Luci not available anymore. SSH works


I have a Xiaomi Router 3G running openwrt 18.06. I configured it using Luci, and this worked fine. However, after a few days, Luci is not available anymore. SSH works and the router does its job, I just cannot access Luci (tried on multiple browsers, PC, etc.). Chrome gives me an error "ERR_CONNECTION_RESET" or "ERR_CONNECTION_REFUSED"

I have reset the router completely and now it happened again after some days of usage. How can I reactive Luci? I tried rebooting the router, reinstalling Luci, but none of those things work.

Any commands I can use via SSH to get Luci to work again?

Are you installing and/or running any additional programs?

/etc/init.d/uhttpd restart

1 Like

Thanks for the quick replies!

  • UHTTPD restart doesn't solve the problem...
  • I don't have anything custom installed (i did have some packages the first time I've set up the router, but I didn't install them the second time since I thought those might be the culprit).

The network is set up as follows:
Cable modem --> Xiaomi 3G OpenWRT router which has one 5G and one 2.4G network (with the same name). a second Xiaomi 3G OpenWRT router is connected to this one and acts like a dumb AP (with IP This second router is still available on Luci.

Confirm that you're not running out of resources that is causing uhttpd to hang or crash. CPU or memory would be the first to look at if your logs aren't giving you a clue.

netstat -anl will also show if there is a listener present.

1 Like

So, you do not use https, but just the plain http?
(SSL libraries / wrong certs etc. are a possible reason for strange LuCI connectivity problems)

What should I be looking for in the netstat -anl command?
75% of memory is free.

Just plain HTTP indeed. i also tried typing explicitly http//192.168.11.

# netstat -nlp | grep uhttpd
tcp        0      0    *               LISTEN      3019/uhttpd
tcp        0      0   *               LISTEN      3019/uhttpd
tcp        0      0 :::80                   :::*                    LISTEN      3019/uhttpd
tcp        0      0 :::443                  :::*                    LISTEN      3019/uhttpd

you should try :smiley:

1 Like

I don't have any listeners for uhttpd ... what is wrong?

opkg update; opkg --force-maintainer --force-reinstall install uhttpd
service uhttpd restart; sleep 5; logread -l 20
1 Like

thank you @vgaetera, this worked for me, at least with this i could access to Luci, BUT the root password didn't work anymore, not eaven changing ir from ssh with "passwd".

Finally did a firstboot and configured all again, but why could this happen?? i wasnt eaven at home, just arrived and no internet.

Are you running a router with 8 MB or less of flash?

See, for example,

thank you for your answer, i think that is not the case, got the same router as who started the thread, xiaomi 3G. It's got 128Mb flash.


is It ok "/dev/root" to be 100%??

Yes, /dev/root being 100% full is normal (it's the read-only squashfs part). There have been two bugs haunting predominantly NAND based devices (like yours) with kernel 4.14, affecting ubi and overlayfs. These bugs should be fixed in the master snapshots, but they didn't make it for 18.06.1.

Leaved home 9AM, everything working normally, arrived home nos 9PM, got no internet connection, can't acces yo luci, can't acces to

BUT i can acces all Google sites, ONLY google sites... what can i do? if i ifconfig & traceroute i see:

so router is there, but i can't access to It, why? what could have changed? nobody else lives at home lol

I didn't see proof you accessed a Google site, just proof that you can do a DNS lookup.

  • Do you assign Public DNS servers in your LAN via the DHCP Option No. 6?
  • What happens if you allow the traceroute to complete?
  • Can you ping

Well i didn't put any probe of that, but i can assure you that I can, YouTube, Google, Google drive, all of them work fine in chromium.

  1. I'm accessing to router by wireless, and no, there's no DNS configured, it's all in automatic. if I do it wired, happens exactly the same.

2.traceroute never completes, i think it loses trace. you can see it on the next pic.

  1. you see next pic.

by the way, you can see YouTube working in the background.

Show the output of route command

You default go Google on ipv6 means your ipv6 talked to Google dns I which is on ipv6
My client faced similar issue 3 month back

As far as I remember it was due to ipv6 only

Just disable ipv6 where ever you had left it enabled
Disable wan6 intergace too if
Make uhttpd listen on ipv4 only
Try this and let me know . I ll get back on this in evening
If you intentionally enabled ipv6 or have it's requirements then also tell here