ArangoDB corruption issue

In some rare cases, ArangoDB might fail to restart after a compaction operation.

In this situation, docker ps shows that the graphdb (arangodb) container is restarting continuously and the webservice container becomes unhealthy because of that:

To see the underlying issue, please check the logs of the graphdb container and look for the fatal error: "Corruption: Compaction sees out-of-order keys."

Copy

This is a known issue in ArangoDB: https://github.com/arangodb/arangodb/issues/20841

Until ArangoDB releases an automated solution to recover from this corrupted state, the simplest solution is purging the Arango database files from the /data/graphdb/fsio folder (after stopping the sandbox service):

Purging the database files will impact the Advanced Search (Hunting) functionality: All previous scan reports will remain available in the system, but older scans will be excluded from advanced search operations.

Bash
Copy

After this purge, ArangoDB will start from a clean slate and the Sandbox webservice can function properly.

Type to search, ESC to discard
Type to search, ESC to discard
Type to search, ESC to discard