Resolved,
I just found a new version of MC, installed and everything works fine.
Printable View
Resolved,
I just found a new version of MC, installed and everything works fine.
Want to come back on the upnp daemon.
It's starting during boot, but it's also crashing every now and then, and I don't have fastnat enabled;)
there's no log about it happening either:confused:
The result:
I'm using the Toolchain (gcc 4.3.5 uClibc 0.9.30.1) K26 x86_64 build 2011.05.05 version to compile it all.Code:(gdb) run
Starting program: /usr/sbin/upnp -L br0 -W vlan2
(no debugging symbols found)
(no debugging symbols found)
(no debugging symbols found)
(no debugging symbols found)
(no debugging symbols found)
(no debugging symbols found)
calling upnp_main
Program received signal SIGPIPE, Broken pipe.
0x2ab39cf4 in write () from /lib/libc.so.0
(gdb)
anyone know if it is safe to import config file exported from -d version in -rtn version?
yeah, should be fine.
The thing is, I did have some configuration problems when I tried it, so in the end it might be better just to do a manual reconfiguration when you experience the same problems:)
I must say that the rtn branch runs more stable than the d branch on my wl-500w:D
I use my WL-500W as a wireless ethernet bridge to my RT-N16.
so you plug in ethernet cables and it'll send the packages over wireless to the RT-N16.:)
I noticed:
- the connection is always alive, under the -d firmware it would stop working sometimes.
- Better sensitivity of the network
- Higher throughput
so yeah, you could say so:)
Never mind, I think I got it working in the other way:pQuote:
I'd like to play around with tcp_vegas a bit so I loaded the module with modprobe, but somehow the files don't appear in the /proc/sys directory:
the log shows that vegas is registered tho:Quote:
echo 1 > /proc/sys/net/ipv4/tcp_vegas_cong_avoid
-sh: can't create /proc/sys/net/ipv4/tcp_vegas_cong_avoid: nonexistent directory
also the available methods show vegas:Quote:
kernel: TCP vegas registered
any ideas?:)Quote:
cat /proc/sys/net/ipv4/tcp_available_congestion_control
cubic reno vegas
doc: http://neal.nu/uw/linux-vegas/
Code:modprobe tcp_vegas
cat /proc/sys/net/ipv4/tcp_available_congestion_control
cubic reno vegas
#setting it to vegas
echo vegas > /proc/sys/net/ipv4/tcp_congestion_control
cat /proc/sys/net/ipv4/tcp_congestion_control
vegas
wow.. I thought broadcom driver is not compatible with 2.6 kernel but this means it is? Does moving to this firmware from the regular 2.4 based images released over at http://code.google.com/p/wl500g have any impact on apps that are installed or is it a straight upgrade? (I currently have wl500g-1.9.2.7-d-r1825 on my wl-500w)
Can anyone tell me how to install WL500gpv2-1.9.2.7-rtn-r2972.trx? I have no problem with the -d release, but that one doesn't seem to flash. The power light starts blinking and then it gets stuck. Obviously into a WL500 GP v2. Thanks.
The one from Windows? (un)fortunately I use a MacBook. I tried with a virtual machine but obviously it messed up. Now I installed Win7 in a partition and that may help in case.
What i did has been to upload the fw from the web interface. it works fine (and fast) with the -d release. With the -rtn I don't know what it happens.
I managed to unblock the router using the wl500g-clear-nvram.trx and wl500g-recover.trx via tftp, fortunately I've found some instructions in the ww-drt site.
So, my only guess is that it really should take 5 minutes to install? How do i know if the firmware has been installed? I am not sure because I messed up, but I thought that in one try I left the router on for 10 minutes or so, but I can remember wrong. Anyway, what do you suggest? The Windows tool?
Well you could use the mac osx build in tftp client to upload the firmware.
In the end... that's all the windows tool does:)
If you're running a VM of windows you could try to bridge the network adapters, that way it might work as well:)
That's what I did indeed (upload via tftp). Still the rtn version seemed to not load. I check the M5 number and it is correct, so it isn't a file corrupted problem. I will try maybe tomorrow, waiting few 10 minutes this time. I know how to restore the old firmware anyway if anything goes wrong again. Thanks. I'll let you know.
Nothing. I am totally unable to install the rtn release of the wl500gp v2 fw. I tried also with the restore utility from Windows and it got stuck waiting for an IP address. I gave a static IP address (192.168.1.2) to Windows and same thing. Uploaded via tftp, binary mode (octet) it got uploaded (I enabled trace as well) and that's it. Left on for 10-20 minutes, nothing happened.
How do you guys flash the rtn version?! I need it because I read it might fix a problem with the WiFi (that in my case it can disappear for weeks, but the light is on and then sometime it reappears). This is very frustrating.
I'm sorry it doesn't work for you:(
I only have a wl-500w and a rt-n16, both just flash fine trough the web-interface.
Also the wl-500gPv2 should be compatible with the rtn branch.
You've also tried restoring to factory defaults before flashing?
have you also tried newer builds: http://wpte.kicks-ass.net/downloads/...101/wl-models/ ?
I think I tried everything. It just seems it doesn't like that file. But the checksum is correct, so i have no clue.
I will give the night builds a look. Thanks
I flashed the WL500W-1.9.2.7-rtn-r3121.trx on my asus wl-500w this weekend. Everything seems to be working great but I see one issue. UPnP no longer works. None of my programs that use upnp for opening ports are able to do so. Even transmission (running on router itself) is not able to open the incoming port and shows it as closed. On checking the Status & Log -> Port forwarding page, it is completely blank, not even a single port forwarded, which is strange because I run a lot of services on my PC, PS3 and router that forward ports through UPnP and earlier this page used to be filled with atleast 15-20 ports always.. I've enabled UPnP (yes, report wan address) under IP Config->Miscellaneous. Any ideas?
Have you tried resetting the router to factory defaults to check if it's not a configuration problem?
Also, if this tool (I made) can find it nothing should be wrong: http://code.google.com/p/csharp-upnp-portmapper/ (it's still very simplistic)
Also you could try these settings:
IP Config - Miscellaneous -> Efficient Multicast Forwarding: Disabled
Wireless - Advanced -> Enable WMF? Yes
Your firmware version has a upnp patch that stopped the upnp daemon from crashing (an issue I experienced myself)
You can use this script to reset the upnp daemon:
(thanks to ryzhov_al for this script)Quote:
#!/bin/sh
killall upnp
for line in `nvram show | grep "forward_port" | sed "s/ //"`
do
i=`expr index $line "="`
let i=i-1
rule=`expr substr $line 1 $i`
nvram unset $rule
done
upnp -D -L `nvram get lan_ifname` -W `nvram get wan_ifname`
please check upnp with this tools (win32)
http://www.zultrax.com/forward-port/upnp.htm
WMF is about wireless multicast only, it's not related with UPNP.
FW has miniupnpd, not upnp as daemon, script is useless.
And, that's why leases are not saved after restarts.
Ok, another installer, and not even a dedicated upnp-nat program
When I disabled it I couldn't find any media server wirelessly anymore, since that happens using SSDP (what upnp uses) it is a thing to try out.
True, forgot that.
Thanks wpte and theMiron. I'll try out your suggestions and report back.
Hi wpte/theMiron
I tried all your suggestions.
Efficient multicast forwarding and WMF are already disabled.
When I run wpte's tool, it is able to map the ports well..
However, when I run the tool suggested by theMiron, it gives following output:
TEST 1 - Operating System Support - PASSED
TEST 2 - SSDP Service Running Check - FAILED
TEST 3 - SSDP Service Automatic Check - FAILED
TEST 4 - UPnPHost Service Running Check - FAILED
TEST 5 - UPnPHost Service Automatic Check - FAILED
TEST 6 - UPnP Framework Firewall Exception Check - PASSED
TEST 7 - Adapter #0 - 192.168.0.7 - PASSED
TEST 8 - Get External IP Address (Result: x.x.x.x) - PASSED
These are all run on my PC. Router is connected directly to my ISP's cable without any modem in between and gets the external IP assigned correctly. Also, I downloaded utorrent and ran that and that is also working fine for adding ports through upnp.
However, the programs that I run on my router itself (e.g. transmission) are not able to open the ports through upnp and the port always shows up as closed. I'm very confused by all these results :(. Unfortunately I'm pretty naive with upnp/nat-pmp etc and couldn't debug anything more. Please let me know if I can provide any more information.
Yeah, I can do that but I have set transmission to use random ports..I can set it to use fixed port for now and get by this issue but just interested in seeing if the problem can be fixed..
Anyways, I did some more searching around and debugging. Used a tool from http://bitlet.org/upnp and it reports no UPnP enabled gateway devices found. Also ran transmission with debug logging mode and saw that UPnpDiscover and UPnP_getValidIGD calls were failing. Searching for this on transmission forums lead me to this ticket https://trac.transmissionbt.com/ticket/3452 where they mention that this problem started happening when they "fixed" their code to use proper return values from UPNP_GetValidIGD. Earlier they were just checking for >1 and now they check specially for 1.
miniupnp says:
so miniupnp is getting confused and returning 3, basically not finding any valid IGD. Searching on miniupnp forums, it is mentioned that this is a router problem, not a miniupnp problem.Quote:
/* UPNP_GetValidIGD() :
* return values :
* 0 = NO IGD found
* 1 = A valid connected IGD has been found
* 2 = A valid IGD has been found but it reported as
* not connected
* 3 = an UPnP device has been found but was not recognized as an IGD
Just thought of providing this information in case you can make some connections :)
please try following (on the router):
1. killall miniupnpd
2. edit /etc/miniupnpd.conf:
search for
and insert lines as following:Code:...
listening_ip=...
...
allow 1024-65535 ... 1024-65535
...
3. miniupnpdCode:...
listening_ip=127.0.0.1
listening_ip=...
...
allow 1024-65535 127.0.0.0/8 1024-65535
allow 1024-65535 ... 1024-65535
...
4. check if it works
Another query: Is it ok to use inotify calls on rtn builds? (since they are built on 2.6 kernel)?
I keep getting the message:
and ports entered are not forwarded:(Quote:
miniupnpd[526]: Can't find in which sub network the client is
that was the only post I made about the problem.
I'm not quite sure, I just use upnp-nat every now and then and noticed it didn't work:p
shantanugoel (a few posts back) had issues, I remember that before that I had no problems with miniupnpd. after that there's only this update: http://code.google.com/p/wl500g/source/detail?r=3169 as far as I can see in the svnlog.
the configuration seems correct as well:
presentation url is on the same port as my web-admin, that's ok I guess?Quote:
# automagically generated
ext_ifname=vlan2
listening_ip=192.168.2.1/255.255.255.0
port=0
enable_upnp=yes
enable_natpmp=yes
lease_file=/tmp/upnp.leases
secure_mode=no
presentation_url=http://192.168.2.1:8080/
system_uptime=yes
notify_interval=60
clean_ruleset_interval=600
uuid=002618a1-3cfb-0026-18a1-3cfb00000000
model_number=RT-N16
allow 1024-65535 192.168.2.0/24 1024-65535
deny 0-65535 0.0.0.0/0 0-65535
Anyway, I checked it just now, and suprisingly it works:confused: there is no more message about the sub network either...
I didn't change anything either:o
yeah..the issue I had with miniupnpd was that programs running on router itself are not able to open any ports but programs running on my PC can open ports. Using 127.0.0.1 in listening_ip in the configuration didn't solve the problem as well but I noticed that if I run miniupnpd with the listening IP specified on the command line as 127.0.0.1 instead of specifying in config file, then the programs running on the router are able to open ports.
My miniupnpd crashed today...
wel it stayed in a 100% cpu usage loop displaying the messages
over and over again.Quote:
Oct 20 20:12:16 miniupnpd[526]: send(res_buf): Broken pipe
I wasn't able to kill it for some reason:(
I've version 1.9.2.7-rtn-r3422
wpte
100% cpu usage already fixed in r3479 ;) It is a bug inside miniupnpd. I would be glad if someone opened a problem at miniupnpd's forum, I haven't enough time for it.
True root of problem should be investigated, but I can't reproduce this bug myself. Probably, recipient close other end of pipe too early and miniupnpd either don't parse response correctly or it is an hole in workflow...
My bad, I looked over that update:o
I've only had it once as well... it's a rare problem I guess...
Looks like a problem with the upnp request parsing, I don't understand why they don't use the intel upnp toolkit (it's a bit bigger in size I guess) since it has been running completely stable here in C here.
I experiment with it to control the QoS rules based on what programs the client PC is running, so that you'll always have optimal rules :D