Newsletters
Abstract
Special Attention: DB2 11.5.8 and 11.5.9 upgrade to Latest CSB Level to Avoid Possible Memory and Disk Corruption
Content
- Db2 11.5.8 with cumulative special build level special_37055 or later.
- Db2 11.5.9 with cumulative special build level special_37506 or later.
Db2 11.5.8 with CSB level special_37055 or later and Db2 11.5.9 with CSB special_37506 or later may experience possible memory corruption when the amount of free space changes on a page in a table with a clustering index. This issue has been fixed by KI DT381034. To resolve this issue, upgrade to the latest published CSB level. It can be found on Db2 V11.5. 8 published cumulative special build downloads and Db2 V11.5.9 published cumulative special build downloads page.
These CSB levels have the fix:
1158 - special_42458
1159 - special_42449
When free space on a data page is modified, for example, through a delete, update or insert of a row and a clustering index is defined on the table, in some cases, memory corruption could occur. The memory corruption could potentially lead to subsequent corruption of table data. Various symptoms could be encountered including traps and errors when memory is being freed internally.
One of the symptoms is a trap with the following stack.
sqldUpdateFreeSpace
sqldUpdateRow
sqldRowUpdate
sqlriupd
sqlriSectInvoke
sqlrr_execute_immediate
sqlrr_execimmd
sqljs_ddm_excsqlimm
sqljsParseRdbAccessed
sqljsParse
sqljsDrdaAsDriver
sqeAgent
sqzEDUObj
sqloEDUEntry
As mentioned, the KI affect databases with clustering index. It also affects databases where clustering indexes existed but were subsequently dropped and the database has not been deactivated/activated after all clustering indexes were dropped. The following query can be used to determine if there are currently any clustering indexes in the database (note that it should only be used if it is certain that no clustering indexes were previously dropped or the database was deactivated/activated after the clustering indexes were dropped)
select substr(indschema,1,20) as INDSCHEMA, substr(indname,1,20) as INDNAME, substr(tabschema,1,20) as TABSCHEMA, substr(tabname,1,20) as TABNAME, indextype from syscat.indexes where indextype='CLUS'
Note: Installing a cumulative special build that includes Known Issue DT379179 will help prevent the possibility of memory or disk corruption from occurring. This KI fixes the root cause of the issue, on top of defensive code DT379179.
Recommendation
Customers who are currently on the impacted releases need to move to the latest CSB level as soon as possible to avoid potential on-disk corruption.
Temporary Workaround
N/A
Was this topic helpful?
Document Information
Modified date:
15 July 2024
UID
ibm17156281