OpenWrt Forum Archive

Topic: Build for WNDR3700/WNDR3800

The content of this topic has been archived between 9 Jul 2013 and 6 May 2018. Unfortunately there are posts – most likely complete pages – missing.

ceri wrote:

Not sure if this is a build-specific issue, but I noticed an issue with SQM. I setup a guest WLAN network and wanted to rate limit it to 4096/512. However, in order to get this to work properly, I had to switch the download and upload rates with each other in /etc/config/sqm, otherwise it imposed the opposite limits to what I intending to have

It is not build-specific. SQM is designed for the WAN interface. When you try to apply it to a interface towards LAN, the logic of traffic flow designation naturally changes:
something downloaded from WAN (from wan to router) is uploaded (from router to wifi) to your guest network in LAN.

hnyman wrote:
ceri wrote:

Not sure if this is a build-specific issue, but I noticed an issue with SQM. I setup a guest WLAN network and wanted to rate limit it to 4096/512. However, in order to get this to work properly, I had to switch the download and upload rates with each other in /etc/config/sqm, otherwise it imposed the opposite limits to what I intending to have

It is not build-specific. SQM is designed for the WAN interface. When you try to apply it to a interface towards LAN, the logic of traffic flow designation naturally changes:
something downloaded from WAN (from wan to router) is uploaded (from router to wifi) to your guest network in LAN.

Ah ok. Thanks for the clarification.

hnyman wrote:
ceri wrote:

Not sure if this is a build-specific issue, but I noticed an issue with SQM. I setup a guest WLAN network and wanted to rate limit it to 4096/512. However, in order to get this to work properly, I had to switch the download and upload rates with each other in /etc/config/sqm, otherwise it imposed the opposite limits to what I intending to have

It is not build-specific. SQM is designed for the WAN interface. When you try to apply it to a interface towards LAN, the logic of traffic flow designation naturally changes:
something downloaded from WAN (from wan to router) is uploaded (from router to wifi) to your guest network in LAN.

Hi ceri,

yes Hnyman is correct (as always), the upload and download directions are always from the point of view of the router, so for a WAN interface their are aligned directly with wan-download and wan-upload, but on a LAN/internal interface they flip direction, stuff you upload reach the router via the ingress direction and stuff you download need to be sent as egress via the internal interface. We had some discussions how to name things in the luci GUI, and opted for simple terms that for the usual situation are correct. We might want to revisit this in the future...

Best Regards
        M.

it's been several days now so i like to share CC r48238 appears stable.  i really like the wifi button.  thanks you hnyman for the build.

NTP (under "System" section, I only use LUCI for settings) questions:

I had been using both OpenWRT NTP Client and Server to have all LAN machine point to the router for time but for some reason machines were getting off by even 4-6 minutes!  So I recently decided to go back to using pool.ntp.org at each machine.

Q1) I see NTP client is enabled by default (just flashed newer version, I guess to allow router to have correct time?).  If it's disabled does that have any negative affect on the router?  Is it alright to have both NTP client & server disabled in OpenWRT?

Q2) When OpenWRT does has NTP Server enabled it's just for your LAN devices correct?  It doesn't work with anything WAN (pool.ntp.org, etc...)

Thanks!

(Last edited by Beaker1024 on 28 Jan 2016, 17:46)

how do i install this luci web gui to my phantom 2 vision plus v3 i tried to ssh in the repeater for on the control and did everything that im suppose to im wondering if im still using the old method i dont know hmm plz help me ??

phantom3 wrote:

how do i install this luci web gui to my phantom 2 vision plus v3 i tried to ssh in the repeater for on the control and did everything that im suppose to im wondering if im still using the old method i dont know hmm plz help me ??

That has nothing to do with my build. Ask that is a separate thread, please.

Hnyman, on the first page I read the reset button works. Did you use the 'restorefactory' package or something else to get it to work?

On the following page it said to use 'restorefactory' but when I opkg it is not found:
https://wiki.openwrt.org/toh/netgear/wndr3700

I'm using the Chaos Calmer branche to build an image.

Thanks in advance.

bladeoner wrote:

Hnyman, on the first page I read the reset button works. Did you use the 'restorefactory' package or something else to get it to work?

Well, you could read the answer from my release sources, as I provide rather compact patch files of all the changes...

But in any case, even the default reset button script from base-files should probably work without any changes. But that default script requires a 5 seconds button push, which is unnecessary as the reset button is so well protected in 3700/3800 that it does not get pushed accidentally. So I have simplified the reset script for my own firmware (but done just minimal changes to the script).

Default script in /etc/rc.button/reset :
https://dev.openwrt.org/browser/branche … tton/reset

#!/bin/sh

[ "${ACTION}" = "released" ] || exit 0

. /lib/functions.sh

logger "$BUTTON pressed for $SEEN seconds"

if [ "$SEEN" -lt 1 ]
then
    echo "REBOOT" > /dev/console
    sync
    reboot
elif [ "$SEEN" -gt 5 ]
then
    echo "FACTORY RESET" > /dev/console
    jffs2reset -y && reboot &
fi

return 0

My shortened script for /etc/rc.button/reset

#!/bin/sh

[ "${ACTION}" = "released" ] || exit 0

. /lib/functions.sh

logger "$BUTTON pressed for $SEEN seconds"

        echo "FACTORY RESET" > /dev/console
        jffs2reset -y && reboot &

return 0

hnyman in your latest Chaos build (48629) it seems that DnsMasq is ignoring the cachesize option. It seems to be defaulting to 1000, as I recall that used to be 150. When I add the cachesize option to /etc/config/dhcp it always defaults to 1000 no matter what number I add to the config file. Do you know if this setting has changed or do I need to add it to another config file?
Thx

Romulus wrote:

it seems that DnsMasq is ignoring the cachesize option. It seems to be defaulting to 1000, as I recall that used to be 150. When I add the cachesize option to /etc/config/dhcp it always defaults to 1000 no matter what number I add to the config file.

Make sure that you edit the existing cachesize option line, not add a second line.
I added the cachesize 1000 option there a few months ago.

(Last edited by hnyman on 5 Feb 2016, 10:22)

That solved it, I totally overlooked the already existing cachesize option.
Thx hnyman.

hnyman wrote:
bladeoner wrote:

Hnyman, on the first page I read the reset button works. Did you use the 'restorefactory' package or something else to get it to work?

Well, you could read the answer from my release sources, as I provide rather compact patch files of all the changes...

But in any case, even the default reset button script from base-files should probably work without any changes. But that default script requires a 5 seconds button push, which is unnecessary as the reset button is so well protected in 3700/3800 that it does not get pushed accidentally. So I have simplified the reset script for my own firmware (but done just minimal changes to the script).

Default script in /etc/rc.button/reset :
https://dev.openwrt.org/browser/branche … tton/reset

#!/bin/sh

[ "${ACTION}" = "released" ] || exit 0

. /lib/functions.sh

logger "$BUTTON pressed for $SEEN seconds"

if [ "$SEEN" -lt 1 ]
then
    echo "REBOOT" > /dev/console
    sync
    reboot
elif [ "$SEEN" -gt 5 ]
then
    echo "FACTORY RESET" > /dev/console
    jffs2reset -y && reboot &
fi

return 0

My shortened script for /etc/rc.button/reset

#!/bin/sh

[ "${ACTION}" = "released" ] || exit 0

. /lib/functions.sh

logger "$BUTTON pressed for $SEEN seconds"

        echo "FACTORY RESET" > /dev/console
        jffs2reset -y && reboot &

return 0

Thank you for your explanation, I couldn't find anywhere that it's already included in the source.

About your post about the reset button I think I read over it, sorry for that.

(Last edited by bladeoner on 6 Feb 2016, 11:41)

it's been over 11 days and i like to share r48558 appears rock solid on the wndr3700v2.  thanks you hnyman for the build.

wrtboy wrote:

it's been over 11 days and i like to share r48558 appears rock solid on the wndr3700v2.  thanks you hnyman for the build.

You are welcome. The WDNDR3700v1/v2/3800 has been so long supported that there are not many surprises regarding the router itself.  The possible problems may (and will) surface time to time from the new/changed functionalities introduced in the core Openwrt.

(Last edited by hnyman on 13 Feb 2016, 14:30)

- r48717 trunk: QoS support for "cake" qdisc changed slightly

For those, who are trying the cake qdisc for QoS: the original cake-aware "tc" as the tc-adv package from dtaht's repo was getting stale as that is still based on iproute2 v 4.1, while Openwrt trunk is already using 4.4.

So I made a comparison of the changes made by tc-adv vs. official tc, and created a simple patch for those changes. That got rid of the tc-adv module and simplified source changes a bit.

In case anybody is interested in specifics, the full changes are naturally visible in my source package of r48717, but can also be found at Github (both the new 950-cake patch for iproute2 and as background information the original comparison of tc-adv to tc): https://gist.github.com/hnyman/b35bf1e2501ad7dd08d4

(Last edited by hnyman on 13 Feb 2016, 14:38)

- r48730 trunk: Add support for 'adblock' with LuCI interface

Dibdot has written a nice ad-blocking module for which I have created a simple LuCI interface. I added those to the trunk build.
(The ad-block is initially disabled, so you need to enable it to use the functionality.)

EDIT:
For the r48735 build I backported adblock also to CC15.05.

(Last edited by hnyman on 17 Feb 2016, 13:21)

hnyman, thank you for the ad-blocking addition this is a very helpful feature.

I have a question though: I am using my WNDR3700 as a DHCP server and DNS forwarding/caching server on my LAN. This mean that a have a different router that is connected to Internet. The WNDR3700 only has a LAN side. When I enable the ad-blocking feature I see the following Log entries:

Wed Feb 24 09:36:29 2016 user.notice adblock[6544] info : domain adblock processing started (0.80.0, 15.05-r48749, 24.02.2016 09:36:29)
Wed Feb 24 09:36:29 2016 user.notice adblock[6544] error: LAN only (lan) network, no valid IPv4/IPv6 wan update interface found, rc: 125
Wed Feb 24 09:36:29 2016 user.notice adblock[6544] info : domain adblock processing failed (0.80.0, 15.05-r48749, 24.02.2016 09:36:29)

It seems that the ad-blocking process requires a WAN interface to update. Can this be made working in any way with just a LAN interface?

Thx

Romulus wrote:

I have a question though: I am using my WNDR3700 as a DHCP server and DNS forwarding/caching server on my LAN. This mean that a have a different router that is connected to Internet. The WNDR3700 only has a LAN side.
...
It seems that the ad-blocking process requires a WAN interface to update. Can this be made working in any way with just a LAN interface?

You should ask that from the adblock's author in the feature-specific thread: https://forum.openwrt.org/viewtopic.php?id=59803

But I do not think that it is possible. Adblock uses three different components, first it sets ad sites DNS info to the special backhole, then it sets firewall rules to block ads (handle the traffic by routing it to the blackhole address) and finally sets up an extra www server at the router (to provide 1 pixel image instead of ads). The DNS server and the www server can be any computer in LAN, but adblock's new firewall rules need to be applied in the router taking care of NAT and outgoing traffic.

First of, thanks for the great build you share with us!

In light of the recent hosting issues, would it be possible for you to include tcpdump into your trunk builds, please? When testing I often need to create packet captures to figure out whether things work as expected, but for the trunk builds this is a bit tricky... Currently I can not install tcpdump at all since the repository seems elusive, but even in the normal state this was tricky enough, due to the race with trunk's build-bot. Often I only managed to install your builds in the evening when the trunk packages where already newer and did not want to install any more. I know, I should install earlier, but due to $DAYJOB I only work on this in my spare time..

Anyway, thanks again for sharing your great builds with us.

Best Regards

moeller0 wrote:

In light of the recent hosting issues, would it be possible for you to include tcpdump into your trunk builds, please?

It adds 256 kB to the image, so for WNDR3700v1 the free space gets limited by a fifth. I think that I will not include it the default build, but I did include it for the 48808 build.

Btw, based on https://forum.openwrt.org/viewtopic.php … 56#p312756 there seems to be an unofficial copy of the Openwrt downloads at ftp://ftp.stw-bonn.de/pub/openwrt/snaps … x/generic/ containing the early-February stuff (r48648)

(Last edited by hnyman on 27 Feb 2016, 11:11)

hnyman wrote:
moeller0 wrote:

In light of the recent hosting issues, would it be possible for you to include tcpdump into your trunk builds, please?

It adds 256 kB to the image, so for WNDR3700v1 the free space gets limited by a fifth. I think that I will not include it the default build, but I did include it for the 48808 build.

Btw, based on https://forum.openwrt.org/viewtopic.php … 56#p312756 there seems to be an unofficial copy of the Openwrt downloads at ftp://ftp.stw-bonn.de/pub/openwrt/snaps … x/generic/ containing the early-February stuff (r48648)

Hello Hnyman,

thanks a lot, I had not looked at tcpdump's size (and I had completely forgotten about the v1's limits), sorry. I fully agree that this is a bit large for unconditional inclusion...

Thanks for the special build!

Best Regards
        M.

r49007-20160316  (opkg downloads packages from CC15.05.1 directory)

Now that Chaos Calmer 15.05.1 has been released, I have update my CC15.05 build to download opkg packages from the .../15.05.1/... directory and to reflect 15.05.1 in the version information.

   OpenWrt Chaos Calmer 15.05.1 r49007 / LuCI for-15.05 branch (git-16.076.51324-e8b6cbb)

(Note that the official 15.05.1 released today is actually already rather old and is based on r48532. It was compiled at the end of January, but the release was delayed by the server hardware problems at the Openwrt main site. My r49007 build contains several post-15.05.1 fixes. Sadly also the kernel in the CC15.05 source repo has also been updated from 3.18.23 to 3.18.27, so the kernel packages in the 15.05.1 repo are incompatible with my current build :-(

Looks like Openwrt is abandoning svn and is moving completely to git. The old svn.openwrt.org source repo is not accessible any more. Thus I have simplified my build scripts to be git-only since r49035.

Just wanted to say: thank you very much for your openwrt builds. I have been using them for quite some time now and thought it was time to let you know your work is much appreciated!