Dirkjan Bussink

> ### Binary Version > ```shell > v18 > ``` @olyazavr There's no v18 yet, which specific commit / sha is this? > executor.go:2082] cancelMigrations: cancelling f3387b12_5323_11ee_8c7e_0a4b134e1d27; reason: task error:...

https://github.com/vitessio/vitess/blob/794b91207ccb2a55275453c3ee696e4d4317f980/go/vt/vttablet/onlineddl/vrepl.go#L497-L498 > > I think that logic just needs to be dropped entirely, since the JSON marshaller already does the right thing and this wrapping breaks stuff. Hmm, this is...

https://github.com/planetscale/schemadiff/pull/16 is a good example of what we can achieve with the PRs linked back to this issue. All these improvements should also be in v19, but they are already...

> Related issue: [99designs/keyring#112](https://github.com/99designs/keyring/issues/112) Fwiw, I don't think this issue is just related, but it's exactly this problem that we're hitting here. @iambillmccann You probably end up debugging that package...

Sadly cleaning up that replace is a very big lift so not something that’s happening any time soon, if ever to be honest.

> * Maintain old behavior on "bad" cluster names (still spin up VTAdmin, vtctld and vtgate won't work) _but_ add a log/notification letting the user know they should use a...

Looks like https://github.com/vitessio/vitess/pull/14553 doesn't not properly fix this, as there can be a newline in the file and it then fails with: ``` E1124 14:11:50.284750 1 mysqld.go:450] Mysqld.Start(1700835110): error parsing...

Reopening this one as there's still cases where this can happen. One theory is that it could be that `mysqlctld` gets assigned the same PID as in the lock file....

> 3\. When the client issues **_SET @@vitess_metadata.read_my_write = , @@vitess_metadata.read_my_write_timeout = _**. VTGate will store this information in the session. Is it possible to have automatic fallback to the...

> In these use cases, we are still thinking that the user has to tell us if she wants to hit the primary or the replica, and we would not...