redis.kappenhagen.io

RKB Monitor

Read-only live diagnostics for any Redis. Connect once, pull slowlog, latency, memory, cluster topology, and more — nothing is written to the database.

Download
v0.4.1 — current release
rkb-enterprise.exe
Windows · single executable · ~88.5 MB
rkb-toolkit-0.4.1.zip
Linux / macOS · Node.js 18+ · ~5 MB
Windows — run rkb-enterprise.exe, open localhost:7070.

Linux / macOS — unzip, then:
node server.js

No Redis agent. No egress. No secrets on disk. The connection is made from your machine to the database and nowhere else.
What it does
● Read-only
  • Pull live state from any Redis over a single read-only connection — AUTH, INFO, CONFIG GET, CLUSTER, and the diagnostics commands only. Nothing is written.
  • Reconstruct a full deployment plan from whatever is actually running, then diff it against best practice and your house standard.
  • Surface the slowlog with argument redaction, latency events, per-command profiles, memory doctor output, client aggregates, and cluster shard health.
  • Generate ready-to-use artifacts: redis.conf, Helm chart, Terraform, Kubernetes YAML, Python/Go/Node client pools, backup runbooks, upgrade runbooks, ADRs, and ops briefs.
  • Works offline once downloaded. Zero telemetry, zero egress after the font load.
Live diagnostics pulled
Slowlog (redacted)
Latency events
Command profile
Per-command latency
Memory doctor
Memory stats
Client aggregates
Error stats
Per-DB keyspace
Cluster shards
Slot coverage
Node topology
Search indexes
Pub/Sub channels
Modules & versions
ACL list
How to reach a private Redis
--via-kubectl <ns>/<svc>
RKB runs your own kubectl port-forward to a Kubernetes Service, dials the local end, and tears it down on exit.
The answer to a ClusterIP that isn't routable from your laptop. No iptables, no tunnel setup.
--ssh-jump user@host[:port]
Tunnel through an SSH jump host. Uses your agent or ~/.ssh keys by default; override with --ssh-key.
Requires: npm install ssh2 in the toolkit directory (optional dep).
--socks5 127.0.0.1:1080
Tunnel through any SOCKS5 proxy. Pairs with ssh -D 1080 user@bastion or a corporate proxy.
Zero extra dependencies — the SOCKS5 handshake is built in.
direct (default)
Plain TCP or TLS (rediss://) to any routable host — Redis Cloud, Upstash, or your VPN-exposed private Redis.
When a direct pull times out, RKB tells you which flag to try next.
Runbooks
Diagnose a slow instance
Download .md ↓
Confirm the symptom, separate latency-bound from throughput-bound, find blocking commands in the slowlog, read memory fragmentation and eviction counts, check persistence fork timing and replication lag. Includes module-specific surfaces for Search, JSON, and TimeSeries.
Recover from AOF corruption
Download .md ↓
Stop and back up the AOF before touching anything. Run redis-check-aof read-only to find the bad offset, then --fix to truncate at the last clean command. Covers multipart AOF (Redis 7+), partial rewrites left behind by a crash, and when to restore from RDB instead.
Safe reshard a live cluster
Download .md ↓
Pre-flight checks (cluster_state, slot coverage, replica count), snapshot topology before and after, capacity-check the target, then redis-cli --cluster reshard with explicit flags. Covers adding a master, draining a master, mid-move failure recovery, and --cluster fix for half-migrated slots.