Monitoring recording storage so you never run out of disk

A VICIdial server that runs out of disk space does not warn you politely. It stops writing Call recording files, database writes fail silently, and in some configurations Asterisk itself becomes unstable. The good news is that disk usage grows at a predictable rate once you know your call volume, and there are clear signals to watch well before you hit the limit.

How recordings accumulate

VICIdial writes recordings as WAV files on the server during the call, then optionally converts them to MP3 for archival. A raw two-channel WAV at standard telephony quality runs roughly 1 MB per minute of audio. On a box where 30 Agent seats are each averaging four hours of connected time per shift, that is around 7 GB per day before any compression. MP3 conversion at a typical bitrate brings that down to roughly 700 MB to 1 GB per day, but compression itself uses CPU during the conversion window.

On a VPS with 100 GB of disk, 1 GB per day means the recording partition fills in roughly three months if nothing is cleaned up. A busier box or a larger agent count shortens that window significantly. On a Single tenant managed box, recordings stay on the customer's own server — there is no shared storage that absorbs the overflow.

Where to check disk usage

The Server Versions page inside VICIdial's admin section lists disk space alongside server name, IP address, current load, and Channel count. This gives you a quick at-a-glance view of remaining capacity without needing to SSH into the box. It refreshes on page load so you can check it between shifts as part of a daily routine.

For a more granular breakdown at the operating system level, the df command shows filesystem usage by mount point. On most VICIdial installs the recordings directory sits on the root filesystem or a dedicated partition. Watch the Use% column — once it crosses 80 percent, you have enough time to act but not unlimited runway. At 90 percent you should be actively archiving or deleting.

Setting a retention policy

A retention policy answers one question: how long does your business or your regulatory environment require recordings to be on the server? Common requirements range from 30 days for internal quality monitoring to 90 days for compliance-driven industries. Once you have that number, you can build a deletion or archival script that runs nightly and removes files older than the threshold. On a VICIdial box this is typically a cron job that finds files in the recordings directory older than N days and removes or moves them.

flowchart TD
  A["New recording written to disk"] --> B["Nightly cron runs"]
  B --> C{"File older than retention window?"}
  C -->|No| D["Keep file — no action"]
  C -->|Yes| E{"Archive to external storage?"}
  E -->|Yes| F["Upload to object storage, delete local copy"]
  E -->|No| G["Delete file from disk"]
  F --> H["Disk space freed"]
  G --> H

Signals that disk is becoming a problem

• Recordings stop appearing for certain calls — when disk is full, VICIdial cannot open new files and silently skips recording.
• Database errors in the log — MariaDB needs disk space for tmp files during sorting and grouping; a full filesystem causes query failures.
• Rising SYSTEM CPU during conversion windows — MP3 compression of large WAV backlogs can spike the SYSTEM CPU percentage noticeably.
• Keepalive errors writing temporary state files — some Perl keepalive scripts write small flag files to disk; a full filesystem blocks these and causes process restart loops.

Checking disk alongside other server health metrics

Disk usage, system load, and Concurrent calls count are three separate but related dimensions of server health. A disk that is 85 percent full on a lightly loaded box is a scheduling problem. The same disk on a heavily loaded box that is also running at IDLE below 10 percent means you have two problems converging at once. Check all three in the same daily review. The Server Versions page gives you the disk number; the Server Performance Report gives you load and Channel count.

For the full monitoring routine, see the server health and capacity guide. For instructions on reading disk space directly from the Server Versions page, monitoring disk space via the Server Versions page covers that view in detail.

On a VICIfast managed box, your recordings stay on your own dedicated server. If your call volume is growing and you need a larger disk, a resize is a straightforward operation. See VICIfast plans and storage options to find a configuration that fits.