ArangoDB Server Agency Options
Activate
agency.activate
Activate Agency.
Compaction
agency.compaction-keep-size
Keep as many indices before compaction point.
Election
agency.election-timeout-max
Maximum timeout before an Agent calls for new election in seconds.
agency.election-timeout-min
Minimum timeout before an Agent calls for new election in seconds.
Endpoint
agency.endpoint
Agency endpoints.
My address
agency.my-address
Which address to advertise to the outside.
Pool size
agency.pool-size
Number of Agent pool.
Size
agency.size
Number of Agents.
Supervision
agency.supervision
Perform ArangoDB cluster supervision.
agency.supervision-frequency
ArangoDB cluster supervision frequency in seconds.
agency.supervision-grace-period
Supervision time, after which a server is considered to have failed, in seconds.
The default value is 10
seconds, which is recommended for regular cluster deployments. For active failover deployments it is recommended to use a higher value for the grace period to avoid unnecessary failovers. In active failover setups, the leader server needs to handle all the load and is thus expected to get overloaded and unresponsive more easily than a server in a regular cluster which needs to handle only a part of the overall load.