Subbu and I are working on 4213. It is distributed and supports master failover.
During his test, Subbu found 4213 to be significantly faster than 1730. Stay tuned. On Oct 28, 2011, at 2:33 AM, Mikael Sitruk <[email protected]> wrote: > Hi Doug > The architecture chapter is not really clear on this feature (unless i > missed something), nevertheless i can understand from your answer that the > time of the operation is related to the number of region. > > I found for the 0.92 release > https://issues.apache.org/jira/browse/HBASE-4213 (which is not resolved yet) > https://issues.apache.org/jira/browse/HBASE-1730 (resolved) > 1730 is the generic answer for the above (perhaps 4213 should be also marked > as resolved?) > Do you know when the 0.92 is planed to be released? > > Thanks > Mikael.S > > On Thu, Oct 27, 2011 at 3:34 PM, Doug Meil > <[email protected]>wrote: > >> Online table changes is a feature on TRUNK/.92 (dev-list, please comment). >> However, for .90 and before you need to disable. Long story short, this >> has to do with the fact that the CFs metadata needs to be propagated to all >> the regions. >> >> For background, see: http://hbase.apache.org/book.html#architecture >> >> >> ________________________________________ >> From: Mikael Sitruk [[email protected]] >> Sent: Thursday, October 27, 2011 3:19 AM >> To: [email protected] >> Subject: Adding a column family restriction >> >> Hi >> >> I have table in which I need to add column family. Nevertheless while >> trying >> todo this via API or from shell this operation cannot be complete without >> disabling the table first. From the business point of view i cannot affort >> disabling the table. >> >> So my questions are: >> Why such restriction exist? I can understand for deleting a column family >> by >> why for creating a new one? >> Is there a plan to enable this in the future? >> I have noticed that disabling/enabling table time is different from table >> to >> table, what is the factor affecting the disable/enable time (number of >> regions/volume of the table...)? >> >> I also suggest to update the Javadoc to reflect this restriction on the >> operation (currently no trace of this restriction in the javadoc) >> >> Thanks >> Regards, >> Mikael.S >> >> > > > -- > Mikael Sitruk
