swamirishi commented on PR #7402: URL: https://github.com/apache/ozone/pull/7402#issuecomment-2483305066
> > This is not an expensive operation. We anyways keep the entire containerMap in memory. There is no point in adding an overhead of rocksdb get, when we have the entire data in memory. > > The point is - either use RocksDB, or don't use it - don't use half in memory and half on RocksDB. We are managing two structures (a map and a table), when one can do it. We are just using Rocksdb as a ledger and nothing more than that. I didn't want to change the existing data structure. All of these have have been working before when things are completely in memory. I am justing adding an operation to also add it to rocksdb and recreate the entire data structure on startup by doing so. > You must read the entire small table from RocksDB on startup - so it will pull it into RockDB memory naturally. After that, writes will get buffered in RocksDB memory, so doing an existence check should be very fast. Removing the reliance on the missing containers map simplifies the code making it cleaner overall. > > > This can be done but we would go back to the same problem of changing all the test files in that case. > > Not if we keep the original add / remove for the common case and then rename the overloaded versions to better specify their intent. -- 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] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
