2

I'm trying to work out why my VPN won't connect when the command is initiated from a cron triggered script when that same script and command otherwise work as expected.

I have a remote server which uses a secure VPN tunnel (pptp) to connect to a database behind our firewall. It's quite a stable system (especially with the persistent option set), and generally runs without any issue. However, from time to time the connection from our ISP to our office drops out, and this disconnect is long enough to prevent the VPN tunnel from staying open.

I've set up a simple script to detect if the firewalled database is still available over the VPN and, if not, it attempts to reopen the VPN.

#!/bin/bash

DATE=`date`

HOST=10.1.2.1

PING_RESULT=`ping -c4 $HOST`
# gets the percentage of lost packets
PING_LOSS=`echo $PING_RESULT : | grep -oP '\d+(?=% packet loss)'`

echo "$DATE : Loss Result : $PING_LOSS"

# if 100% packet loss on the ping - assume connection lost
if [ "100" -eq "$PING_LOSS" ];
then
        echo "$DATE : Connection Lost"
        pon VPN_TUNNEL
        echo "$DATE : Restarted Connection"
else
        echo "$DATE : Connection OK"
fi

I've saved the script as /root/cron/pptp-monitor and set the permissions as -rwxr--r-- root root

The script works great when run manually (using sudo) - but the cron I've configured doesn't work properly:

*/5 *     * * *     root   [ -x /root/cron/pptp-monitor ] && /root/cron/pptp-monitor >> /var/log/pptp-monitor.log 2>&1

The monitor script runs - I see the log entries every 5 minutes - but the pon command doesn't actually seem to fire off.

In the syslog, I see this every 5 minutes:

Apr 17 08:45:01 bombur CRON[774]: (root) CMD (  [ -x /root/cron/pptp-monitor ] && /root/cron/pptp-monitor >> /var/log/pptp-monitor.log 2>&1)
Apr 17 08:45:14 bombur pppd[784]: pppd 2.4.5 started by root, uid 0
Apr 17 08:45:14 bombur pppd[784]: Using interface ppp0
Apr 17 08:45:14 bombur pppd[784]: Connect: ppp0 <--> /dev/pts/0
Apr 17 08:45:14 bombur pppd[784]: Modem hangup
Apr 17 08:45:14 bombur pppd[784]: Connection terminated.

The lines "using interface ppp0" to "connection terminated" are repeated 10 times before exit - indicating that the tunnel is trying to reopen unsuccessfully. Note - the network connection is fine by this time, and as soon as I run the command manually, it connects on the first attempt.

What did I miss that is causing the cron trigger to prevent the VPN connection?

2
  • did you find any solution ? – MKT Nov 19 '14 at 10:43
  • @MKT not as yet – HorusKol Nov 19 '14 at 22:10
2

You need to declare the PATH variable in your scripts, afterwards it will work. At the top of the files (both the crontab and the script that's going to be executed):

SHELL=/bin/sh
PATH=/usr/local/sbin:/usr/local/bin:/sbin:/bin:/usr/sbin:/usr/bin
1
  • Why not #!/bin/sh? – Donarsson Jan 26 '15 at 15:47
0

I'd prefer run this:

# Here we can specify which user do crontab use.
# And, we can change $PATH by edit this file directly

vi /etc/crontab

# I'm using centos, restart cron daemon

service crond restart

Than run this:

# Edit crontab directly, we can't specify the user.
# And I'm not sure changing $PATH is available or not.
# The default value of $PATH may be `/sbin:/bin:/usr/sbin:/usr/bin'

crontab -e

Because pppd will run pptp, but pptp was located in `/usr/sbin/', so set $PATH can be necessary.

1
  • Sorry but this is simply wrong. There is no reason to use /etc/crontab. When running contab -e, you are editing a file directly, it's just your user's personal crontab, so the username is assumed. If you want to do this for another user, run sudo -u username crontab -e. As for $PATH, there is absolutely no difference between using the system-wide /etc/crontab file or the local one. – terdon Feb 2 '17 at 8:54

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.