Low disk space health checks

The GraphDB engine implements disk space validation health checks in order to avoid failure scenarios and unexpected behavior due to low disk space. These checks preserve GraphDB’s integrity even when a machine runs out of disk space.

Whenever a certain threshold of available disk space is reached, GraphDB logs a caution specifying the free space availability. There are two levels of warnings logged, followed with a fatal state which halts all transactions and requires manual addressing.

  1. Warnings are issued whenever the available disk space falls below 20% or less.

  2. Errors are issued whenever the available disk space falls below 10% or less.

  3. Fatal state is reached whenever the available disk space fall below 5% or 1GB, whichever comes first. At this point GraphDB will prevent new transactions from being started and log a fatal state.

The default values of the three admonition levels can be changed through their respective configuration properties.

Note

Only the percentage value of the critically low threshold can be changed. The 1GB threshold cannot be changed.

Recovering from a fatal state

Whenever the available disk space falls below the critically low threshold, GraphDB prevents new transactions from being started until either enough disk space has been freed, or enough additional disk space is allocated to the drive (if running on a dynamic drive).

Tip

To free additional disk space, you can delete server logs and large server files.