Ben Pfaff
Ben Pfaff
> This will need help from the profiling infrastructure to collect data for storage too. The right way to disable the test is to use `@Ignore`. But this will work...
Probably the solution is https://github.com/feldera/dist-design/issues/13
I think this is fixed.
> The simplest solution from the connector development perspective is to give the connector the ability to delay the snapshot by returning some type of a TRY_AGAIN status. Something like...
> > The simplest solution from the connector development perspective is to give the connector the ability to delay the snapshot by returning some type of a TRY_AGAIN status. >...
> > > The simplest solution from the connector development perspective is to give the connector the ability to delay the snapshot by returning some type of a TRY_AGAIN status....
Yes, you'll need to include part of the `Cargo.toml` from the feldera repo, e.g.: ``` [workspace] members = [ "crates/dbsp", "crates/nexmark", "crates/pipeline-types", ] resolver = "2" [workspace.metadata.release] release = false...
@aehmttw Since we decided against going this direction, do you think we should close this PR without merging?
@aehmttw Thanks! I appreciate your work on this. It really convinced me that we do not want to go this route.
The new version of the multithreaded, fault-tolerant Kafka connector never reorders within a partition, so this bug is fixed.