ArangoDB v3.12 is under development and not released yet.
This documentation is not final and potentially incomplete.
How to produce an Agency Dump
One can read out all information of an Agency in the following way:
curl -k https://<any-coordinator>:<port>/_api/cluster/agency-dump > agency.json
When authentication is enabled, the user provides either an authentication
header to access every server or uses the root user credentials. The
authentication header is generated using the following
AUTH=$(arangodb auth header --auth.jwt-secret <path-to-cluster-jwt-secret>)
The generated authentication header is then used in the following way with
curl, to produce the Agency dump:
curl -kH"$AUTH" https://<any-coordinator>:<port>/_api/cluster/agency-dump > agency.json
Or using username and password like below, where one is prompted to type in the password. It is best practices to not specify the root password on command line considering the risks associated with finding passwords in the shell history. The jwt secret method or username password method if done like above leaves no such traces.
curl -k --username root https://<any-coordinator>:<port>/_api/cluster/agency-dump
Should the Agency be down, an Agency dump can still be created starting from the database directory of (one of) the Agents. Contact ArangoDB Support, in this case, to obtain more detailed guidelines on how to produce the dump.