Tnx for the help. Im using david's latest build. On the 1st day. I'd encountered disconnection issues with my iphone6+, but on my ipad the wifi is stable. I was thinking of using the official lede build. However, last night the wifi is now stable i think. Will try a few more days if i still will have wifi issues.

I have the same problem with pages occasionally failing to load in Safari and Chromium, I've never seen it happen on my mobile devices, I'm not on Davids build, but I've seen it in LEDE 17.01.2 and Trunk, it wasn't present in 17.01.0, I only briefly used 17.01.1 and can't remember if it was present there or not, using a wrt1200 v1. (links removed in quote so I could post)

So from what you and 2 others recently posted, it affects @david's builds (built from development/trunk branch) and official LEDE 17.01.2 and it affects @beachbum's 1200AC V1.  I also have a 1200AC V1, but have never encountered this issue.

I'm assuming those that this affects is in the minority or there would be many more complaints.

Does DD-WRT use dnsmasq like LEDE?  If so, what version does it use?  Also, have you compared /etc/config/dhcp in LEDE vs in DD-WRT (assuming it exists in DD-WRT and uses dnsmasq -- not very familiar with DD-WRT, sorry)?

The main thing I don't understand is why this issue affects only a few.  What OS are those of you with the issue running?  Have you tried manually setting the DNS server from inside the OS (to bypass the router) and see if that fixes it?

Thanks,

David

(Last edited by davidc502 on 5 Aug 2017, 18:49)

Hi,

Can someone say how can i make sure the openvpn log size ,(which is in my case inside:
/var/log/openvpn.log )

bigger ? it is always rotating and give me max of ~20 lines...

tnx !

Yes DD-WRT uses dnsmasq like LEDE. Not sure what version. Some syslog on DD-WRT shows dnsmasq being loaded, but not version. OS in question is Windows 10 Pro in 2 cases, one Ethernet attached to the router the other wireless 5GHz. Have not set a local DNS server in the OS and tested. All good ideas/points. If I find out the dnsmasq version in DD-WRT I will update. Brainslayer has his own way of doing things and not necessarily layed out the same as "normal" Linux.

I have been running davidc502's snapshots on my wrt3200acm for about a month now. For the most part it has been stable. Lately, I've had to reconnect the lan interface through Luci to get it to work. The weird part is that it seems to affect devices differently. I am unable to access the Luci Config Page from most machine (OSX, android, IOS), but was able to gain access through a wifi connected windows 10 machine. Directly connecting a machine to the router doesn't seem to make a difference. This happened at least 3 times since I upgraded 4 days ago.

Additionally, I somehow managed to flash both partitions with the latest version of LEDE. Is there a way to revert the current partition to the Factory image or to at least dd-wrt?

thanks

Sent.  Let me know how it goes.

Nothing abnormal in that. When you flash from either OEM, Openwrt or LEDE, you always flash the other partition, so that the current one stays as the fallback option. If you have flashed LEDE at least twice, you now have LEDE on both partitions. (no idea about the dd-wrt behaviour)

LEDE has currently slightly different image verification system, so the OEM firmware will give a hiccup for the sysupgrade.

You need to first download the image to the router (with wget, curl, aria, whatever...) and then use from console sysupgrade with "-F" or "--force" switch to force using the image even when if fails the image format check. You can't do that from LuCI.

(you may also add "-n" to prevent storing the current LEDE config to /tmp/syscfg partition, from where it might be inadvertently found when you next time flash LEDE)

sysupgrade -n -F imagename

For those having internet issues and are using dnscrypt-proxy:  it is my experience that wan glitches cause dnscrypt-proxy to hang, preventing lookups effectively resulting in no internet access.  Some of the symptoms listed in this thread don't fit this model but I'm convinced this tool needs some form of hotplugging (or a more robust startup method).  Below is an extremely beta version of a hotplug script you may want to try.

Without the hotplug my router will not come up with internet access if both cable modem and router are powered on at the same time (power failure for example).

Before adding the hotplug script to your router:  If you see problems with dns or wan access, you may want to execute the kill command in the ifdown section, then restart dnscrypt-proxy.  If this clears things up then this may be your issue.

Be warned!!!
- pgrep is required (to run the script as-is -- if your build doesn't include it you'll need to modify the syntax or install it)
- router startup timing (init.d scripts vs. hotplug enabling) may result in race conditions (thus the extreme beta nature)
- as written this only operates on wan interface -- if you use wan6 you'll need to modify the initial exit test
- after adding the hotplug script you'll likely want to robustly test:  power down cable modem/DSL/etc AND router, start both together; disconnect WAN port of router then reconnect; perform "ifdown wan ; sleep 10 ; ifup wan" ; cycle power on router alone; etc -- and make sure everything plays nicely
- EDIT: I don't think this handles the case of losing internet connectivity on the WAN port (wan interface is up but cable modem/etc isn't ready yet) -- any ideas on how to monitor/detect/hotplug that are welcome

Obviously all the above would need to be fixed for an official hotplug script.  Name it 50-dnscrypt-proxy, make it executable and place in /etc/hotplug.d/iface.

#!/bin/sh

[ $INTERFACE = "wan" ] || exit 0                                # only operate on WAN state changes

case "$ACTION" in
  ifup)
    [ $(ls /etc/rc.d | grep dnscrypt | wc -l) -gt 0 ] || exit 0                    # only start dnscrypt-proxy if enabled
    [ $(pgrep dnscrypt | wc -l) -gt 0 ] && for DP in $(pgrep dnscrypt); do kill -9 $DP; done    # ensure no running processes before starting
    logger -t dnscrypt_wan "Starting dnscrypt-proxy due to $ACTION of $INTERFACE ($DEVICE)"
    /etc/init.d/dnscrypt-proxy start
    ;;
  ifdown)
    [ $(pgrep dnscrypt | wc -l) -gt 0 ] || exit 0                        # only stop dnscrypt-proxy if running
    logger -t dnscrypt_wan "Stopping dnscrypt-proxy due to $ACTION of $INTERFACE ($DEVICE)"
    for DP in $(pgrep dnscrypt); do kill -9 $DP; done
    ;;
esac

(Last edited by InkblotAdmirer on 5 Aug 2017, 17:22)

Got tired of fighting these fires and abandoned in favour of unbound. Speed, and plays nicely with dnsmasq and adblock.

@starcms
Just to keep me honest. When I setup LEDE I configure the DNS servers to be used on the LAN interface. I believe this is correct. If not how should be be configured. Was trying to think of things I could be mucking up and that was one of them. Does get a little confusing as you can set DNS servers on the WAN interface as well.

Have not found the dnsmasq version in dd-wrt yet by the way.

(Last edited by bill1228 on 5 Aug 2017, 19:16)

DNS set in WAN Advanced Settings
dnsmasq -v

@Villeneuve
Thanks for the info. Maybe that is my slow at times DNS lookup. Somewhere I saw just the opposite somewhere on the Openwrt site.  Am running DD-WRT r33006 right now, but when I get a chance I will reinstall LEDE and report. Right now I have teen girl doing her Internet thing and wife doing email. Having the wrath of one is bad, both is too the dog house in the backyard.

(Last edited by bill1228 on 5 Aug 2017, 20:56)

Ok. back on LEDE r4651 without the driver patch. Did the DNS on the WAN interface, not intuitive obvious to us neophytes. Working, but tooooooo early to know it this addresses my browser DNS lookup issue. Guessing it will.
Will update IF I still have failures, in other words no news is good news.

(Last edited by bill1228 on 5 Aug 2017, 22:43)

As long as you have your router NTP client set up to always sync the time (does so by default), and have two different dnscrypt resolvers set in the dnscrypt config file (in case one goes out), you shouldn't have any issues.  Also very important is to follow the dnscrypt instructions here: https://wiki.openwrt.org/inbox/dnscrypt and most importantly add an entry for     

list server                 '/pool.ntp.org/208.67.222.222'

below the entries for list server added for dnscrypt in your dhcp config file so that your router can sync the time without using dnscrypt.  Otherwise, it'll just get stuck in a loop.  Can't use dnscrypt because time is wrong ---> cant sync time because of no dns access --> can't use dnscrypt because time is wrong  --> for infinity and beyond

If you turn on cable modem and router at same time, router won't be able to sync time for a few minutes, so it won't be able to connect to the dnscrypt resolver(s) properly.  However, once internet is available, NTP should sync the time and dnscrypt will begin working (assuming you following the directions and set it up so that the NTP server can access the internet without using dnscrypt; if you skip that one step you'll have nothing but headaches).  NTP and dnscrypt auto-retry in the background automatically and continuously until they succeed.

(Last edited by starcms on 6 Aug 2017, 04:43)

There isn't really an option in LuCi on what interface to configure the DNS servers to (I don't think).  If you don't change any major default settings, you should be fine.  The default DNS servers come from your ISP, so that would be coming via WAN, if that is what you are asking. 

Updated/Corrected:

@ all those having problems recently described about the dns server being slow, not resolving right away, etc, please read this

However, I just found something comparing the dhcp config file (/etc/config/dhcp; which controls dnsmasq/DNS as well as dhcp) that is currently included with LEDE (and @david's build) vs. mine and one thing caught my eye:

My config file has that set to 0.  The default now has it set to 1.  Try setting it to 0 and rebooting and see if that helps.  I bet it will!!!   

This would explain why only some are experiencing the issue.  Because originally, the default setting was 0.  Those of us who kept upgrading and saving/restoring our config, kept it that way.  However, it was changed in the past few months to be set to 1 by default so anyone with a fresh install started experiencing the issue.

Addition: https://wiki.openwrt.org/doc/uci/dhcp also shows the default of nonwildcard to be 0.  I don't know when the entry for nonwildcard to be 1 was added (some stupid commit probably in the last 3-4 months or so), but I bet that's your issue.

Also compare to the config file in DD-WRT and see what differences you see.  I bet DD-WRT also has it set to 0 and that is why there is no issue with DD-WRT.

This also explains why the issue was not present in official LEDE 17.01.0, but is in official LEDE 17.01.2 (and in the snapshot/trunk branch).  There must have been a commit which changed the value in the config file of nonwildcard from 0 to 1 for some idiotic reason sometime after LEDE 17.01.0.

(Last edited by starcms on 6 Aug 2017, 05:08)

Also, the driver patch has nothing to do at all with any of these dns problems a few are facing.  The driver patch disables amsdu which fixes ping issues that definitely exist on all 1200/1900 models without the patch.  It is highly, highly recommended and will be included in @david's next build.

(Last edited by starcms on 6 Aug 2017, 04:52)

So what exactly did you change vs how you had it configured before?  By default, LEDE automatically fetches the DNS servers from WAN/your ISP.

(Last edited by starcms on 6 Aug 2017, 05:02)

@starcms

My setting is for nonwildcard is set to 0 (I always restore my config files) and I have experienced the dns resolving issue. Though i do want to add it does not occur all the time. About a month ago there would be times that the loading issue was quite apparent, but then it would go away for days.

To be honest i haven't noticed it for maybe a week.

Is it possible in your case it's caused by your ISP's DNS servers (or the dnscrypt resolver you are using)?  And not the router or LEDE?  Possibly we have multiple causes of a very similar issue?

(Last edited by starcms on 6 Aug 2017, 05:12)

Hi,

I'm using the latest david build with the new wifi driver ,
since then i noticed 5G is not exceeding 1mb of download ! it is weird...
2.4 GHZ looks good ... i tried everything i thought its channel issue or something else...

but after all things i checked i assumed its firmware issue... is anyone experiencing the same issue ?

tnx to @david and all of you.

In practice... this doesn't appear to work all the time.  If the wan interface is up on the router but the cable modem does not yet provide internet access, dnscrypt-proxy hangs and never recovers.  This has happened to me on multiple routers with different types of wan connections and since it's rare-ish I haven't gotten to the bottom of it yet.

As an interim solution I'm using this to check resolving, you can use any command that requires a DNS lookup:

ping -w 3 <url_on_wan> > /dev/null || { <commands to execute when ping doesn't work, like restart dnscrypt-proxy> }

There is a package pingcheck for a more formal implementation but I'm just doing something like the code snippet above some time after boot -- measure how long it takes for internet access to come up, delay something a bit more than that after initial boot, then execute the command.

@starcms
I use Luci for configuration, keeps me from mucking up config files.
On the LAN interface there is a obvious place for adding DNS servers. On the WAN interface  under advanced settings there is a place for "use custom DNS servers". Have to make a selection before this window opens for input. Not as obvious as the one on LAN. I read somewhere on a Openwrt site that you should use the LAN interface for DNS settings which is what I was doing. This is wrong as now DNS lookup is faster and no more of the cannot lookup messages when browsing.  Just goes to show you do not always get correct info on forums.

From past experiences with slow ISP provided DNS servers I have for years configured router firmware for DNS servers I want to use. Currently for me, Level 3 DNS servers are working well. So that is my reason for the custom configuration. 

Have not investigated the DHCP option you point out in another post.

Thanks for your assistance to me and others that are having a similar experience. Wonder if any of them are doing the same configuration I was.

UPDATE
Made the DHCP option setting to 0 that you suggested above now. Things seem to be running fine, better then has in recent tests. Was going back to DD-WRT because of issues, but now I think that LEDE is my favorite as it was for many builds.

(Last edited by bill1228 on 6 Aug 2017, 16:49)

Hey all,

I'm having some trouble with the firmware and I'm hoping someone might be able to help me troubleshoot it. I recently purchased a WRT3200ACM. I originally installed the latest version of OpenWRT and then upgraded to Davidc502's firmware (the July 29th version). Everything was working fine, until I upgraded to the July 30 firmware. After that, I had no DNS resolution on my local network. The router can ping and access the Internet without a problem, but none of the machines on the LAN were able to resolve hosts.

Even downgrading back to the original OpenWRT firmware didn't fix the problem, so I re-flashed the stock firmware and everything was fine. I just tried flashing Davidc502's factory image from stock and ended up with the same problem. I did notice that my computer, which is connect directly to the router via an Ethernet cable, listed the DNS server as 192.168.1.14, even though the router's address is 192.168.1.1, but I'm not sure how to get the router to give out the proper address. Does anyone have any idea how to fix this?

UPDATE: I tried setting  "option nonwildcard    '0'" in /etc/config/dhcp and setting custom DNS servers for the LAN interface, but it did not fix the issue.

(Last edited by access on 6 Aug 2017, 22:05)