A backup strategy for a VICIdial server

A VICIdial backup is not one file. The system splits across three very different kinds of data, and a backup that grabs only one of them gives you a false sense of safety. You need the database, the recordings, and the configuration — captured on a schedule, kept for a sane window, and copied somewhere off the box. Miss any leg and a real recovery turns into a rebuild.

The three things to back up

First, the MySQL database. This holds leads, campaigns, dispositions, users, and every report. A nightly logical dump of the VICIdial database is the backbone of the whole strategy — without it you have lost your leads and your history. Second, the Call recording files on disk, which depending on your Recording format (WAV/MP3) and volume can be the largest and slowest-changing data you hold. Third, the config: the Asterisk conf files, the VICIdial settings files, and your firewall rules, so a rebuilt box comes back configured rather than stock.

flowchart LR
  DB[MySQL dump nightly] --> OFF[Offsite storage]
  REC[Recordings rsync] --> OFF
  CFG[Config files] --> OFF
  OFF --> R{Restore test}
  R -->|works| DONE[Backup verified]
  R -->|fails| FIX[Fix and rerun]

Schedule and retention

Dump the database nightly, because it changes constantly and is small enough to back up in full every night. Sync recordings incrementally — only the new files — since the old ones never change. Capture config on every change or at least weekly. For retention, keep enough daily database dumps to recover from a problem you did not notice for a few days, then thin to weekly and monthly copies further back. Recordings follow your Recording retention policy: there is no point backing up audio you are legally required to delete. Set both windows on purpose, not by accident.

Offsite and the restore test

A backup on the same box is not a backup — when the disk dies, both copies die together. Push at least the database dumps and config to separate storage: object storage, a Server snapshot held by your provider, or another machine entirely. A full-disk Server snapshot is a fast way to capture an entire box, though it is coarse; pair it with logical dumps so you can restore a single table without rolling back everything.

Then do the part everyone skips: restore the backup somewhere and confirm it works. An untested backup is a guess. Load last night's dump into a scratch box, point a test VICIdial at it, and check the leads and campaigns are intact. The post-install checklist is a good template for verifying a restored box behaves, and the VICIdial install guide tells you which files a config backup must include.

What managed hosting handles

On VICIfast your box is Single tenant and you keep root SSH, so your own backup scripts run unimpeded and recordings stay on your machine. We provision in under 40 seconds and can snapshot the VPS at the infrastructure level, but the disciplined database-plus-recordings-plus-config routine — and the restore test — is yours to own. We make it easy; we do not replace your responsibility to verify a restore.

Back up all three, keep a copy offsite, and restore-test it before you need it. Pick a box with snapshot options on our pricing page and build your backup routine on top of it.