"SIP failure when DNS fails" or "internal phones don't work when internet is down" -- DIY solution

[Anxifer]

Yes I know, we have dnsmasq as caching dns-proxy installed on the system which should take care of the annoying problem, that you can't use your internal phones when internet is down, but it doesn't. At least not always Or even at all. I don't know. I personally never had an raspbx installation where internal sip registrations "survived" the internet link going down.

So what can we do?
First take a look at what we've got: dnsmasq
Which is a good thing because it's capable of using more than one hosts file.

So we "just" need to write a little quick 'n dirty shell script that creates this additional hosts file for us, which contains the addresses of our external sip servers and tell dsnmasq to use it. Simple, isn't ist? ;-)

OK, but before we start scripting we need some prerequisites:
1. nslookup (part of the dnsutils package)
2. rcconf (for managing system services the easy way)
3. daemon (to run our tiny script as a service)
=> apt-get install dnsutils rcconf daemon

Then we can put together our little "dns-resolve-robot", let's call it lookup_sip_servers and put it in /usr/local/bin, which first checks if a given external (!!!) dns server (e.g. google-dns) is up and then looks up the IPs of our external sip servers, echoing them and their respective hostnames into our additional hosts file.

So here we go:

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

#!/bin/bash SLEEPTIME=10s SLEEPMULTIPLIER=12 DNS_SERVERS="8.8.8.8" HOSTS="sip.easybell.de tel.t-online.de" TEMPFILE=/dev/shm/dns_answer HOSTFILE=/var/run/dnsmasq/hosts.dnsmasq while [ true ]; do COUNTER=0 for h in $HOSTS; do for d in $DNS_SERVERS; do if [ $(nc -zvn $d 53 2>&1 | egrep -ic "succeeded") -gt 0 ]; then nslookup $h $d >$TEMPFILE IP=$(cat $TEMPFILE | grep Address | awk '{print $2}' | tail -n 1) ALIASES=$(cat $TEMPFILE | egrep -ic "canonical name") if [ $ALIASES -lt 1 ]; then if [ $COUNTER -lt 1 ]; then echo -e $IP'\t\t'$h >$HOSTFILE (( COUNTER ++ )) else echo -e $IP'\t\t'$h >>$HOSTFILE fi else ALIASLIST=$(cat $TEMPFILE | grep "canonical name" | awk '{print $5}' | rev | cut -c 2- | rev) if [ $COUNTER -lt 1 ]; then echo -e -n $IP'\t\t'$h'\t' $ALIASLIST >$HOSTFILE (( COUNTER ++ )) else echo -e $IP'\t\t'$h'\t'$ALIASLIST >>$HOSTFILE fi fi fi rm -f $TEMPFILE done done for (( i=1; i<=$SLEEPMULTIPLIER; i++ )); do sleep $SLEEPTIME done done rm -f $TEMPFILE exit 0

|
OK, an endless loop ist definetely not the most soophisticatdt way to do it, but it gets the job done.
Originally I intended tu use multiple external DNS servers, hence the respective for-loop still being there.
EDIT: I don't know why parts of the code are not visible, but if you're interested in it just copy and paste the whole section into your local editor.
Our created additional hosts file then looks like this:.:

212.172.97.124          sip.easybell.de
217.0.23.100            tel.t-online.de ims.voip.t-ipnet.de ims001.voip.t-ipnet.de b-epp-001.isp.t-ipnet.de

|
Now we tell dnsmasq to use this addtional hosts file. Therefore we put a dnsmasq.conf in /etc/dnsmasq.d containing the following line:

addn-hosts=/var/run/dnsmasq/hosts.dnsmasq

|
While still having momentum we can now install our little script as a system (startup) service by puttin another tiny script, let's also call it lookup_sip_servers, in /etc/init.d:

### BEGIN INIT INFO
# Provides:             lookup_sip_servers
# Required-Start:       $syslog
# Default-Start:        2 3 4 5
# Required-Stop:
# Default-Stop:         0 1 6
# Short-Description:    lookup_sip_servers
# Description:          periodically checks DNS records of certain hostnames via nslookup and echos them
#                       to a custom hostz file
### END INIT INFO

PIDFILE=/var/run/lookup_sip_servers.pid
# LOG=/var/log/messages

case "$1" in

    start)
        if [ -e $PIDFILE ]; then
            echo -e "\nDaemon seems to be running already.\n
                    Use: '/etc/init.d/lookup_sip_servers restart' to restart."
        else
            echo -en '\nStarting lookup_sip_servers...'
            daemon -U -n lookup_sip_servers -X /usr/local/bin/lookup_sip_servers -F $PIDFILE
            while [ ! -e $PIDFILE ];do
                sleep 1s;
            done
            PID=$(cat $PIDFILE)
            echo -e ' done. (PID='$PID')'
        fi
    ;;

    stop)
        if [ -e $PIDFILE ]; then
            PID=$(cat $PIDFILE)
            echo -en '\nStopping lookup_sip_servers (PID='$PID')...'
            kill $PID
            while [ -e $PIDFILE ];do
                sleep 1s
            done
            echo -e ' done.'
        else
            echo -e "\nDaemon seems not to be running."
        fi
    ;;

    restart)
        $0 stop
        $0 start
    ;;

    *)
    echo -e "\nUsage: /etc/init.d/lookup_sip_servers (start|stop|restart)\n"
    exit 1

esac

exit 0

|
'and activate its execution at boot time with rcconf.

The end.

I hope somebody finds this to be useful...

[advocate]

I cannot help you with the script because I don't know anything about programming, but consider this:

You don't need dnsmasq to provide real IP addresses. If your internet is down, there's no need for a real IP address because you're not going to get through anyway. You just need dnsmasq to return SOMETHING so that Asterisk doesn't freeze. That something could be 10.10.10.10 for all you care.

That fact may make the script easier to write..