|
Problem IC82472
|
Status: Closed |
WHEN DB MEMBER FAILURE OCCURS, READ ONLY TRANSACTIONS THROUGHPUT MAY DEGRADE FOR SEVERAL SECONDS. |
| product: |
DB2 FOR LUW / DB2FORLUW / 980 - DB2 |
| Problem description: |
When a DB member failure occurs, one of the surviving DB members
may not process transactions for several seconds.
It may occur when there is a large number of transactions.
A few seconds after the DB member failure, the following
messages appear in cfdiag.128.log.
2012-03-13-19.14.04.0292198000+540 E123456789A310 LEVEL :
Error
PID : 9633960 TID : 3599
HOSTNAME : host22
FUNCTION : CA trace, log_error
MESSAGE : CA server has encountered an error.
DATA #1 :
process_ignore_msg() Multi-HCA invocation failed on HCA ID: 0 of
2 HCAs.
status: 0x8006002e
2012-03-13-19.14.05.0159808000+540 E123456789A318 LEVEL :
Error
PID : 9633960 TID : 3599
HOSTNAME : host22
FUNCTION : CA trace, log_error
MESSAGE : CA server has encountered an error.
DATA #1 :
process_reject_msg_phase_1() Multi-HCA invocation failed on HCA
ID: 0 of 2
HCAs. status: 0x8006002e
2012-03-13-19.39.59.0468850000+540 E123456789A310 LEVEL :
Error
PID : 9633960 TID : 1800
HOSTNAME : host22
FUNCTION : CA trace, log_error
MESSAGE : CA server has encountered an error.
DATA #1 :
process_ignore_msg() Multi-HCA invocation failed on HCA ID: 0 of
2 HCAs.
status: 0x8006002e
2012-03-13-19.40.00.0398972000+540 E123456789A318 LEVEL :
Error
PID : 9633960 TID : 1800
HOSTNAME : host22
FUNCTION : CA trace, log_error
MESSAGE : CA server has encountered an error.
DATA #1 :
process_reject_msg_phase_1() Multi-HCA invocation failed on HCA
ID: 0 of 2
HCAs. status: 0x8006002e |
| Problem Summary: |
****************************************************************
* USERS AFFECTED: *
* Pure Scale *
****************************************************************
* PROBLEM DESCRIPTION: *
* See Error Description *
****************************************************************
* RECOMMENDATION: *
* Upgrade to v9.8.5. *
**************************************************************** |
| Local Fix: |
|
| Solution |
After the node failure, the EDU continues to wait for the
reception of the IGNORE notification for the failed member.
Which only happens around 8 seconds later, blocking all
transactions, that caused throughput degradation. |
| Workaround |
not known / see Local fix |
| Timestamps |
Date - problem reported :
Date - problem closed :
Date - last modified :
| 03.04.2012
13.06.2012
13.06.2012 |
| Problem solved at the following versions (IBM BugInfos) |
9.8.5
|
| Problem solved according to the fixlist(s) of the following version(s) |
| 9.8.0.5
|
 |
