refactor(gluster): keep one cluster's values out of brick_servers #28
Loading…
Add table
Add a link
Reference in a new issue
No description provided.
Delete branch "refactor/brick-servers-tenant-vars"
Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?
Why
group_vars/brick_servers.ymlset rpcbind on and narrowed the brick port range to49152:49161for every tenant's brick servers, with comments describing one cluster's three single-brick nodes. The shared layer holds only what every tenant should get (AGENTS rule 10). A tenant with more than ten bricks on a node would see a new brick come up Online locally and stay unreachable to its peers, with nothing pointing at the firewall.What changed
gluster_enabled, and says that a cluster's own values (rpcbind, a narrower brick range) go in its tenant's inventory asbrick_serversgroup vars. ADVANCED_USAGE's group row says the same.91e2c0a).Verification
ansible-inventory --hostand--list, compared without printing values), so no host changes.make checkpassed.make testwas not run (no podman in the agent sandbox); no role or test changed.Risks and follow-ups
91e2c0ais on itsmain. That push was refused because the branch is protected. A checkout with the new castrum but the old inventory would drop the home brick servers' rpcbind rule and widen their brick range to the role default.