eolivelli commented on issue #1436: BP-14 force() API - client side implementation URL: https://github.com/apache/bookkeeper/pull/1436#issuecomment-393094971 @sijie In the limited scope of this PR I think that it is good that we are not breaking ExplicitLAC and it is working as expected (I have added a test case as proof). I think that in general, despite BP14, the fact that LAC is only piggybacked on writes (and written finally on close/recovery...) is very annoying for tailing readers. The best would be to enable ExplicitLAC by default but as far as I can see it is not fully integrated in the normal flow of a tailing reader (you have to explicitly read ExplicitLAC from the reader side). In my opinion it is better for us to keep ExplicitLAC RPC as it is now. We have choosen to keep force RPC away from ledger storage (no masterKey on wireprotocol), but in the future we can add an optional payload with masterKey and current client-side LAC. We could make force() to really write a meta entry which contains the LAC and integrate this in the story and so ExplicitLAC won't be necessary anymore if we perform force() in background, even for non-DEFERRED_SYNC writers. So a recap for this PR and my view of the story after BP14: - the story we are writing is healthy and it does not break existing assumptions - it opens the door to future enhancements, and maybe to the deprecation of ExplicitLAC, in favour of standard LAC (we are introducing a mechanism to let LAC advance on writer side without "writes") I will be happy to continue personally this work and make BK more like a storage system cc @jvrao
---------------------------------------------------------------- This is an automated message from the Apache Git Service. To respond to the message, please log on GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
