How to upgrade VICIdial safely

A VICIdial upgrade is not one command. It is a sequence: snapshot, back up the database, pull new code, run the schema upgrade, restart the dialer, then test. Skip a step and you can end up with code that expects database columns that do not exist yet, which breaks the agent screen and the dialer at the same time. The good news is the sequence is short and well-trodden. Follow it in order.

Snapshot before you touch anything

The first move is a full Server snapshot. A point-in-time image of the whole box means a bad upgrade is a one-click revert, not a rebuild. Then dump the database with mysqldump so you have a separate, restorable copy of the data even if you only need to roll back the schema. Two safety nets, because an upgrade that goes sideways during business hours is expensive.

stateDiagram-v2
  [*] --> Snapshot
  Snapshot --> DBBackup
  DBBackup --> SvnUpdate
  SvnUpdate --> DBUpgrade
  DBUpgrade --> Restart
  Restart --> Test
  Test --> Live
  Test --> Rollback
  Rollback --> [*]
  Live --> [*]

Pull the new code

VICIdial ships from Subversion. From the source directory you run svn update to pull the revision you want, then build and install the Perl scripts, web admin, and agent files. The dialer is a pile of Perl and PHP plus an Asterisk layer underneath, so a VICIdial code update and an Asterisk update are separate decisions. Move one variable at a time. If you are also bumping Asterisk, our Asterisk version compatibility post covers which pairings are sane.

Run the database upgrade

New code often expects new or changed columns. VICIdial ships an upgrade SQL script that applies those changes in order, bringing your schema up to match the code revision you just pulled. Run it once, against the live database, after the code is in place. This is the step people skip, and it is the step that breaks the agent screen with database errors when they do. Code and schema have to move together.

After the schema is current, restart the keepalive-managed processes so the new code is loaded. The cron-driven Keepalive cycle picks the scripts back up within a minute. Reload Asterisk if you changed dialplan or modules.

Test, then go live

Do not declare victory until you have dialed. Log an agent in, place a test call, set a Disposition, and confirm a Real-time report updates. Watch one full Agent session end to end. If anything throws a database error, your schema upgrade did not finish, and that is your cue to revert to the snapshot rather than debug live. Test on a low-traffic window so a rollback costs you minutes, not a shift.

For where upgrades fit in the bigger picture of running a box, see the installation guide, which lays out the install and maintenance lifecycle.

We handle the Asterisk side of this for you: we patch Asterisk CVEs and keep the platform secured so you are not chasing security updates on a dialer. You keep root SSH, so you choose when to move your VICIdial revision and run the schema script yourself. And because every box is a fresh Single tenant build, an upgrade on your server never touches anyone else's. See plans and pricing.