RE: Oracle 12c extended VARCHAR - LOB segment grows indefinitely
From: Jonathan Lewis <jonathan_at_jlcomp.demon.co.uk>
Date: Mon, 25 Jan 2016 16:25:37 +0000
Message-ID: <CE70217733273F49A8A162EE074F64D9282BF4FC_at_EXMBX01.thus.corp>
Date: Mon, 25 Jan 2016 16:25:37 +0000
Message-ID: <CE70217733273F49A8A162EE074F64D9282BF4FC_at_EXMBX01.thus.corp>
Thomas, I said you would get the same problem with a CLOB because I ran your model on 12.1.0.2 then changed the varchar2(32000) to CLOB and got the same size problem. I've just run the test on 11.2.0.4 - and the same thing occurs - massive overallocation for ASSM, relatively small overallocation for freelist management. How have you defined the CLOB in 11.2, and what's the tablespace definition ? (and which version of 11.2?) Regards Jonathan Lewis http://jonathanlewis.wordpress.com _at_jloracle ________________________________________ From: oracle-l-bounce_at_freelists.org [oracle-l-bounce_at_freelists.org] on behalf of Thomas Kellerer [thomas.kellerer_at_mgm-tp.com] Sent: 25 January 2016 16:04 To: oracle-l_at_freelists.org Subject: Re: Oracle 12c extended VARCHAR - LOB segment grows indefinitely Jonathan, > A CLOB column will have the same problem. I think I came across this > problem quite a long time ago so it may be logged as a bug > somewhere. We have been running this on 11.2 for years now using a CLOB column and never encountered a problem. > If you can't find it as a bug on MoS raise it. I already searched for it and there either is no bug, or I am not able to find the right search terms ;) > If you're licensed for the partition option you could avoid the growth and the down-time > by adding in a step to recreate the table: > (I haven't tested this with a CLOB/extended varchar2 in place): > > Create the table as a hash partitioned table with one partition > Create a non-partitioned clone > Do your delete/insert > truncate the clone > insert append from the partition to the clone > exchange partition with table - online (12c feature). Thanks for the hint, I will test that. But I think if this indeed is a bug, then we will probably switch back to CLOBs rather than changing the process that fills the tables. Thomas -- http://www.freelists.org/webpage/oracle-l --http://www.freelists.org/webpage/oracle-l Received on Mon Jan 25 2016 - 17:25:37 CET