Restore data

InfluxDB OSS only

Use the influxd restore command to restore backup data and metadata from InfluxDB. You must stop InfluxDB before restoring data.

When restoring data from a backup file set, InfluxDB temporarily moves existing data and metadata while the restore process runs. Once the process completes, the temporary data is deleted. If the restore process fails, InfluxDB preserves the data in the temporary location. See Recover from a failed restore.

Restore data with the influxd CLI

1. Stop the influxd server.

2. Use the influxd restore command and specify the path to the backup directory using the --backup-path flag.

   # Syntax
   influxd restore --backup-path <path-to-backup-directory>
   
   # Example
   influxd restore --backup-path ~/backups/2020-01-20_12-00/

   For more information about restore options and flags, see the influxd restore documentation.

Customize the TSI rebuild process

By default, InfluxDB rebuilds the index and series file when restoring data. When rebuilding the Time Series Index (TSI), it uses the default build-tsi options. To customize the Time Series Index (TSI) rebuild process:

1. Disable rebuilding the index and series files when restoring data:

   influxd restore --rebuild-index false

2. Manually run influxd inspect build-tsi with any custom options.

   influxd inspect build-tsi \
     --max-log-file-size=1048576 \
     --max-cache-size=1073741824

   Manually rebuild the TSI index to adjust the performance of the TSI rebuild process.

Recover from a failed restore

If the restoration process fails, InfluxDB preserves existing data in a tmp directory in the target engine path (default is ~/.influxdbv2/engine).

To recover from a failed restore:

1. Copy the temporary files back into the engine directory.
2. Remove the .tmp extensions from each of the copied files.
3. Restart the influxd server.