Parallel Query Fails with ORA-600 [15764] [ID 839536.1]

 

In this Document
  Symptoms
  Cause
  Solution
  References

---

This document is being delivered to you via Oracle Support's Rapid Visibility (RaV) process and therefore has not been subject to an independent technical review.

Applies to:

Oracle Server - Enterprise Edition - Version: 10.2.0.3 to 11.2.0.1 - Release: 10.2 to 11.2
Information in this document applies to any platform.

Symptoms

A select statement running in parallel failed with ORA-600 [15764].

The tracefile showed the following call stack:

----- Call Stack Trace -----
kxfprigdb <- kxfpqrgdb <- kxfxgs <- kxfxcw <- qerpxFetch <- opifch2 <- kpoal8 <- opiodr <- ttcpip <- opitsk <- opiino <- opiodr <- opidrv <- sou2o <- opimai_real <- main <- start

The tracefile also showed the process state was busy holding a child latch:

=================================================== 
PROCESS STATE 
------------- 
Process global information: 
     process: 7000004748eaa00, call: 700000443d33870, xact: 0, curses: 700000474e203c8, usrses: 700000474e203c8 
  ---------------------------------------- 
  SO: 7000004748eaa00, type: 2, owner: 0, flag: INIT/-/-/0x00 
  (process) Oracle pid=157, calls cur/top: 700000443d33870/700000443d33870, flag: (0) - 
            int error: 0, call error: 0, sess error: 0, txn error 0 
  (post info) last post received: 0 0 249 
              last post received-location: kxfprienq: QC 
              last process to post me: 700000474914f40 186 0 
              last post sent: 0 0 250 
              last post sent-location: kxfprienq: slave 
              last process posted by me: 700000474914f40 186 0 
    (latch info) wait_event=0 bits=10 
      holding    (efd=7) 700000472e4a838 Child process queue reference level=4 child#=99   
      Location from where latch is held: kxfprigdb: KSLBEGIN: addr qref  <---
        Context saved from call: 504403177372952360 
        state=busy, wlstate=free  <----
    Process Group: DEFAULT, pseudo proc: 700000474a384f0 
    O/S info: user: oracle, term: UNKNOWN, ospid: 2351144 
    OSD pid info: Unix process pid: 2351144, image: oraclePPSOLTP1@psoldbap003


Cause

The error is caused by unpublished Bug 9825798: "DBMV2:SQL EXEC PART:ORA-00600: INTERNAL ERROR CODE, ARGUMENTS: [15764]," which was closed as unpublished Bug 9792010: "DBMV2: STARETL_EXADATA: INSTANCE TERMINATES AFTER MULTIPLE PQ ERRORS." 

This error has been seen as early as 10.2.0.3 and is fixed by the 11.2.0.2 patch set.  The error is not RAC specific.

Solution

The following solutions are available:

1. Upgrade to 12.1 (Future Release).
2. Apply the 11.2.0.2 Server Patch Set.
3. Apply the 11.2.0.1 Bundle Patch 8 (Exadata Database only).
4. Apply one-off Patch 9792010 if available for your RDBMS version and OS platform.
5. Workaround:
   Increase parallel message size to avoid join message fragmentation.
   The join message will be sent in one message buffer making the join atomic
   from the QC point of view.

Note:  Although the patch name contains the word "Exadata," this is a generic RDBMS parallel execution bug and is not exclusive to Exadata databases.

References

NOTE:9792010.8 - Bug 9792010 - Dumps / ORA-600 in PQ slaves due to race condition

相关内容

---

产品

---

• Oracle Database Products > Oracle Database > Oracle Database > Oracle Server - Enterprise Edition

关键字

---

PARALLEL EXECUTION; PARALLEL QUERY; FAILED ACTIVITY错误

---

ORA-600[15764]; ORA-600[[15764]; ERROR 0