|
Problem IC61333
|
Status: Closed |
SERVER MAY HANG IN "WAITING FOR NEXT LOGICAL LOG FILE TO BE FREED" DURING FAST RECOVERY WHEN GLOBAL TRANSACTION IS PRESENT |
| product: |
IBM IDS ENTRP E / 5724L2304 / B15 - IDS 11.50 |
| Problem description: |
The server may hang in "Waiting for Next Logical Log File to be
Freed" during fast recovery under the following circumstances:
- There is an old heuristically completed transaction.
- The log where the transaction begun is already overwritten.
- Another transaction is present which is rolled back during
fast recovery.
- The current log is almost full so that the rollback has to
aquire a new one.
in online.log:
03:43:36 10 recovery worker threads will be started.
03:43:39 Logical Recovery has reached the transaction cleanup
phase.
03:43:39 Checkpoint Completed: duration was 0 seconds.
03:43:39 Fri May 1 - loguniq 239, logpos 0x3be018, timestamp:
0x54a2e8 Interval: 392
03:43:39 Maximum server connections 0
03:43:39 Checkpoint Statistics - Avg. Txn Block Time 0.000, #
Txns blocked 0, Plog used 0, Llog used 1
03:43:39 Waiting for Next Logical Log File to be Freed
> onstat -x
Transactions
address flags userthread locks
beginlg curlog logposit isol retrys coord
7000000103ff028 A---- 7000000103bf028 0 0
239 0x3bf018 COMMIT 0
7000000103ff2f0 A---- 7000000103bf820 0 0
0 0x0 COMMIT 0
7000000103ff5b8 A---- 7000000103c0018 0 0
0 0x0 COMMIT 0
7000000103ff880 A---- 7000000103c0810 0 0
0 0x0 COMMIT 0
7000000103ffb48 A---- 7000000103c1008 0 0
0 0x0 COMMIT 0
7000000103ffe10 A-R-- 7000000103c1008 0 239
239 0x21d64c COMMIT 0
7000000104000d8 -LH-G 0 0
227 234 0x123c COMMIT 0
The global transaction started in logical log 227. This log was
already resused.
> onstat -l
address number flags uniqid begin
size used %used
700000010415550 1 U-B---- 235 1:263
1000 1 0.10
7000000104155b8 2 U-B---- 236 1:1263
1000 2 0.20
700000010415620 3 U-B---- 237 1:2263
1000 1 0.10
700000010415688 4 U-B---- 238 1:3263
1000 880 88.00
7000000104156f0 5 U---C-L 239 1:4263
1000 1000 100.00
700000010415758 6 U-B---- 230 1:5263
1000 1 0.10
7000000104157c0 7 U-B---- 231 1:6263
2000 1 0.05
700000010415828 8 U-B---- 232 1:8263
2000 1 0.05
700000010415890 9 U-B---- 233 1:10263
2000 2 0.10
7000000104158f8 10 U-B---- 234 1:12263
2000 3 0.15
10 active, 10 total |
| Problem Summary: |
****************************************************************
* USERS AFFECTED: *
* IDS customers who have applications that generate global *
* transactions, either through I-Star or the XA protocol *
****************************************************************
* PROBLEM DESCRIPTION: *
* When switching to a new log file during fast recovery IDS *
* checks to see if the next log file can be freed. There are *
* many conditions that might prevent freeing a log file. *
* When a global transaction is heuristically aborted by IDS, *
* it is completed, but the transaction structure remains in *
* memory so that the transaction coordinator can be notified *
* of its heuristic completion. This sort of transaction was *
* appearing, to the routine responsible for freeing a log *
* file, like an open transaction and in certain cases was *
* preventing the log file from being freed. *
* *
* The symptom are a hang during fast recovery with a message *
* in the log indicating that the server was waiting for the *
* next log file to be freed. *
****************************************************************
* RECOMMENDATION: *
* Upgrade to IDS 11.50.xC5 and above. *
**************************************************************** |
| Local Fix: |
|
| Solution |
The problem is first fixed in IDS 11.50.xC5. |
| Workaround |
not known / see Local fix |
| Timestamps |
Date - problem reported :
Date - problem closed :
Date - last modified :
| 02.06.2009
19.01.2010
19.01.2010 |
| Problem solved at the following versions (IBM BugInfos) |
11.50.xC5
|
| Problem solved according to the fixlist(s) of the following version(s) |
| 11.50.xC4.W1
|
 |
| 11.50.xC5
|
|
