Files
cozystack/examples/backups/redis/README.md
T
Andrey Kolkov b0eb104276 feat(backups): add Redis backup/restore strategy driver
The spotahome RedisFailover operator ships no backup CRD, so add a
dedicated strategy.backups.cozystack.io/Redis driver that runs a one-shot
Job: it dumps an RDB snapshot from the current master (discovered via the
operator's sentinel Service) to the system bucket over S3, and on restore
replays it (RDB -> RESP via redis-cli --pipe) into the target's master.

Structurally a specialisation of the Job/Altinity script-Job pattern with
a Redis applicationRef gate and an app-readiness precondition. Backup
parameters round-trip through the Backup's DriverMetadata (no
underlyingResources snapshot); restore drives from .Backup.ApplicationRef
so a to-copy restore reads the source's object-storage prefix while
writing into a differently-named target. Both in-place and to-copy are
supported; PITR is not (RDB is a discrete snapshot).

- api: Redis/RedisList strategy type + generated CRD and deepcopy
- driver: reconcileRedis / reconcileRedisRestore + pure helpers, wired
  into both dispatch switches, supportedBackupStrategyKinds and the no-op
  restore cleanup case
- default: gated cozy-default-redis strategy (system-bucket coords via
  cozy-backups-creds env secretKeyRef, https:// scheme + optional CA) and
  the Redis route in the cozy-default BackupClass
- image: images/redis-backup bundling redis-cli + curl + an RDB->RESP
  converter, wired via backupStrategyController.redisBackupClientImage
- tests: driver unit tests (gate, precondition, render, param round-trip,
  both reconcile paths) and the helm rendering-gate coverage
- example: examples/backups/redis honest marker round-trip for both
  restore flows

Signed-off-by: Andrey Kolkov <androndo@gmail.com>
Assisted-By: LLM
2026-09-08 17:56:20 +04:00

2.7 KiB

Redis backup / restore demo

Proves an honest data round-trip for a Cozystack-managed Redis (spotahome RedisFailover) through the platform system bucket: a sentinel marker key is written, backed up, and asserted to return byte-for-byte after both restore flows.

The spotahome operator ships no backup CRD, so the strategy.backups.cozystack.io/Redis driver runs a one-shot Job that dumps an RDB snapshot to S3 and replays it on restore (RDB → RESP via redis-cli --pipe). This is the cozy-default-redis strategy the cozy-default BackupClass binds Redis to.

Prerequisites

  • A Cozystack cluster with the backupstrategy-controller running and its system bucket provisioned, so cozy-default (and the cozy-default-redis strategy) are present and the controller can project cozy-backups-creds into the app namespace.
  • kubectl pointed at the cluster; the tenant namespace defaults to tenant-test (override with NAMESPACE).

Run

./run-all.sh            # steps 01-04: provision, back up, restore in-place, restore to-copy
SKIP_RESTORE=1 ./run-all.sh   # stop after a successful backup
./cleanup.sh            # remove everything the demo created

Or step through them:

Step Script What it does
01 01-create-redis.sh Provision Redis, seed the marker on the master
02 02-create-backupjob.sh BackupJob (backupClassName: cozy-default) → RDB dump to S3
03 03-restore-in-place.sh Corrupt the marker, RestoreJob back into the source, assert it returns
04 04-restore-to-copy.sh RestoreJob with targetApplicationRef into a fresh copy, assert the marker landed

Notes

  • Master discovery goes through the operator's sentinel Service rfs-redis-<name> (master name mymaster), so writes and the dump/replay always target the current master even after a failover.
  • Faithful restore. The restore loads the RDB into a throwaway local redis and replays what a real engine emits, so all logical databases (SELECT per db), stream last-ids and consumer groups survive — a merge onto db0 does not happen.
  • Endpoint TLS. The projected endpoint in cozy-backups-creds is always scheme-stripped, so the strategy prepends https:// and the transport is always TLS (there is no http:// path). It verifies against the image's system CA store by default — the platform bucket's ACME cert needs nothing more — honours a private CA from backupStorage.endpointCASecretName when set, and skips verification only when backupStorage.insecureSkipTLSVerify: true is set explicitly.
  • No PITR. RDB is a discrete snapshot; the engine ships no continuous log to replay to an arbitrary time, so spec.options.recoveryTime is not supported (the etcd-like case).