How to read Safenix backup failure logs on a server

Read the Safenix agent log with journalctl. Each log line is structured JSON. A failed cycle includes the activity type, error text, duration, and time until the next retry.

1. View the latest log entries

Run this command as root to display the last 50 lines:

sudo journalctl -u safenix-agent -n 50

To watch new entries as they are written, use:

sudo journalctl -u safenix-agent -f

To show only error entries, use:

sudo journalctl -u safenix-agent -p err

2. Find the failed cycle

Look for the exact message cycle failed; the next one will run as scheduled. Read the activity type and the accompanying error text. The same entry also shows how long the activity took and when it will be retried.

A single failed cycle does not stop the agent. The cycle is retried at the next scheduled interval, normally one hour later. Safenix sends an email alert only after three consecutive failures.

3. Match the error text to the likely cause

“no space left on device”

This indicates that the server has run out of space. Safenix backups need temporary disk space on the server in addition to the space occupied by the data.

Network error, such as “no such host” or a timeout

This means that the control plane could not be reached. This is counted as a failed cycle even when the data was already uploaded successfully and only the final communication failed. In that case, the backup data is safe despite the alert.

An unreadable file

An unreadable file appears as a warning, not an error, and does not fail the backup. Other files are protected normally, but the affected file is not included in the backup.

Search for these warnings with:

sudo journalctl -u safenix-agent | grep -i "item error"

A database connection or credential problem

If the MySQL user password has changed, the agent can no longer connect. Run the dedicated check:

sudo safenix-agent check-mysql

4. Decide whether the alert represents a failed backup

Use the activity type and error text in the JSON entry to distinguish a failed cycle from a warning about one unreadable file. For a network error, check whether the message indicates that only final communication failed after the data was uploaded; in that case, the backup is safe even though the cycle is counted as failed.