【MOS】Index Rebuild Is Hanging Or Taking Too Long (文檔 ID 272762.1)

Oracle Database - Enterprise Edition - Version 8.1.7.4 to 9.2.0.4 [Release 8.1.7 to 9.2]

Information in this document applies to any platform.

 .

(AuthWiz 1.2) Created from <<TAR:3405564.995>>

<a></a>

Oracle Server - Enterprise Edition - Version: 9.2.0.4

This problem can occur on any platform.

Problem:

========

- Online Index rebuild takes a long time.

- ONLINE INDEX REBUILD SCANS THE BASE TABLE AND NOT THE INDEX

Symptoms:

=========

Performance issues while rebuilding very large indexes.

- The offline rebuilds of their index is relatively quick -finishes in 15 minutes.

- Issuing index rebuild ONLINE statement =&gt; finishes in about an hour.

- This behavior of ONLINE index rebuilds makes it a non-option for large tables 

as it just takes too long to scan the table to rebuild the index. The 

offline may not be feasible due to due to the 24/7 nature of the database.

- This may be a loss of functionality for such situations.

- If we attempt to simultaneously ONLINE rebuild the same indexes we may encounter

hanging behavior indefinitely (or more than 6 hours). 

DIAGNOSTIC ANALYSIS:

--------------------

We can trace the sessions rebuilding the indexes with 10046 level 12.

Comparing the IO reads for the index-rebuild and the index-rebuild-online 

reveals the following:

-ONLINE index rebuilds

It scans the base table and it doesn't scan the blocks of the index.

-OFFLINE index rebuilds

It scans the index for the build operation.

- This behaviour is across all versions.

TEST CASE:

----------

--connect as scott

--

sqlplus scott/tiger

--create some dummy table from the dba_objects view

create table objects as select * from dba_objects;

--create an index on the table

create index object_idx on objects(object_id,object_name);

--check out the file_id and block_id for the table and index

set linesize 150

set pagesize 4444

col segment_name format a40

select segment_name,file_id,block_id from dba_extents where owner = 'SCOTT' 

and segment_name like 'OBJECT%';

--trace the ONLINE index rebuild first

alter session set events '10046 trace name context forever, level 12';

alter index object_idx rebuild online;

--exit here and pull up the trace file. You'll see the reads

--for the cursor representing the cursor performing the rebuild that they

--are scanning the file and blocks belonging to the --base table

--called OBJECTS

exit

--Log back in and retry the same with the OFFLINE index rebuild

--get the new block_ids because the index has been rebuilt since we did this 

--last time

--trace and rebuild the index

alter index object_idx rebuild;

--exit out and check the trace file again. You'll see that we read the index 

--blocks for the rebuild

On analyzing the trace file generated, we will notice that there are lots of

'db file scattered read' wait events.

Cause/Explanation

=============

When you rebuild index online,

- it will do a full tablescan on the base table.

- At the same time it will maintain a journal table for DML data, which has 

changed during this index rebuilding operation. 

So it should take longer time, specially if you do lots of DML on the same table,

while rebuilding index online.

On the other hand, while rebuilding the index without online option, Oracle will grab

the index in X-mode and rebuild a new index segment by selecting the data from 

the old index. So here we are

- not allowing any DML on the table hence there is no journal table involved 

- and it is doing an index scan

Hence it will be pretty fast.

Solution/Conclusion:

===========

- The ONLINE index rebuild reads the base table, and this is by design.

- Rebuilding index ONLINE is pretty slow.

- Rebuilding index offline is very fast, but it prevents any DML on the base table.

<a href="http://blog.itpub.net/26736162/viewspace-1624453/"></a>