Updates, backups and rollback

Version tags#

Bekci images are published for amd64 and arm64 on Docker Hub as kadirsungurlu/bekci (the same images are on GitHub Container Registry as ghcr.io/kadirsungurlu/bekci):

Two-part tags such as 1.0 follow the latest patch of that series (e.g. once 1.0.1 is out, 1.0 points to it).

Before you update#

1. Take a backup. The full backup below is the safest. With SQLite, Bekci also takes its own backup before updates that change the database structure (automatic backups).
2. Read the release notes: releases on GitHub.

Updating#

cd /opt/bekci
docker compose pull
docker compose up -d

docker pull kadirsungurlu/bekci:latest
docker stop bekci && docker rm bekci

Pinning the version#

The latest tag brings the newest version with every pull. If you want to decide when to update, use a specific version tag:

• Docker Compose: in docker-compose.yml, change image: kadirsungurlu/bekci:latest to image: kadirsungurlu/bekci:1.0.0.
• Caddy install: set UPTIME_TAG=1.0.0 in /opt/bekci/.env (for embedded PostgreSQL 1.0.0-postgres).
• Coolify: enter 1.0.0 in the Tag field.

To move to a new version, change the tag and run docker compose up -d (redeploy on Coolify).

Automatic backups#

Bekci takes two kinds of backups on its own. Both live inside the data volume, in /data/backups/.

• Change how many nightly backups are kept under Settings → General → Nightly backups to keep (0 turns nightly backups off).
• If the pre-update backup can't be taken (e.g. the disk is full), the update isn't applied and Bekci doesn't start; it starts once there's space again. These files are never deleted automatically; you can delete old ones by hand.
• With external PostgreSQL (DATABASE_URL) Bekci takes no backups; back up on the database side (below).

To list the backups:

docker exec bekci ls -la /data/backups

In a Compose install the container is named bekci-bekci-1; or, from inside /opt/bekci, use docker compose exec bekci ls -la /data/backups.

Copying backups off the server#

Copy the backup folder from the container to the server:

cd /opt/bekci
docker compose cp bekci:/data/backups ./bekci-backups

(For a docker run install: docker cp bekci:/data/backups ./bekci-backups.)

Then download it from your own computer:

scp -r root@SERVER-IP:/opt/bekci/bekci-backups ./

We recommend automating these steps with a cron job or your backup tool (restic, rsync, your provider's backup service…).

Full backup (the whole data volume)#

To capture the whole data volume, database included, in a single file, stop Bekci briefly (a database file copied while it's open can be inconsistent):

cd /opt/bekci
docker compose stop bekci
docker run --rm -v bekci_bekci-data:/data -v "$PWD":/yedek alpine \
  tar czf /yedek/bekci-data-$(date +%F).tgz -C /data .
docker compose start bekci

The volume is named bekci_bekci-data in a Compose install and bekci-data in a docker run install (docker volume ls shows it). The command creates a file such as /opt/bekci/bekci-data-2026-09-28.tgz.

Restoring#

cd /opt/bekci
docker compose exec bekci ls /data/backups          # pick the file name
docker compose stop bekci
docker run --rm -v bekci_bekci-data:/data alpine sh -c \
  'cd /data && cp backups/uptime-2026-09-28.db uptime.db && rm -f uptime.db-wal uptime.db-shm && chown 1000:1000 uptime.db'
docker compose start bekci

cd /opt/bekci
docker compose stop bekci
docker run --rm -v bekci_bekci-data:/data -v "$PWD":/yedek alpine sh -c \
  'find /data -mindepth 1 -delete && tar xzf /yedek/bekci-data-2026-09-28.tgz -C /data'
docker compose start bekci

docker exec bekci pg_restore -h /run/postgresql -U postgres --clean --if-exists --no-owner -d uptime /data/backups/uptime-2026-09-28.dump
docker restart bekci

cd /opt/bekci
docker compose exec -T db pg_dump -U bekci --format=custom bekci > bekci-$(date +%F).dump

cd /opt/bekci
docker compose stop bekci
docker compose exec -T db pg_restore -U bekci --clean --if-exists --no-owner -d bekci < bekci-2026-09-28.dump
docker compose start bekci

JSON backup of your configuration#

Settings → Backup / Restore → Download backup exports monitors, notification channels, tags, status pages and settings into a single JSON file. Users, check history and the audit log aren't included. You can import this file into another Bekci install (SQLite or PostgreSQL, it doesn't matter) with Restore in the same section. The file contains passwords and tokens in plain text; keep it somewhere safe.

Going back to an older version#

If a new version causes trouble, you can go back to the tag of the old one:

1. Find the tag to go back to: published versions (e.g. 1.0.0) are listed on Docker Hub. The old sürüm= (version) value from the log (e.g. 0ae7b6d) works as a tag too: kadirsungurlu/bekci:0ae7b6d.
2. If the new version upgraded the database (the log has a migration uygulandı line), the old version may not be able to open the new database. First restore the pre-update backup (pre-migrate-…db). Changes made after that backup are lost.
3. Switch the tag back to the old version and start it, e.g. image: kadirsungurlu/bekci:1.0.0 and docker compose up -d.

PostgreSQL major version upgrade#

This section only applies to the postgres image with embedded PostgreSQL. If Bekci ever moves to a new major PostgreSQL version (e.g. from 18 to 19), the data folder can't be opened directly. The container then won't start and writes a step-by-step explanation to its log under the heading HATA: PostgreSQL ana sürümü uyuşmuyor. (“ERROR: PostgreSQL major version mismatch.”). In short:

1. Stop the container and go back to the old image (the tag that contains the old PostgreSQL version).

2. While the old image is running, take a backup and copy the file off the server too:

   terminalCopy

   docker exec bekci pg_dump -h /run/postgresql -U postgres --format=custom --file=/data/backups/tasima.dump uptime

3. Stop the container and rename the /data/postgres folder instead of deleting it (e.g. /data/postgres-18-old).

4. Start the new image; an empty database is created.

5. Restore the backup and restart the container:

   terminalCopy

   docker exec bekci pg_restore -h /run/postgresql -U postgres --clean --if-exists --no-owner -d uptime /data/backups/tasima.dump
   docker restart bekci

6. Once everything works, delete the old folder.

Moving to another server#

1. Take a full backup on the old server and copy the file to the new server.
2. Install Bekci the same way on the new server, then stop it and restore the backup.
3. Point the domain's DNS record to the new server.

Because the address stays the same, server agents reconnect without being reinstalled: the IP lock looks at the agent's IP address, not the panel's.