Both Percona and Maria allow MySQL to be configured to listen on an extra port when their thread pool is enable.
* https://www.percona.com/doc/percona-server/5.7/performance/threadpool.html
* https://mariadb.com/kb/en/the-mariadb-library/thread-pool-in-mariadb-51-53/
This is valuable because if the table has a lot of traffic (read or write load), gh-ost can end up starving the thread pool as incomming connections are immediately blocked.
By using gh-ost on the extra port, MySQL locking will still behave the same, but MySQL will keep a dedicated thread for each gh-ost connection.
When doing this, it's important to inspect the extra-max-connections variable. Both Percona and Maria default to 1, so gh-ost may easily exceed with its threads.
An example local run using this
```
$ mysql -S /tmp/mysql_sandbox20393.sock -e "select @@global.port, @@global.extra_port"
+---------------+---------------------+
| @@global.port | @@global.extra_port |
+---------------+---------------------+
| 20393 | 30393 |
+---------------+---------------------+
./bin/gh-ost \
--initially-drop-ghost-table \
--initially-drop-old-table \
--assume-rbr \
--port="20395" \
--assume-master-host="127.0.0.1:30393" \
--max-load=Threads_running=25 \
--critical-load=Threads_running=1000 \
--chunk-size=1000 \
--max-lag-millis=1500 \
--user="gh-ost" \
--password="gh-ost" \
--database="test" \
--table="mytable" \
--verbose \
--alter="ADD mynewcol decimal(11,2) DEFAULT 0.0 NOT NULL" \
--exact-rowcount \
--concurrent-rowcount \
--default-retries=120 \
--panic-flag-file=/tmp/ghost.panic.flag \
--postpone-cut-over-flag-file=/tmp/ghost.postpone.flag \
--execute
```
- Rowcopy time is bounded by copy end-time
- Retries are configurable via `--default-retries` (default: `60`)
- `migrator` notes the hostname
- `applier` and `inspector` note `impliedKey` (`@@hostname` and `@@port`)
- Added lots of code comments
- Adding documentation for "triggerless design"
- Supporting `--allow-nullable-unique-key`
- Tool will bail out if chosen key has nullable columns and the above is not provided
- Fixed `OriginalBinlogRowImage` comaprison (lower/upper case issue)
- Introduced reasonable streamer reconnect sleep time
- Operation would terminate after events lock noticed but before applying all events: race condition where the event would be captured asynchronously. The event is now handled sequentially with the DML events, hence now safe.
- Multiple rowcopy operations would still write to `rowCopyComplete` channel. This is still the case, but now we only wait for the first and then just flush (read and discard) any others, to avoid blocking
- Events DML listener is only added after table creation: the problem was that with very busy tables, the events func buffer would fill up, and the "tables-created" event would be blocked.
- `waitForEventsUpToLock()` unifies the waiting on all variants of complete-migration
- With `--test-on-replica`, now stopping replication "nicely", using `master_pos_wait()`
- With `--test-on-replica`, not throttling on replication after replication is stopped (duh)
- More debug output
- Using heartbeat
- Throttling works based on heartbeat
- Refactored binlog_reader stuff. Now streaming events (into golang channel, which makes for nice buffering and throttling)
- Binlog table listeners work
- More Migrator logic; existing logic for waiting on `state` events (e.g. `TablesCreatedState`)
