Just a small correction re: > Major compactions combine all files into a single file. Minor compactions > select a subset of files can combines them into a file.
Minor compactions are when in-memory data is written to files. Major compactions are when files are combined. Major compactions can involve a subset of the files assigned to a tablet, or all of them (usually referred to as a full major compaction). Flush forces minor compaction(s) to happen immediately vs when the in-memory tables fill up. See: https://accumulo.apache.org/1.10/accumulo_user_manual.html#_tablet_service <https://accumulo.apache.org/1.10/accumulo_user_manual.html#_tablet_service> https://accumulo.apache.org/1.10/accumulo_user_manual.html#_compactions <https://accumulo.apache.org/1.10/accumulo_user_manual.html#_compactions> > On Apr 14, 2022, at 9:36 AM, dev1 <d...@etcoleman.com> wrote: > > Flush and compactions are different actions. > > Flush – sorts and writes current, in-memory changes to a file. This can > reduce the amount of recovery in case of a failure because the flushed > entries do not need to be processed from the WAL. > > Compactions combine multiple files into a single file. Major compactions > combine all files into a single file. Minor compactions select a subset of > files can combines them into a file. > > See: https://accumulo.apache.org/1.10/accumulo_user_manual.html#_compaction > <https://accumulo.apache.org/1.10/accumulo_user_manual.html#_compaction> > > Flushing will increase the number of files generated, this will potentially > increase the number of compactions. There are tradeoffs. If you are asking > will frequent flushes reduce the time required to perform a major compaction? > Probably not much, if at all. > > Ed Coleman > > From: Ligade, Shailesh [USA] <ligade_shail...@bah.com > <mailto:ligade_shail...@bah.com>> > Sent: Thursday, April 14, 2022 9:14 AM > To: user@accumulo.apache.org <mailto:user@accumulo.apache.org> > Subject: minor compaction same as flush > > Hello just wanted to some clarification, > > Is the flush same as minor compaction? Is flush better (performance wise) > than running say range compaction? > Having flush often, will it help major compaction performance or no > difference?? > > Thanks > > -S
