Yes, phoenix.schema.dropMetaData is a client-side setting. You can create a VIEW instead of a TABLE and your data won't be deleted. However, performance won't be as good as with a TABLE under some circumstances. Also, you wouldn't be able to use mutable secondary indexing with VIEWs. Note also, the old data is still there - we've just added Delete markers. You can actually connect at an "earlier" timestamp and continue to query the old data: http://phoenix.incubator.apache.org/faq.html#/Can_phoenix_work_on_tables_with_arbitrary_timestamp_as_flexible_as_HBase_API
Thanks, James On Thu, Feb 13, 2014 at 1:11 PM, Justin Workman <[email protected]>wrote: > I'm looking for some guidance on the proper way to drop a table from > Phoenix yet preserve the underlying HBase table and data? I have read here, > https://github.com/forcedotcom/phoenix/issues/219, that I should be able > to set a value to false and have the table preserved in HBase. Is this a > client side or server side setting? > > I added that value, phoenix.schema.dropMetaData, to my clients > hbase-site.xml and set it to false and when I dropped the table in Phoenix, > it did preserve the table in HBase, however it truncated all of the data. > > Is there anyway to preserve both the table and data when dropping a table > in Phoenix? > >
