docs: add postgres upgrade description

Author: ziggie1984

docs/postgres.md

@@ -67,3 +67,146 @@ List of key-value tables/relations:
 ```
 
 Native SQL tables include invoices and graph data, representing the ongoing migration toward a fully native SQL implementation.
+
+## Upgrading Postgres Version
+
+This guide describes the recommended approach for upgrading your Postgres database to a newer version. This process requires planned downtime for your LND node.
+
+### Prerequisites
+
+- Access to both old and new Postgres instances
+- Sufficient disk space for database dumps
+- Administrative access to stop/start LND
+
+### Migration Steps
+
+#### 1. Prepare the New Postgres Instance
+
+Install and configure the new Postgres version on your target system. Ensure the new instance is running and accessible.
+
+Create an empty database for LND on the new Postgres instance:
+
+```shell
+$ psql -h new-postgres-host -U postgres
+postgres=# CREATE DATABASE lnddb;
+postgres=# CREATE USER lnduser WITH PASSWORD 'your-password';
+postgres=# GRANT ALL PRIVILEGES ON DATABASE lnddb TO lnduser;
+```
+
+#### 2. Stop LND
+
+Before beginning the migration, stop your LND node to ensure data consistency:
+
+```shell
+$ lncli stop
+```
+
+Verify that LND has fully stopped before proceeding.
+
+#### 3. Backup the Current Database
+
+Create a full backup of your current database using `pg_dump`:
+
+```shell
+$ pg_dump -h old-postgres-host -U lnduser -d lnddb -F c -f lnddb_backup.dump
+```
+
+The `-F c` flag creates a custom format dump which is recommended for `pg_restore`. This backup serves as your safety net during the migration.
+
+**Important**: Store this backup in a safe location. Without a backup, you cannot recover if something goes wrong during the migration.
+
+#### 4. Restore to New Postgres Instance
+
+Restore the database to your new Postgres instance:
+
+```shell
+$ pg_restore -h new-postgres-host -U lnduser -d lnddb -v --on-conflict-do-nothing --exit-on-error lnddb_backup.dump
+```
+
+The `-v` flag provides verbose output so you can monitor the restoration progress.
+
+**Critical**: Always use the `--exit-on-error` flag during restoration. Without this flag, `pg_restore` may continue even if errors occur, which can lead to incomplete schema migration and data inconsistencies. The restore process must fail fast if any errors are encountered to ensure data integrity.
+
+#### 5. Verify the Migration
+
+After restoration completes, perform basic verification checks:
+
+```shell
+$ psql -h new-postgres-host -U lnduser -d lnddb
+```
+
+Check that all expected tables exist:
+
+```sql
+-- Verify key-value tables
+SELECT tablename FROM pg_tables WHERE schemaname = 'public' AND tablename LIKE '%_kv';
+
+-- Check row counts for a sample key-value table (e.g., channeldb_kv)
+SELECT COUNT(*) FROM channeldb_kv;
+
+-- Check row counts for native SQL tables (if applicable two example tables are
+-- depicted in the following)
+SELECT COUNT(*) FROM invoices;
+SELECT COUNT(*) FROM graph_nodes;
+```
+
+Compare the row counts with the old database to ensure data was migrated successfully.
+
+#### 6. Update LND Configuration
+
+Update your LND configuration to point to the new Postgres instance:
+
+```
+[db]
+db.backend=postgres
+db.postgres.dsn=postgresql://lnduser:your-password@new-postgres-host:5432/lnddb
+db.postgres.timeout=0
+```
+
+#### 7. Start LND and Verify
+
+Start your LND node:
+
+```shell
+$ lnd
+```
+
+Monitor the logs during startup to ensure LND connects successfully to the new database. Verify basic functionality:
+
+```shell
+$ lncli getinfo
+$ lncli listchannels
+$ lncli listinvoices
+```
+
+If LND starts successfully and you can query your channels and invoices, the migration is complete.
+
+### Troubleshooting
+
+If LND fails to start or you encounter issues:
+
+1. Check LND logs for database connection errors
+2. Verify the connection string in your LND configuration
+3. Ensure the new Postgres instance is accessible from your LND host
+4. Verify that all tables were restored correctly in the new database
+
+### Post-Migration
+
+Once you've verified everything is working correctly:
+
+1. Keep the backup file (`lnddb_backup.dump`) for at least a few days
+2. Monitor your LND node closely for the first 24-48 hours
+3. The old Postgres instance can be decommissioned once you're confident in the new setup
+
+**Note**: There is no automated rollback procedure. If you need to revert, you must restore from your backup to the old Postgres instance and reconfigure LND.
+
+### Replication Considerations
+
+If you are running Postgres with replication (as discussed in the "Important note about replication" section), additional steps may be needed:
+
+1. Ensure replication is properly configured on the new Postgres instance before starting LND
+2. Verify that synchronous replication is working correctly after the migration
+3. Test failover scenarios to ensure the replica has been correctly synchronized
+4. Monitor replication lag to ensure replicas stay in sync with the primary
+
+**Note**: Always consult and follow the official PostgreSQL best practices for version upgrades and database migrations.