Updated: 25th August 2015
A problem with using old technology alongside newer systems is that updates to the newer systems can break things silently. This is a followup to my earlier article on using vgetty with a new voice modem.
Vgetty stopped working after the server was upgraded to Debian 8. The reason was that the old /etc/inittab line:
s0:23:respawn:/usr/sbin/vgetty ttyS1
no longer had any effect. To control the service under systemd, I created a new file /lib/systemd/system create vgettyS1.service containing:
[Unit]
Description=vgettyS1 daemon
After=network.target
[Service]
Type=simple
User=root
Group=root
ExecStart=-/usr/sbin/vgetty ttyS1
Restart=always
RestartSec=0
KillSignal=SIGHUP
KillMode=process
[Install]
WantedBy=multi-user.target
Then, the service can be started using:
systemctl start vgettyS1.service
The command to set the service to run at boot is:
systemctl enable vgettyS1.service
Hopefully, the voicemail service will now continue to work.
Why persist with using POTS (Plain Old Telephone Systems)? The current system is working, so why change? The capital cost of moving to IP phones is significant. The recurrent cost is uncertain: service providers are keen to offer a free value-added service package (e.g. voice mail, call waiting) for two years, but don't guarantee the pricing afterwards. Then there is loss of control - using the service provider's voicemail, potentially sensitive messages are stored outside of our control. Our current system is a Panasonic KX-TD1232 "Digital Super Hybrid" system with 20 extensions and 6 external lines (3 in a hunt group, 3 direct dial in [DID]). It also supports an entryphone and door release. Fax and voicemail services are provided by PC modems, delivering directly to internal email - service providers use their voice-to-email services as a selling point, as though it is an innovative service when we've had this for years with this old POTS solution. Finding a IP phone solution that matches all of the current features at a reasonable capital and recurrent cost looks like a challenge. Perhaps it is better left until we've migrated to IPv6.