Advanced setup : how to set a WAN disconnect timeout?

Fleur · February 2, 2021, 2:00pm

What is the reason behind this need though? I mean what will happen if you disconnect the wan after 5 minutes?

My WISP seems to disconnect/unauthenticate my link if unused, so I've to restart the router or the connection (disable/enable via /etc/init.d/network restart) to force link UP...
I can manage this without too much effort but my family not!

I can use this connection without any issue via Smartphone/Tablet/Notebook (double-boot Ubuntu 20.10 and Windows 10 20H2) but, of course, I can't share the link with my LAN (I don't want to use my notebook as a network bridge between the wifi and ethernet port)

WiFi seems to get disconnected and reconnected (or only re-authenticated) as soon as traffic doesn't pass through ...

Another option may be to monitor (with a continuous ping ?) the first hop and then when check /ping fails, restart the network interface.

Any idea on how to implement it?

Thkz
Fleur

vgaetera · February 2, 2021, 2:05pm

Fleur · February 2, 2021, 3:30pm

It seems exactly what I was looking for!

I was working on this script I've found on another topic (your suggestion is a better one and also manageable from the GUI):

Install packages

opkg update
opkg install hub-ctrl

Save connectivity checking script

cat << "EOF" > /root/wan-watchdog.sh
#!/bin/sh

#Fetch WAN gateway
. /lib/functions/network.sh
network_flush_cache
network_find_wan NET_IF
network_get_gateway NET_GW "${NET_IF}"

#Check WAN connectivity
TRIES="0"
while [ "${TRIES}" -lt 5 ]
do
if ping -c 1 -w 3 "${NET_GW}" &> /dev/null
then exit 0
else let TRIES++
fi
done

#Restart network
/etc/init.d/network stop
#hub-ctrl -h 0 -P 1 -p 0 # commented, originally used to disconnect the USB port used for tethering
sleep 1
#hub-ctrl -h 0 -P 1 -p 1 # commented, after 1 second reconnect the USB port ...
/etc/init.d/network start
EOF

Set the script as executable

chmod +x /root/wan-watchdog.sh

Add cron job

cat << "EOF" >> /etc/crontabs/root
* * * * * /root/wan-watchdog.sh
EOF

Fleur · February 2, 2021, 5:22pm

So finally I used it: watchdog -> sorry watchcat

opkg update
opkg install watchcat luci-app-watchcat
opkg list-installed | grep watchcat

GUI config web page @ http://[router-ip]/cgi-bin/luci/admin/services/watchcat
or
vi /etc/config/system

config watchcat
option mode 'ping'
option forcedelay '30' # after 30sec, if soft-reboot doesn't works then hard-reboot
option period '15' # if no ping for 15 seconds then reboot
option pinghosts '192.168.64.1' # host to ping to check wan connection, it's my first hoop, multiple hosts allowed!
option pingperiod '1m' # ping check remote $pinghosts every 1 min

@vgaetera How can I monitor/log how many times watchcat triggers a reboot?

I saw that the package install this stuff:
opkg files watchcat

Package watchcat (1-7) is installed on root and has the following files:
/etc/uci-defaults/50-watchcat
/lib/upgrade/keep.d/watchcat
/usr/bin/watchcat.sh
/etc/init.d/watchcat

and inside the check script /usr/bin/watchcat.sh, raw 62, there is some logging on ping lost:

logger -p daemon.info -t "watchcat[$$]" "no internet connectivity for $time_diff seconds. Reseting when reaching $period"

vgaetera · February 2, 2021, 6:30pm

https://openwrt.org/docs/guide-user/base-system/log.essentials#local_file_logging

Fleur · February 3, 2021, 3:42pm

HI @vgaetera
watchcat is installed and configured BUT it doesn't work.
No trace on log, no reboot on connection lost.

How can I check that's all ok?

trendy · February 3, 2021, 3:47pm

uci show system.@watchcat[0]

Fleur · February 3, 2021, 3:47pm

root@ArcherC7:/tmp/run# uci show system.@watchcat[0]

system.cfg08e014=watchcat
system.cfg08e014.mode='ping'
system.cfg08e014.forcedelay='30'
system.cfg08e014.period='15'
system.cfg08e014.pinghosts='192.168.64.1'
system.cfg08e014.pingperiod='1m'

trendy · February 3, 2021, 3:48pm

To whom does it belong?

Fleur · February 3, 2021, 3:52pm

It's my WAN first hop

root@ArcherC7:/tmp/run# route -n

> Kernel IP routing table
> Destination     Gateway         Genmask         Flags Metric Ref    Use Iface
> 0.0.0.0         192.168.64.1    0.0.0.0         UG    0      0        0 wlan0
> 192.168.64.0    0.0.0.0         255.255.192.0   U     0      0        0 wlan0
> 192.168.143.0   0.0.0.0         255.255.255.0   U     0      0        0 eth0.1

root@ArcherC7:/tmp/run# ping 192.168.64.1

> PING 192.168.64.1 (192.168.64.1): 56 data bytes
> 64 bytes from 192.168.64.1: seq=0 ttl=255 time=5.778 ms
> 64 bytes from 192.168.64.1: seq=1 ttl=255 time=12.342 ms
> 64 bytes from 192.168.64.1: seq=2 ttl=255 time=7.657 ms
> 64 bytes from 192.168.64.1: seq=3 ttl=255 time=6.202 ms
> ^C
> --- 192.168.64.1 ping statistics ---
> 4 packets transmitted, 4 packets received, 0% packet loss
> round-trip min/avg/max = 5.778/7.994/12.342 ms

trendy · February 3, 2021, 3:55pm

Maybe the connection is not lost on your first hop.
Use some internet IP, like 8.8.8.8 or 1.1.1.1

Fleur · February 3, 2021, 10:51pm

Change pinghost to 8.8.8.8 but nothing changed...
During connection loss, I can't ping the remote host (8.8.8.8) from both my notebook and an ssh terminal session on my ArcherC7 router

I think that watchcat isn't active, looking at crontab or at active process I can't find the monitoring process.

@trendy How can I verify that wd is ok ?

Fleur · February 4, 2021, 12:12pm

Hi @vgaetera, @trendy

I think that somethings goes wrong

root@ArcherC7:~# head /usr/bin/watchcat.sh

#!/bin/sh
#
# Copyright (C) 2010 segal.di.ubi.pt
#
# This is free software, licensed under the GNU General Public License v2.

My watchcat.sh script isn't the latest, it looks quite old indeed:

github.com

openwrt/packages/blob/master/utils/watchcat/files/watchcat.sh

#!/bin/sh
#
# Copyright (C) 2010 segal.di.ubi.pt
# Copyright (C) 2020 nbembedded.com
#
# This is free software, licensed under the GNU General Public License v2.
#

get_ping_size() {
    ps=$1
    case "$ps" in
    small)
        ps="1"
        ;;
    windows)
        ps="32"
        ;;
    standard)
        ps="56"
        ;;

This file has been truncated. show original

I'm using the latest official openwrt build:
Powered by LuCI openwrt-19.07 branch (git-20.341.57626-51f55b5) / OpenWrt 19.07.5 r11257-5090152ae3

trendy · February 4, 2021, 2:12pm

How did you understand that? The copyright year doesn't mean anything. What is the output of opkg find watchcat ?
Also change pingperiod 60 and period 300

Fleur · February 4, 2021, 2:34pm

I suppose this 'cause watchdog.sh files are quite different... (the master one has a reference to a 2020 copyright author)

Changed timeout to a longer one as suggested (check every 300seconds -> 5minutes, connection is lost if after 60 seconds there're no ping reply)

I've also executed some command:

root@ArcherC7:~# /etc/init.d/watchcat enable
root@ArcherC7:~# /etc/init.d/watchcat start
root@ArcherC7:~# /etc/init.d/watchcat status

Syntax: /etc/init.d/watchcat [command]

Available commands:
        start   Start the service
        stop    Stop the service
        restart Restart the service
        reload  Reload configuration files (or restart if service does not implement reload)
        enable  Enable service autostart
        disable Disable service autostart

Can I suggest a status option ?

trendy · February 4, 2021, 2:46pm

It's vice versa, check every 60 seconds, reboot if there is no reply for 5 minutes.
Connect to the device and check the pings with tcpdump -i wlan0 icmp and host 8.8.8.8
Disconnect the access point and wait until you have five missed pings.
Let us know if the device doesn't reboot for further troubleshooting.

Fleur · February 4, 2021, 4:09pm

I'm wrong, your right!

After today 6PM I'll give it a try and I 'll confirm (or not) that's working as expected.

If WAN connection lost, restart USB WAN Device - #17 by vgaetera

@vgaetera suggest ( and I think so) that pinging the default GW (something near your, but of course, outside of home/biz devices you manage) is a better option...

@trendy maybe ping 8.8.8.8 is a guarantee* like www.cisco.com ping 2 decades ago, but why do you suggest this one?

*guarantee because this service are offered with high-available, load balancing setup so you never get a false positive response (rebooting your router without connection loss)

trendy · February 4, 2021, 4:25pm

It's a google anycast address, so you'll get an answer for sure from the closest server.
It's used from other vendors as well for reach-ability tests.

vgaetera · February 4, 2021, 5:25pm

Unlike your ISP gateway which usually supports ICMP traffic, hosts on the Internet are not obligated to reply to your pings and some internet services explicitly block ICMP traffic.
Moreover, there's typically no point to restart the router when the Internet connectivity fails, but the ISP connectivity is fine.
The only valid reason to rely on higher level connectivity is a dynamic ISP gateway which might require scripting to keep up-to-date.

system · February 14, 2021, 5:25pm

This topic was automatically closed 0 minutes after the last reply. New replies are no longer allowed.