Hi Arul, What is your table descriptor? Use hbase shell and scan the table with timerange (1 week older). Validate if any data returns? Validate hdfs space by hbase partition? /hbase/data, /hbase/archive, /hbase/oldWals, /hbase/MasterProcWals, etc.. Which is most consumed? Validate hdfs table folder to find hfiles that are older than 7 days(major compaction threshold)? Validate If any HBase snapshot created regularly? Validate regionserver log if compaction error? Validate hbase row counter vs Ingested data count? Also, Validate if you have any HDFS snapshot on /hbase partition?
-Karthik On Wed, Jul 14, 2021 at 5:20 PM Arul Ramachandran <arkup...@gmail.com> wrote: > We have a small use case for HBase but the data consumed by HBase is > growing in size comparable to the rest of the HDFS usage. > > I noticed the hbase table TTL setting 5 days does not seem to be honored - > the data stays much longer or is never purged. The major compaction is set > for 7 days. > > Any suggestions on what to look out for to make sure TTL works as expected? > > > Thanks, > Arul >
