Mustafa Dagher

From the [comments on the PR](https://github.com/camunda/zeebe/pull/17877#issuecomment-2082915980) for #17637 , @oleschoenburg, @Zelldon and @deepthidevaki brought to my attention that > the generated `requestId` might be unique across partitions, although this was...

For trasnparency: However the implementation in https://github.com/camunda/camunda/pull/17877 was finalized and the PR was approved to have a unique requestId from exported with the records, after discussions with @korthout, we agreed...

> @Zelldon reminded me that we should verify that we do not make wrong assumptions about `requestId` being set. For example, maybe we assume somewhere that records with a `requestId`...

I'm trying to look for possible ways to test/verify the changes here. But I couldn't come up with good ones yet. I'd appreciate any hints.

> It might be unique across partitions, although this was not the intention. Please check [here](https://github.com/camunda/zeebe/blob/main/zeebe/transport/src/main/java/io/camunda/zeebe/transport/impl/AtomixServerTransport.java#L45). But it is easy to break this assumption if we run this code per...

> Very nice job @mustafadagher, thanks! 🚀 > > I added a few comments to improve the implementation, please take a look. 👀 > > ❌ Let's add a test...

@korthout @deepthidevaki @oleschoenburg @Zelldon Could I ask you guys to review my last commit. The commit refactors the requestIdGenerator to be part of the broker startup steps, introducing a guarantee...

> @mustafadagher With #18669 merged, is there anything from this PR that we want to keep? @korthout, I wanted to open that up with @npepinpe, but didn't get the chance...

@rodrigo-lourenco-lopes No, not that I know of! Excuse my ignorance, I was just going through the main epic for context to another topic when I saw these issue archived in...

ZPA triage: - @berkaycanbc to look into it, and involve ZDP if needed.