Merge pull request #108 from planetscale/document-fetching-gtids

Add information on how to find starting GTIDs to connector

README.md

@@ -169,4 +169,29 @@ You can start replication from a specific GTID _per keyspace shard_, by setting
 }
 ```
 
-**Note:** When `starting_gtids` is specified in the configuration file, _and_ a `--state` file is passed, the `--state` file will always take precedence. This is so incremental sync continues working.
+**Note:** When `starting_gtids` is specified in the configuration file, _and_ a `--state` file is passed, the `--state` file will always take precedence. This is so incremental sync continues working.
+
+**How to get starting GTIDs**
+
+You can get the latest exectued GTID for every shard by querying your database. 
+1. Access your PlanetScale database. One way to do so is to use `pscale shell`.
+2. Target the keyspace and shard that you would like the latest GTID for by doing `use keyspace/shard`.
+    - i.e. `use my_sharded_keyspace/-10`
+    - If your database is _unsharded_, you don't have to target a keyspace or shard. Skip this step.
+3. Execute `select @@gtid_executed;`
+4. You'll get a result that looks something like:
+```
+my_sharded_keyspace/-10> select @@gtid_executed\G
+*************************** 1. row ***************************
+@@`gtid_executed`: 16cec08d-f91b-11ee-8afb-aacaf4984ae5:1-5639808,
+b13c2fe0-f91a-11ee-aa81-6251c27c9c24:1-2487289,
+fe8e2a3c-f91a-11ee-9812-82f5834c1ba7:1-46602355
+1 row in set (0.01 sec)
+```
+5. Use the GTIDs returned to form the starting GTID for that shard (in this example, shard `-10`):
+```
+{"my_sharded_keyspace": {"-10": "MySQL56/16cec08d-f91b-11ee-8afb-aacaf4984ae5:1-5639808,b13c2fe0-f91a-11ee-aa81-6251c27c9c24:1-2487289,fe8e2a3c-f91a-11ee-9812-82f5834c1ba7:1-46602355"}}
+```
+6. Repeat this process for all your shards, if your database is sharded.
+
+**Note**: Remember to prepend the prefix `MySQL56/` onto your starting GTIDs.