One: Version information
Operating system version: AIX 61009
Database version: 11.2.0.3.11 (RAC)
Two: Error description
Research and development reaction, many of their processes today reported "ORA-12801: Parallel Query Server p004,ora-08103: Object no longer exists" error. Check out their process specific error part found
All are insert into test select/*+ Parallel (a,4) */* from test1 A similar statement of the error
Three: Causes of errors and solutions
For this before the same treatment has been normal, one day suddenly the problem of failure, the analysis of ideas is still very clear. (usually due to changes to the database)
First, we looked at Oracle's alert logs and didn't find any errors. Then asked a few colleagues, the results found that in the time of the error he was on some partition table compression processing. Further checks found
The table that was involved in the error is the one he did the compression processing. Analysis here has been able to boldly guess the cause of the failure is because of his compression operation.
Take our guess to the MOS to inquire about the ORA-08103 error, find the following two note
1) "Getting ORA-1410 or ORA-8103 when Running Select against Partition or subpartition with Active Exchange (Doc ID 727597. 1) "
2) "oerr:ora-8103" Object no longer exists "Master note/troubleshooting, Diagnostic and solution (Doc ID 8103.1)".
From the two articles we know that the related objects in the database (Entity objects, such as tables and indexes, etc.) do truncate Table,alter index. Rebuild,alter table. Move and so on will change the data_object_id of the object.
If we are accessing related objects in a session, it is possible to trigger a ORA-08103 error if the object is in several operations. The partition of the compressed table is actually a move operation (ALTER TABLE AAA move partition PM201407 compress;)
can cause data_block_id to change.
Quad: Solutions
Try not to make changes to the programmed table during the program's operation.
ora-12801,ora-08103