echo "Checking for duplicate email templates..."
psql "$CLOUDRON_POSTGRESQL_URL" -v ON_ERROR_STOP=1 -c "
SELECT name, template_type, locale, account_id, inbox_id, COUNT(*)
FROM email_templates
GROUP BY name, template_type, locale, account_id, inbox_id
HAVING COUNT(*) > 1;
"

This was pasted in one go.

1- Should the app be updated to the broken 1.54 or stick with the working 1.53.1?
2- Should I paste the entire code above in one-shot? If not, then my other attempt by pasting one line at a time didn't work either? I'll try pasting in sections as they're pasted above.

Fixed.

To answer my own questions.
1- Yes, apply the code to broken package 1.54.
2- Apply code one section at a time.

App package 1.54.0 now starts as it should. Thank you @andreasdueren

1. create backup of chatwoot in case something goes bad
2. put app into recovery mode
3. enter web terminal and enter

set -euo pipefail

source /app/data/env.sh

export FRONTEND_URL="${CLOUDRON_APP_ORIGIN}"
export REDIS_URL="redis://${CLOUDRON_REDIS_HOST}:${CLOUDRON_REDIS_PORT}"
export REDIS_PASSWORD="${CLOUDRON_POSTGRESQL_PASSWORD}"
export POSTGRES_DATABASE="${CLOUDRON_POSTGRESQL_DATABASE}"
export POSTGRES_USERNAME="${CLOUDRON_POSTGRESQL_USERNAME}"
export POSTGRES_PASSWORD="${CLOUDRON_POSTGRESQL_PASSWORD}"
export POSTGRES_HOST="${CLOUDRON_POSTGRESQL_HOST}"
export POSTGRES_PORT="${CLOUDRON_POSTGRESQL_PORT}"
export RAILS_ENV=production

echo "Checking for duplicate email templates..."
psql "$CLOUDRON_POSTGRESQL_URL" -v ON_ERROR_STOP=1 -c "
SELECT name, template_type, locale, account_id, inbox_id, COUNT(*)
FROM email_templates
GROUP BY name, template_type, locale, account_id, inbox_id
HAVING COUNT(*) > 1;
"

echo "Creating missing legacy index if needed..."
psql "$CLOUDRON_POSTGRESQL_URL" -v ON_ERROR_STOP=1 -c "
CREATE UNIQUE INDEX IF NOT EXISTS index_email_templates_on_name_and_account_id
ON email_templates (name, account_id);
"

echo "Running Chatwoot database migrations..."
cd /app/code
bundle exec rails db:migrate

echo "Done. Now disable Recovery Mode / Debug Mode and restart the app."

4. when prompted, take app out of recovery mode

My installation ran into this upgrade issue as well, thanks for the commands. Instead of running the last database migrations myself I just disabled the recovery mode and restarted the app.

Needed a minute to figure the CLI part out as I hadn't used that before (it's on your machine not on the server or the app container for everybody else)

Also make sure to fill in the password env var names in step 4.

Hit the exact same issue yesterday — same migration, same error. Root cause: the 1.54.0 migration AddInboxScopeToEmailTemplates tries to drop the index index_email_templates_on_name_and_account_id, but on instances installed at certain older Chatwoot versions that index never existed. Postgres throws PG::UndefinedObject, and since db:migrate runs on every boot, the app is stuck in a restart loop. (Upstream bug — the migration should use if_exists: true.)

Fix without rolling back (no data is touched — you just create the index so the migration can drop it):

1. Stop the crash loop:

cloudron debug --app <chatwoot-app-id>

2. Safety check — make sure there are no duplicate email templates (the migration later adds unique indices that would fail on duplicates):

cloudron exec --app <app-id> -- bash -c \
  'psql "$CLOUDRON_POSTGRESQL_URL" -c "SELECT name, template_type, locale, account_id, COUNT(*) FROM email_templates GROUP BY 1,2,3,4 HAVING COUNT(*)>1;"'

Expected: (0 rows). If you get rows, dedupe first.

3. Create the missing index:

cloudron exec --app <app-id> -- bash -c \
  'psql "$CLOUDRON_POSTGRESQL_URL" -c "CREATE UNIQUE INDEX index_email_templates_on_name_and_account_id ON email_templates (name, account_id);"'

4. Run the migration manually (in debug mode the start script hasn't populated the Rails env, so map the Cloudron addon vars):

cloudron exec --app <app-id> -- bash -c '
  export POSTGRES_DATABASE="$CLOUDRON_POSTGRESQL_DATABASE" \
         POSTGRES_USERNAME="$CLOUDRON_POSTGRESQL_USERNAME" \
         POSTGRES_PASSWORD="${CLOUDRON_POSTGRESQL_PASSWORD}" \
         POSTGRES_HOST="$CLOUDRON_POSTGRESQL_HOST" \
         POSTGRES_PORT="$CLOUDRON_POSTGRESQL_PORT" \
         REDIS_URL="redis://$CLOUDRON_REDIS_HOST:$CLOUDRON_REDIS_PORT" \
         REDIS_PASSWORD="${CLOUDRON_POSTGRESQL_PASSWORD}" \
         RAILS_ENV=production
  source /app/data/env.sh
  cd /app/code && bin/rails db:migrate'

Watch for AddInboxScopeToEmailTemplates: migrated — the remaining pending migrations run right after.

5. Leave debug mode:

cloudron debug --app <app-id> --disable

App should be back within ~30s. Rolling back to 1.53.1 only postpones this — it'll hit again on the next update until the package/upstream guards the remove_index with if_exists: true.