Details
-
Type:
Bug
-
Status: Closed
-
Priority:
Major
-
Resolution: Fixed
-
Affects Version/s: None
-
Component/s: Storage Engine - InnoDB, Storage Engine - XtraDB
-
Labels:None
Description
This test is (in 10.0.2 at least) unstable. Sometimes SELECT returns incorrect number of pages, namely from before the operation. But if the test is modified to repeat the same SELECT many (e.g. 10) times, then one can see that the number of pages eventually changes to the correct one. There's obviously a race condition here.
One way of fixing the test is to use include/wait_condition.inc and wait for the number of pages to reach the correct value. If the test times out, it would mean a bug.
Gliffy Diagrams
Attachments
Activity
- All
- Comments
- Work Log
- History
- Activity
- Transitions
Ok, I think I found it.
The test manipulates how purge is done, but the manipulating, debug-only code only works for main thread purge, not for background thread purge.
And while plain innodb has innodb_purge_threads=0 by default, xtradb has it default to 1.
Testing a fix now ...
Fix pushed to 5.5.
The fix is to explicitly set innodb_purge_threads=0 for this test case. Since xtradb defaults to 1, and the debug code used by the test does not work correctly with background purge.
Opened to change priority
The test is still disabled in 10.0. When I enable it — it still fails.
revno: 4471
committer: Jan Lindström <jplindst@mariadb.org>
branch nick: 10.0-bugs
timestamp: Mon 2014-11-03 15:47:57 +0200
message:
MDEV-4396: Fix innodb.innodb_bug14676111 test.
I do not understand the suggested fix. All the failures I see in the buildbot test failure database start with this:
+Timeout in wait_condition.inc for SELECT VARIABLE_VALUE < 1 FROM INFORMATION_SCHEMA.GLOBAL_STATUS
+WHERE VARIABLE_NAME = 'INNODB_PURGE_TRX_ID_AGE'
+Id User Host db Command Time State Info Progress
+1 root localhost test Query 0 NULL show full processlist 0.000
So there is already a timeout in a wait_condition, I do not understand how adding another condition can fix this?