|
Problem IC81916
|
Status: Closed |
A prefetch request at a high offset into a large tablespace may lead to a hang situation. |
| product: |
DB2 FOR LUW / DB2FORLUW / 970 - DB2 |
| Problem description: |
Under certain conditions tablespace I/O will be done by db2
prefetcher edus.
When very large tablespaces exist and an I/O request is
submitted to a db2 prefetcher
that is at an offset higher than 2^31 x <the DB2 I/O size>,
then the request submitting edu may hang and the I/O request
never completes.
This can cause the application or eventually even the database
to hang.
A typical stack trace would show :
sqloCSemP
sqlbpfParallelDirectIO
sqlbpfParallelDirectRead
...
Note that this stacks trace is typical for any edu waiting for
an I/O to be done by the prefetchers,
but the stack should not remain the same indefinitely.
This problem is more likely to occur on LOB I/O, when the I/O
size is only 1k.
So any I/O request above 2^31 x 1k = 2TB could cause a hang in
this case. |
| Problem Summary: |
****************************************************************
* USERS AFFECTED: *
* All *
****************************************************************
* PROBLEM DESCRIPTION: *
* Under certain conditions tablespace I/O will be done by db2 *
* prefetcher edus. *
* When very large tablespaces exist and an I/O request is *
* submitted to a db2 prefetcher *
* that is at an offset higher than 2^31 x <the DB2 I/O size>, *
* *
* then the request submitting edu may hang and the I/O request *
* never completes. *
* This can cause the application or eventually even the *
* database to hang. *
* *
* A typical stack trace would show : *
* *
* sqloCSemP *
* sqlbpfParallelDirectIO *
* sqlbpfParallelDirectRead *
* ... *
* *
* Note that this stacks trace is typical for any edu waiting *
* for an I/O to be done by the prefetchers, *
* but the stack should not remain the same indefinitely. *
* *
* This problem is more likely to occur on LOB I/O, when the *
* I/O size is only 1k. *
* So any I/O request above 2^31 x 1k = 2TB could cause a hang *
* in this case. *
****************************************************************
* RECOMMENDATION: *
* Upgrade to V9.7 FP7 or disable the parallel IO *
**************************************************************** |
| Local Fix: |
Disable the parallel I/O for the tablespace where your object is
located.
For example, if the object is in tablespace 3 use the following
registery:
db2set DB2_PARALLEL_IO=3:1 |
| available fix packs: |
DB2 Version 9.7 Fix Pack 6 for Linux, UNIX, and Windows
DB2 Version 9.7 Fix Pack 7 for Linux, UNIX, and Windows
DB2 Version 9.7 Fix Pack 8 for Linux, UNIX, and Windows
DB2 Version 9.7 Fix Pack 9 for Linux, UNIX, and Windows
DB2 Version 9.7 Fix Pack 9a for Linux, UNIX, and Windows
DB2 Version 9.7 Fix Pack 10 for Linux, UNIX, and Windows
|
| Solution |
First fixed in V9.7 FP7 |
| Workaround |
not known / see Local fix |
| BUG-Tracking |
forerunner : APAR is sysrouted TO one or more of the following: IC84529 IT03579
follow-up : |
| Timestamps |
Date - problem reported :
Date - problem closed :
Date - last modified :
| 08.03.2012
05.06.2012
05.06.2012 |
| Problem solved at the following versions (IBM BugInfos) |
9.7.FP7
|
| Problem solved according to the fixlist(s) of the following version(s) |
| 9.7.0.6
|
 |
