Installation

TwinCommit ships as a single static binary with no runtime dependencies beyond the client library of your source database. Install it with the script, a container image, the Helm chart or Homebrew.

Linux

curl -fsSL https://get.twincommit.io | sh

The script detects your architecture, verifies the release checksum and installs twincommit to /usr/local/bin. To pin a version:

curl -fsSL https://get.twincommit.io | sh -s -- --version 1.4.0

Docker

docker run --rm \
  -v $PWD/pipeline.yaml:/etc/twincommit/pipeline.yaml:ro \
  -e DB2_PASSWORD -e PG_PASSWORD \
  -p 9108:9108 \
  ghcr.io/twincommit/twincommit:1.4.0 \
  run -config /etc/twincommit/pipeline.yaml

The image includes the IBM Data Server Client and Oracle Instant Client, so it can reach remote sources directly.

Kubernetes (Helm)

helm repo add twincommit https://charts.twincommit.io
helm repo update
helm install prod-db2 twincommit/twincommit \
  --namespace replication --create-namespace \
  --set-file pipeline=./pipeline.yaml \
  --set existingSecret=prod-db2-credentials

The chart runs two replicas with leader election by default. See high availability.

macOS (Homebrew)

brew install twincommit/tap/twincommit

macOS builds are intended for development against remote sources and targets.

Verify the install

twincommit version

twincommit 1.4.0 (linux/amd64, built 2026-11-03)

System requirements

Resource	Recommendation
OS	Linux x86-64 or arm64 (glibc 2.28+)
CPU	2 vCPU per pipeline
Memory	1 GiB plus batch.max_txn_memory_bytes
Disk	Enough under spill_dir for your largest transaction
Network	Access to the source, the target and (optionally) Prometheus

Note

For Db2 sources, TwinCommit loads libdb2 from the IBM Data Server Client or a local Db2 installation. The Docker image bundles it; for the binary install, set DB2_HOME or make sure libdb2.so is on the library path.

Next steps

Continue with the quickstart or read the configuration reference.