ODA X5-2: When Using DB Flash Cache on RAC, Instance May Crash With ORA-600 [kjbrchkpkeywait:timeout]

Asset ID:	1-72-2075831.1
Update Date:	2016-05-27
Keywords:

Solution Type Problem Resolution Sure

Solution 2075831.1 : ODA X5-2: When Using DB Flash Cache on RAC, Instance May Crash With ORA-600 [kjbrchkpkeywait:timeout]

Applies to:

Oracle Database Appliance X5-2 - Version All Versions and later
Information in this document applies to any platform.

Symptoms

At 12c RAC when db_flash_cache_size setting is not zero, the database may hit ORA-00600 [KJBRCHKPKEYWAIT:TIMEOUT] causing an instance crash.

In the related incident trace file we can find:

BH (0xa1c66c28) file#: 40 rdba: 0x01404884 (5/18564) bscn: 0x215.9a740f63

flags: buffer_reused L2_buf_pinned L2_buf_pinned_remote_read flash_lock

insts 2: mark for cr weak bast: 1 2

On ODA X5-2 the scripts will enabledb_flash_cache by default when using oakcli create database.

Cause

This is Bug 19849781 EXADATA:LMS CRASHED WITH ORA-600 [KJBRCHKPKEYWAIT:TIMEOUT] related to flash lock has not been cleared up as expected cause the instance crash.

Although this is a general RAC bug, on ODA this will only happen with versions on X5-2.

Solution

Workaround: Stop the instances and disable the flash_cache by setting parameter db_flash_cache_size to 0.

This parameter can change dynamically, but sometimes the change the system will hang. In order to prevent this hang, it is recommended to make this change offline.

This fix will be provided in a future ODA patch release. A Single Patch for 22894949 may also be available.

There may be an interim patch available under Patch 19849781.

Setting this at internal as we have new patch 21961754 superceding this fix

Bug 21961754 - INSTANCE CRASHED AFTER LMS HITS ORA-00600 [KJBRSCRPING]...85 - Fix Superseded,

Note that this fix can cause / expose the problem described in Bug:22894949

Note that this fix has been superseded by the fix in Bug:22894949

Bug 22894949 - RAC PERF: CR TEST FOREGROUNDS NOW USE INDIRECT SENDS CAUSING 25% LESS THROUGHPUT

References

<BUG:22046677> - INSTANCE RESTARTED WITH ORA-600[KJBRCHKPKEYWAIT:TIMEOUT] DURING FLASH CACHE READ
<BUG:22129174> - BLR BACKPORT OF BUG 22046677 ON TOP OF DATABASE PSU 12.1.0.2.4
<BUG:19849781> - ORA-600 [KJBRCHKPKEYWAIT:TIMEOUT], WRONG PKEY/OBJD ON IOT
<BUG:22894949> - RAC PERF: CR TEST FOREGROUNDS NOW USE INDIRECT SENDS CAUSING 25% LESS THROUGHPUT

Attachments

This solution has no attachment