Public deploy kit and upgrade guide for self-managed, on-prem Foundry Connect
installations. Platform images are published to quay.io/ummsbiocore/* (public —
no pull credentials required).
Each Foundry Connect release attaches a versioned kit tarball to a GitHub release in this repo:
fc-deploy-<version>.tar.gz
Its contents:
| File | Purpose |
|---|---|
UPGRADING.md |
Step-by-step cutover + upgrade guide (start here). |
cutover.sh |
Dry-run helper that repoints a legacy image registry to quay.io/ummsbiocore/*. |
docker-compose.prod.yml |
Canonical production compose (fresh install / reference). |
.env.template |
Every environment variable the stack needs — copy to .env and fill in. |
install.sh |
Timestamped-backup deploy/rollback helper. |
README.md |
Kit-local quick reference. |
- Download the latest release's
fc-deploy-<version>.tar.gzfrom the Releases page. - Unpack it:
tar -xzf fc-deploy-<version>.tar.gz && cd fc-deploy. - Follow
UPGRADING.md— take a DB backup, dry-runcutover.sh, then deploy.
Rollback reverts the running image, not the database: for the on-prem
2.17-to-latest jump there are no schema migrations at all, so a one-line
image-tag flip is fully lossless. For upgrades that do cross additive
migrations, ALLOW_SCHEMA_AHEAD lets the older image run against the newer
schema without touching your data. The DB backup is insurance for the rare
destructive-migration case. See UPGRADING.md for the full model.
These artifacts are generated from deploy/prod/ in the (private) Foundry
Connect application repository and published here automatically on each release.
Do not edit files here by hand — they are overwritten on the next release.