Disabling of major compactions has very little impact. All that means is the default major compactions that run every 24 hours will not be run. That's it.
It has absolutely no impact on normal compact/split/flush heuristics. Does that help? Immediately following a split, compactions will run on the children to rewrite their data into separate files. Often times this could result in one file for each new region, but this doesn't mean there was a major compaction. Minor compactions _may_ result in all files -> one file. Major compactions _always_ result in all files -> one file. There is more involved with major compactions as well such as the removal of versions beyond the maxVersions setting, removing of versions because of TTL, etc. JG > -----Original Message----- > From: Vidhyashankar Venkataraman [mailto:[email protected]] > Sent: Thursday, June 17, 2010 10:38 AM > To: [email protected] > Subject: Questions on Region split.. > > I am repeating the mail that I sent yesterday: Hope the questions were > non-trivial enough. > > Data from a parent region is copied on to its child regions when > compactions are requested.. Is the compaction referred to here major or > minor? (I have turned major compactions off in my table, but I still > see some child regions each having one big store file which possibly > means they were major compacted from their parent regions: > > And I do see debug messages of the form: > 2010-06-16 20:57:52,840 DEBUG > org.apache.hadoop.hbase.regionserver.Store: loaded > /hbase/DocDB/90955749/bigColumn/6896592738209003301.1155422223, > isReference=true, sequence id=1276651030300, length=2147588214, > majorCompaction=true) > > I am trying to understand what happens to splits when major compactions > are disabled.. > > Would be glad if someone helps.. > Thank you > Vidhya > >
