Skip to main content
RunBook Academy

← All checklists in Ceph

Before deploymentceph-rgw-readiness

RGW Production Readiness Checklist

6 items ·2 critical ·4 warn ·0 info

Purpose

RGW is the front end with an external audience, which changes what the review has to cover. A single gateway is a single point of failure that a load balancer health check will not save you from; a bucket policy that is slightly too generous is reachable by anyone with the endpoint. This checklist is the gate before the object endpoint is published.

Order of work

The list runs in three passes. The availability pass is multiple-gateways and data-pool: more than one gateway behind the balancer, and enough room in the pool behind them. The tenancy pass is quota and bucket-policy, which decide how much one tenant can consume and what one tenant can see of another - the two questions that matter most once the endpoint is shared. The replication and confidentiality pass is multisite and sse, both of which are conditional: skip them explicitly if the deployment is single-site and unencrypted by agreement, rather than leaving them silently unchecked.

Before you start

Have the tenant list, the endpoint and load-balancer configuration, the zone and zonegroup layout if this is a multisite deployment, and whatever the compliance requirement says about encryption at rest.

Verification

Ready when every critical item is checked and a bucket has been created, written to, listed and deleted through the load balancer rather than against a single gateway directly. Mark an item N/A rather than leaving it unchecked when it does not apply to this cluster.

Escalation

A failing multisite sync blocks the endpoint going live. Take it to the storage lead before publishing.

Critical2 items

  1. ceph orch ps | grep rgw
  2. ceph df

Warning4 items

  1. radosgw-admin quota list
  2. radosgw-admin bucket list
  3. radosgw-admin sync status