lordcheng10 edited a comment on pull request #2973: URL: https://github.com/apache/bookkeeper/pull/2973#issuecomment-1008221599
> But I am still not sure , how it addresses the root cause. I understand that throttling with semaphore sort of reduces the pressure on the bookie. And there is a another mechanism for that too, that is percentageOfLedgerFragmentToBeVerified(Slight misnomer here, it actually checks percentage of entries in the ledger fragments). I understand throttling will reduce timeout from bookie. But timeout can still happen and will happen. My question is why not address this issue that occasional timeout should not be considered a failure, or may be should be retried? Thoughts? Looks good otherwise. percentageOfLedgerFragmentToBeVerified is not very useful. This parameter is for a Fragment, but in fact, most Fragments have only one entry, but at least the first and last entry will be checked. percentageOfLedgerFragmentToBeVerified configures the default configuration we use, and the default configuration is 0. Every time when check all ledger is executed, a large number of ledgers will be marked as under replica, obviously this is a wrong judgment。@pkumar-singh -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
