maria-developers team mailing list archive

[Kristian Nielsen <knielsen@xxxxxxxxxxxxxxx>]

Hm, seems I forgot to attach the patch.

 - Kristian.

=== modified file 'mysql-test/r/innodb.result'
--- mysql-test/r/innodb.result	2009-09-07 20:50:10 +0000
+++ mysql-test/r/innodb.result	2009-09-10 13:00:42 +0000
@@ -1738,7 +1738,7 @@ count(*)
 drop table t1;
 SELECT variable_value FROM information_schema.global_status WHERE LOWER(variable_name) = 'innodb_buffer_pool_pages_total';
 variable_value
-512
+511
 SELECT variable_value FROM information_schema.global_status WHERE LOWER(variable_name) = 'innodb_page_size';
 variable_value
 16384
@@ -1784,7 +1784,7 @@ innodb_sync_spin_loops	20
 SET @old_innodb_thread_concurrency= @@global.innodb_thread_concurrency;
 show variables like "innodb_thread_concurrency";
 Variable_name	Value
-innodb_thread_concurrency	8
+innodb_thread_concurrency	0
 set global innodb_thread_concurrency=1001;
 Warnings:
 Warning	1292	Truncated incorrect thread_concurrency value: '1001'

=== modified file 'mysql-test/t/bug46080-master.opt'
--- mysql-test/t/bug46080-master.opt	2009-07-10 12:00:34 +0000
+++ mysql-test/t/bug46080-master.opt	2009-09-10 13:36:22 +0000
@@ -1 +1 @@
---skip-grant-tables --skip-name-resolve --safemalloc-mem-limit=4000000
+--skip-grant-tables --skip-name-resolve --safemalloc-mem-limit=20000000

=== modified file 'mysql-test/t/innodb.test'
--- mysql-test/t/innodb.test	2009-09-07 20:50:10 +0000
+++ mysql-test/t/innodb.test	2009-09-10 12:44:52 +0000
@@ -1317,6 +1317,10 @@ drop table t1;
 
 # Test for testable InnoDB status variables. This test
 # uses previous ones(pages_created, rows_deleted, ...).
+
+# InnoDB aligns the memory for the buffer pool to a page boundary. This may
+# cause actual pool size to be one less than requested depending on exact
+# alignment of obtained memory.
 --replace_result 512 511
 SELECT variable_value FROM information_schema.global_status WHERE LOWER(variable_name) = 'innodb_buffer_pool_pages_total';
 SELECT variable_value FROM information_schema.global_status WHERE LOWER(variable_name) = 'innodb_page_size';

Kristian Nielsen <knielsen@xxxxxxxxxxxxxxx> writes:

> Hi Sergey,
>
> Attached patch fixes two more failures in the merge as seen in Buildbot.
>
> The main.innodb failures are two different issues. One I already fixed in
> another place in XtraDB, I added a comment to explain better. The other looks
> like a merge error, the result file was changed in the merge but the output of
> the test did not change.
>
> The main.bug46080 failure is really due to a poor test case. It checks that an
> out-of-memory condition does not crash the server, but it does so poorly using
> --safemalloc-mem-limit=4000000. This requires all kinds of assumptions about
> how much memory is actually used. MariaDB uses more memory (due to the Maria
> storage engine I assume). I increased the value to something reasonable for
> Maria that should still trigger the out-of-memory condition. The better fix
> would be to re-write the test case to use error injection.