Hi all. I just put this on my 1900ACSv2 a few days ago and today was attempting to setup an openVPN server as per the OpenWRT Openvpn basic setup guide.
It fails when attempting to delete the /etc/easy-rsa/pki directory with this-
# Remove and re-initialize the PKI directory
easyrsa --batch init-pki
Has an I/O error, exactly as per this thread-
and
When re-doing the whole openVPN basic guide setup instead using /tmp, I found I could copy everything to that pki directory except the reqs and private directories (which after the I/O error didn't exist, and threw the I/O error attempting a cp -fr of those 2 directories from tmp).
So I did a factory reset, and sure enough, there is a pki directory with empty reqs and private directories in the default OpenWRT factory default image, and it does not like letting one delete those I presume.
Don't know if this is specific to this (yesterday/recent) builds, or a general OpenWRT issue yet (still on a steep learning curve here).
Still fiddling here, wondering if I should attempt this OpenVPN(server) setup via luci instead of following the OpenVPN basic guide.
edit: If I followed every step in that guide except the "init-pki", it seems to create the files required and starts the server.
I have an issue with name resolution after upgrading to r11398. I've been using dnscrypt-proxy v2 for a long time. After upgrading to r11398, things work fine for a number of hours. Then, name resolution simply starts timing out:
root@OpenWrt:/etc/dnscrypt-proxy2# nslookup cnn.com
;; connection timed out; no servers could be reached
root@OpenWrt:/etc/dnscrypt-proxy2#
I didn't see anything in the dnscrypt-proxy log or the system log which indicates why I am having this problem. I needed to roll back to r11266. Any ideas for what is causing this issue?
I followed the instructions you listed but I am still having the same issue. Nothing is being resolved before or after I made the changes from your instructions.
Here is what I did:
I reset my router to factory settings from within Luci
I flashed the latest firmware " *r11398"
I edited the .toml file and the following changes
listen_addresses = ['127.0.0.1:5300']
server_names = ['cloudflare']
restarted /etc/init.d/dnscrypt-proxy restart
Then I edited /etc/config/dhcp and added list dhcp_option '6,192.168.1.1' under LAN and added ist server '127.0.0.1#5300' under "config dnsmasq"
restarted /etc/init.d/dnsmasq restart and I got a message that it was failing ( see attached screenshot)
But still nothing was being resolved. I restarted my clients and restarted the router but it did not help.
Even if I remove the line (list server '127.0.0.1#5300') from dnsmasq and restart the service it did not make any difference.
In fact, I've never used the firmware provided by Linksys for anything other than flashing OpenWRT.
When using that image, I get significantly better 5ghz performance than I did using David's build. I did not test performance for the 2.4ghz radio, I may or may not have had similar performance patterns on the 2.4ghz radio, but I never tested. When I say "significantly better performance" I am referring to the crocodile pattern mentioned by @sunarowicz. Not only do I see a crocodile pattern with David's build, but an increase in lost packets, and overall throughput is much lower as well.
I don't know why my 5ghz performance is significantly better with the stable mainstream OpenWRT image (linked to above) than it is with Davidc502's build. I was speculating when I said that it might be that a new driver provided by David's build doesn't work well for my router. I thought that one of the features that David provides is the latest drivers. However, @davidc502 said this:
This is the same Wifi Driver. Expect Zero difference between the 2. You are welcome to try it out, but wifi is sketchy for some.
And I think that means that he doesn't pack a newer driver than the firmware that is linked to above.
Like @sunarowicz, I have very little (like zero) interference from other 5ghz routers in my neighborhood. My problems could not have been caused by that kind of interference. I basically have line of sight with my only 5ghz client (It's a server that I want to keep very fast).
I guess it could have been wild coincidence that I happened to have worse performance with David's build than with the stock openwrt (again, linked to above) build. But it did start when I installed and ran David's build, and stopped immediately when I switched back to the stock/stable openwrt image (you guessed it, linked to above).
How does one go about forking this build to the EA9500v1.1? The only difference is the 1900's and 3200's are Marvell and the the EA9500's are BCM. There is a build for it but It's what I deem bloated and not really what I am looking for in a OpenWRT build. Does anyone have any insight on this at all? Thanks in advance.
Sadly no it doesn't. The best you can do is use one of the radios for backhaul and WDS. Then use the other radio for a distant AP. Unless you hardwire them all together.
Thanks for clarifying this. I was considering doing this, but you just saved me some time. Maybe I'll look into moca or power plug with a second cheapo router to kind of fill in the weak spots.
Happy to hardwire the AP's together for the backhaul.. The house has Ethernet ports in every room so this would be the prefered option for the mesh solution.
The open source driver receives and forwards the mesh information element to the proprietary firmware. It is also capable of transmitting this information to userspace. Thus, to iw or hostapd it makes it look like the device is capable of running mesh point. However the real processing is done in the closed source firmware, the binary files in /lib/firmware/mwlwifi/. There, the information gets blackholed and discarded because the firmware doesn't know what to do with it - it is not coded to support it. There once was plans, but now there's none - as alluded to by @anomeome.
Please I report for dnsmasq in wrt1200 v1 router
Has some files missing to start after reboot
This is from 3 months after release new kernel 4.19. Until today
I thinking this problem happened because kernel 4.19
Please check all required files for dnsmasq into new builds
Or can u add unbound DNS instead of dnsmasq it has small files low space
And also u can add softether version 5 it is working good with internal disk device space
Hope you get those and tested dnsmasq or adding unbound instead it
Thank you David
Have a nice work day
Best regards,
Eng. Tarek Herik.
The DNSCrypt-proxy2 is working well without issues, and now myself being on fibre 40/10 i am getting a nice speed as well
root@WRT32X:~# speedtest-netperf.sh
2019-11-08 13:56:26 Starting speedtest for 60 seconds per transfer session.
Measure speed to netperf.bufferbloat.net (IPv4) while pinging gstatic.com.
Download and upload sessions are sequential, each with 5 simultaneous streams.
.............................................................
Download: 32.22 Mbps
Latency: [in msec, 61 pings, 0.00% packet loss]
Min: 18.281
10pct: 18.680
Median: 19.829
Avg: 20.296
90pct: 20.953
Max: 45.960
CPU Load: [in % busy (avg +/- std dev), 59 samples]
cpu0: 8.4 +/- 10.7
cpu1: 11.3 +/- 9.9
Overhead: [in % used of total CPU available]
netperf: 1.9
.............................................................
Upload: 5.02 Mbps
Latency: [in msec, 61 pings, 0.00% packet loss]
Min: 17.779
10pct: 18.017
Median: 19.121
Avg: 19.514
90pct: 20.974
Max: 29.012
CPU Load: [in % busy (avg +/- std dev), 59 samples]
cpu0: 6.1 +/- 10.7
cpu1: 3.6 +/- 5.7
Overhead: [in % used of total CPU available]
netperf: 0.3
As for the WebGUI its working smooth as anything too no more having to spam refesh button time in and time out.
Nice work @davidc502 on this build. As always will let you know if there is any issues, going to use your builds for a while whilst i am fixing my internal builds (Yep still having issues with LUCI-SSL dependancy even using your config).
here is log
Fri Nov 8 16:24:33 2019 daemon.crit dnsmasq[5061]: cannot read /usr/share/dnsmasq/trust-anchors.conf: No such file or directory
Fri Nov 8 16:24:33 2019 daemon.crit dnsmasq[5061]: FAILED to start up
another problem i saw in latest build in front page is local time showing extra 2 hours after current time
