OUTPUT -p ALL -o $iface -j QUEUE
+ iptables -D FORWARD -p ALL -o $iface -j QUEUE
+
+
+
----
Back to [MelbWirelessRouterProject]
[