Skip to content
DiscVaultDiscVault Docs

Update and migrate: Restore

Goal and result

Section titled “Goal and result”

Restore the complete data set for the selected channel rather than copying only one database file.

Applies to

Section titled “Applies to”

For both release channels, restore the matching /data archive and PostgreSQL dump before starting the recorded image digest.

DiscVault v26 stable

Section titled “DiscVault v26 stable”
  • DiscVault v26 stable · ghcr.io/helmerznl/discvault:latest
  • PostgreSQL dump and filesystem archive from the same stopped-writer window
  • restore the matching data before starting the previous image

DiscVault v26 beta

Section titled “DiscVault v26 beta”
  • DiscVault v26 beta · ghcr.io/helmerznl/discvault:beta
  • PostgreSQL dump and filesystem archive from the same stopped-writer window
  • restore the matching data before starting the previous image

Prerequisites

Section titled “Prerequisites”
  • matching backup set
  • maintenance window
  • PREVIOUS_IMAGE · private DiscVault runtime environment file

Procedure

Section titled “Procedure”
  1. Stop: next-api + next-worker + next-mcp
  2. Restore: /data archive · before image startup
  3. Restore: PostgreSQL · pg_restore --exit-on-error
  4. Start: next-api → GET /api/next/health → next-worker + next-mcp

Commands and settings

Section titled “Commands and settings”

Both channels

Section titled “Both channels”
Terminal window
set -euo pipefail
BACKUP_ID=20260714T180000Z
cd /opt/discvault
. "/srv/backups/discvault-$BACKUP_ID.env"
test -n "$PREVIOUS_IMAGE"
export DISCVAULT_IMAGE="$PREVIOUS_IMAGE"
docker pull "$DISCVAULT_IMAGE"
ENV_FILE=/opt/discvault/.env
ENV_TMP="$(mktemp "$ENV_FILE.restore.XXXXXX")"
trap 'rm -f "$ENV_TMP"' EXIT
awk -v image="$DISCVAULT_IMAGE" '
  /^DISCVAULT_IMAGE=/ { print "DISCVAULT_IMAGE=" image; found=1; next }
  { print }
  END { if (!found) exit 42 }
' "$ENV_FILE" > "$ENV_TMP"
chmod 0600 "$ENV_TMP"
docker compose --env-file .env -p discvault -f compose.yaml stop next-worker next-mcp next-api
mv /srv/discvault "/srv/discvault.failed-$(date -u +%Y%m%dT%H%M%SZ)"
tar --acls --xattrs -C /srv -xzf \
  "/srv/backups/discvault-data-$BACKUP_ID.tgz"
docker compose --env-file .env -p discvault -f compose.yaml up -d postgres
docker compose --env-file .env -p discvault -f compose.yaml exec -T postgres \
  sh -c 'dropdb --if-exists --force -U "$POSTGRES_USER" "$POSTGRES_DB"'
docker compose --env-file .env -p discvault -f compose.yaml exec -T postgres \
  sh -c 'createdb -U "$POSTGRES_USER" "$POSTGRES_DB"'
docker compose --env-file .env -p discvault -f compose.yaml exec -T postgres \
  sh -c 'pg_restore --exit-on-error --no-owner -U "$POSTGRES_USER" -d "$POSTGRES_DB"' \
  < "/srv/backups/discvault-$BACKUP_ID.dump"
docker compose --env-file .env -p discvault -f compose.yaml up -d --force-recreate next-api
curl --fail http://localhost:6080/api/next/health
docker compose --env-file .env -p discvault -f compose.yaml \
  up -d --force-recreate next-worker next-mcp
docker compose --env-file .env -p discvault -f compose.yaml ps
mv "$ENV_TMP" "$ENV_FILE"
trap - EXIT

Expected result

Section titled “Expected result”

The original image starts healthy and library counts and media files match the backup.

  • library counts and sampled media match the selected backup
  • same procedure · :latest / :beta

Safety and rollback

Section titled “Safety and rollback”

Next step

Section titled “Next step”

Update and migrate: Container update

Source and status

Section titled “Source and status”