Matt Lord
A property of messaging systems / job queues is that they don't/cannot guarantee order (in part because delivery does not mean handling/completion and they may be redelivered/taken by later workers)....
Another alternative — that I think is clearly better, but would require a new project on our side — is to use the [`performance_schema` replication tables](https://dev.mysql.com/doc/refman/8.0/en/performance-schema-replication-tables.html) instead. The master and...
Ping on this @deepthi — obviously not urgent, but I know it's easy for website things to slip through the cracks. 🙂
> @mattlord can you take a look at https://vitess.io/docs/16.0/overview/whatisvitess/? It seems to cover this topic pretty well. > I do see that we are missing schema management and materialization in...
Hi @J0SAL ! You can always open a PR w/o being assigned to the issue. I'll un-assign myself. Thanks!
> can we just update the installation guidelines with this link https://vitess.io/docs/contributing/build-on-ubuntu/#build-vitess as it already mentions building vitess from source using the `vitessio/vitess main` branch. That's definitely one option. It...
I'm going to make this about VDiff2 for now. It's a general issue for VReplication workflows, and we can start by addressing VDiff2 specifically, e.g. potentially with a new shard/workflow...
@arthurschreiber I'd like to try and better understand what you saw. For VDiff2 specifically, we have a mutex (`snapshotMu`) which should prevent this kind of concurrent modification of the shared...
> To clarify, the use of `Version` is not causing a functional issue in Vitess, right? This change is only to be consistent? > > Since we are watching a...
> OK, I just saw after my comment above that you have a new commit where everything is set as `ModRevision`. I guess this should be good. Yes, I noticed...