Skip to content

Using ArangoDB Starter When Nodes' IP Addresses May Change #442

Open
@SeriousBeetle

Description

@SeriousBeetle

Hello there!
I have been experimenting with setting up an ArangoDB cluster on multiple nodes using the starter, and noticed that in the generated setup.json and arangod_command.txt, even though I specified (in --starter.join) their hostnames, it inevitably resolves to their IP addresses (e.g., the Address field of the peer in setup.json or the --agency.endpoint in the various arangod_command.txt in the coordinator/dbserver directories).
However, it looks like it keeps the leader's address intact in all these places, which is my desired behaviour.

I am currently working in an environment that cannot guarantee the stability of IP addresses, only their hostnames (DNS) (it's k8s, but I cannot use the operator (cluster admins won't allow it) and cannot switch to VMs).

Is there any way to use the starter without it resolving and passing the IP addresses, only the hostnames (as I originally specified in the arangodb --starter.join A,B,C command?)

I can attach a docker-compose.yml that highlights this behaviour if needed.

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions