Files
kube/dragonfly-operator-system/README.md
rasmus efb467e425 dragonfly: scale to 1 instance (hotfix)
eaas: dragonfly has 1 rw instance and applications
don't realize and change over from ro to rw when
the leader changes.
2025-07-17 10:52:21 +03:00

27 lines
1.2 KiB
Markdown

# Dragonfly Operator
Dragonfly operator is the preferred way to add Redis support to your application
as it is modern Go rewrite and it supports high availability.
Following alternatives were considered, but are discouraged:
* Vanilla Redis without replication is unusable during pod reschedule or Kubernetes worker outage
* Vanilla Redis' replication is clunky and there is no reliable operator for Kubernetes
to use vanilla redis
* KeyDB Cluster was unable to guarantee strong consistency
Note that vanilla Redis
[has changed it's licensing policy](https://redis.io/blog/redis-adopts-dual-source-available-licensing/)
# For users
Refer to [official documentation on usage](https://www.dragonflydb.io/docs/getting-started/kubernetes-operator#create-a-dragonfly-instance-with-replicas)
For example deployment see
[here](https://git.k-space.ee/k-space/kube/src/branch/master/passmower/dragonfly.yaml).
To find other instances in this repository use `grep -r "kind: Dragonfly"`
Use storage class `redis` for persistent instances.
To achieve high availabilllity use 2+ replicas with correctly configured
`topologySpreadConstraints`.
# For administrators
See [/argocd/applications/dragonfly.yaml](/argocd/applications/dragonfly.yaml)