Related issue: https://github.com/github/gh-ost/issues/521
- Add --ssl-cert and --ssl-key options to specify SSL public/private
key files
- Allow combining --ssl-allow-insecure with other --ssl* flags.
`mysql.RegisterTLSConfig` allows combining the corresponding
parameters in the `tls.Config` it receives, so gh-ost should
allow this. I found being able to pass --ssl-allow-insecure along
with --ssl-ca, --ssl-cert, and --ssl-key useful in testing.
- Use the same TLS config everywhere. Since the CLI only supports
a single set of --ssl* configuration parameters, this should be
fine -- `mysql.RegisterTLSConfig` documentation indicates the
TLS config given will not be modified, so it can safely be used
in many goroutines provided we also do not modify it. The previous
implementation did not work when the TLS config was duplicated,
which happens when gh-ost walks up the replication chain trying
to find the master. This is because, when the config is duplicated,
we must call `RegisterTLSConfig` again with the new config. This
config is exactly the same, so it's easiest to side-step the issue
by registering the TLS config once and using it everywhere.
- Adding a command line option for users to enforce tls/ssl connections
for the applier, inspector, and binlog reader.
- The user can optionally request server certificate verification through
a command line option to specify the ca cert via a file path.
- Fixes an existing bug appending the timeout option to the singleton
applier connection.
- Identifying textual characters sets; converting into specific type when applying dml events
- Refactored `ColumnsList`: introducing `Column` type
- Refactored `unsigned` handling, as part of `Column`
- `Column` type supports `convertArg()`: converting value of argument according to column data type
- DB URI attempts `utf8mb4,utf8,latin1` charsets in that order (first one to be recognized wins)
- Local tests filter by pattern
- Local tests append table schema on failure
- Local tests do not have postpone flag file
- Added character set local tests: `utf8`, `utf8mb4`, `latin1`
- 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"
- 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
- when `--test-on-replica`, the tested replica is implicitly a control replica
- added `replication-lag-query`, an alternate query to `SHOW SLAVE STATUS` to get replication lag
- throttling takes both the above into consideration
