Service mode released. The next chapter is here.

WOODPECKER · OPEN-SOURCE WAL

Cloud-Native
WAL.

Ordered, durable logs built on object storage.
Deploy a replicated WAL service or embed the engine in your Go application.

Service + EmbeddedObject-storage backedOpen source
INSIDE THE WAL01 / SERVICE
YOUR APPLICATION
↳Woodpecker clientAPPEND / READ
NODE 01LogStoreLocal WAL
NODE 02LogStoreLocal WAL
NODE 03LogStoreLocal WAL
COMPACT & UPLOAD
Object storageCompacted segments · shared durable storage
S3 / CLOUD
01

Replicate, then acknowledge. The client writes to the selected quorum. Acknowledgement follows the configured durability policy; compaction uploads completed data to object storage.

ILLUSTRATIVE TOPOLOGY · THREE REPLICAS SHOWN

01 / THE FOUNDATION

The log is fundamental.
Its infrastructure can be simpler.

Separate the WAL from your application. Use replication for the active write path and object storage for compacted segments.

Replicated by design

Configurable ensemble, write and acknowledgement quorums, with placement controls for availability zones and resource groups.

Understand the quorum

Storage that goes further

Stage active writes on LogStore nodes. Compact completed segments into object storage, separating compute from long-term retention.

Follow the storage lifecycle

Built to be operated

Inspect runtime health, trace slow operations and diagnose replica data with the wp CLI and documented recovery workflows.

Open the CLI handbook

02 / ONE ENGINE, TWO WAYS TO RUN

Your architecture.
Your deployment.

Choose a dedicated LogStore service or an in-process library. Both use etcd for coordination and share the Woodpecker log API.

DEDICATED CLUSTER

Service mode

Move WAL processing into a replicated LogStore cluster. Manage log storage independently from application compute.

APPLICATION→LOGSTORE × N→OBJECT
STORAGE
  • Quorum-based replication and node selection
  • Local staging with object-storage compaction
  • Cluster diagnostics and operational tooling
Deploy a service
IN-PROCESS LIBRARY

Embedded mode

Run the log engine inside your Go application. Use the object-storage backend without a separate LogStore cluster.

APPLICATION
+ WOODPECKER
→OBJECT STORAGE
  • Go library integration
  • No dedicated LogStore service to deploy
  • Etcd and storage backends configured by the application
Integrate the library

03 / FROM SYMPTOM TO ACTION

Understand it.
Diagnose it.
Keep it moving.

Operations are part of the product. Start with cluster health, inspect the affected log and replicas, then follow a recovery procedure grounded in the evidence.

Browse the complete CLI handbook
wp / operator workflowEXAMPLE

# 1. Check runtime health

$ wp cluster health

Cluster Health: YELLOW
active: 3 decommissioning: 0 unreachable: 0
- zero redundancy (active == N)

# 2. Inspect a log's local durable coverage

$ wp log scan orders --mode raw

SEGMENT  STATE      META LAST  REACHES
7        Completed  99         49

# 3. Read through the affected replicas

$ wp segment probe orders 7

Compare each replica's served range and outcome.

# Decide the next step from the findings.

Illustrative excerpt. Configure admin mappings and etcd first.
Three active nodes can be reachable with no spare replica capacity.

04 / RELEASED. AND MOVING FORWARD.

A new chapter.
A clear direction.

With service mode released, Woodpecker is in an ongoing release cycle. The next phase deepens the foundations of a mature cloud-native WAL service.

01

Reliability

Broader failure testing, predictable recovery and safer lifecycle transitions.

02

Performance

Higher throughput, lower tail latency and more efficient resource use.

03

Operations

Clearer diagnosis, controlled repairs and practical maintenance workflows.

04

Observability

More useful metrics, traces and explanations of runtime behavior.

CURRENT RELEASE v0.1.47 ↗See what shipped, and follow what comes next.Follow development

YOUR NEXT STEP

Build on Woodpecker.

Start a cluster, connect your application, or explore the operational toolkit.