When inplace reorg moves a row, after a target page is found and
latched, the page with the row to be moved is latched while the
target page is still latched. If a latch conflict is
encountered, the target page may need to be unlatched and the
pages latched in the appropriate order to prevent a deadlatch
with concurrent applications. Before unlatching the target page,
space is reserved on the page for the row to be moved to prevent
a concurrent application from using the space. If inplace reorg
already has sufficient reserved space on the page, no additional
space is reserved. In the case where inplace reorg did not
reserve any additional space, if, between the time the target
page is unlatched and relatched, a concurrent application
inserts a row onto the target page or updates a row causing it
to grow in size, inplace reorg will overwrite one or more rows
on the page.
Various symptoms are possible, including incorrect data being
returned or abends when reading data from the page. One example
of an error being returned during a query is
the following being dumped to the db2diag.log
FUNCTION: DB2 UDB, data management, sqldFetchOverflow,
probe:2145
MESSAGE : New scanner found to be following a REORGPOINTER!
DATA #1 : Hexdump, 8 bytes
0x00002AAD9F3FAF88 : 7200 0600 0000 0000 |
****************************************************************
* USERS AFFECTED: *
* can possibly occur on pureScale. *
* More likely to hit the issue in v105 which introduced *
* adaptive compression. *
****************************************************************
* PROBLEM DESCRIPTION: *
* Is this specific to pureScale, BLU, DPF or Serial? no *
* Is this on the common or error code path? error *
* Special conditions (platform, configuration option) : n/a *
* Did this require the intersection/combination of several *
* functionalities in order to occur? *
* Likelyhood details: more likely to hit the issue in v105 *
* which introduced adaptive compression. *
* Failing scenario description: *
* 2015-02-11-05.42.06.306042-300 E3827951392E1014 LEVEL: *
* Critical *
* PID : 2620 TID : 46925189474624 KTID : *
* 5815 *
* PROC : db2sysc 1 *
* INSTANCE: eunicec2 NODE : 001 DB : *
* CORAL *
* APPHDL : 1-19256 APPID: *
* 9.26.96.41.43670.150211104146 *
* AUTHID : EUNICEC2 HOSTNAME: coralm242 *
* EDUID : 210 EDUNAME: db2agent (CORAL) 1 *
* FUNCTION: DB2 UDB, data management, sqldFetchOverflow, *
* probe:10 *
* MESSAGE : ADM14001C An unexpected and critical error has *
* occurred: "BadPage". *
* The instance may have been shutdown as a result. *
* "Automatic" FODC *
* (First Occurrence Data Capture) has been invoked *
* and diagnostic *
* information has been recorded in directory *
* *
* "/home/eunicec2/sqllib/db2dump/DIAG0001/FODC_BadPage_2015-02 *
* -11-05.42 *
* .06.277894_0001/". Please look in this directory *
* for detailed *
* evidence about what happened and contact IBM *
* support if necessary to *
* diagnose the problem. *
****************************************************************
* RECOMMENDATION: *
* Upgrade to DB2 version 10.5 *
**************************************************************** |
