October 1, 2006

Backport of micro-time patch to mysql 4.1

Posted by Vadim |

Taking into account 4.1 tree is still popular and is used on many production servers we
decided to make backport of patch to slow-log queries. The patch allows to specify time of slow queries in microseconds and is very helpful in a fight with problematic queries. 4.1 Patch is available here (The original patch was developed by Georg Richter georg at php net)

Related posts: :Microslow patch for 5.0.37::Microslow patch for 5.1.20::Binaries of MySQL 5.0.33 Community release for AMD64 / EM64T:
 

8 Comments »

  1. Thanks. I’m going to use it once I had time.

    Comment :: October 1, 2006 @ 1:39 pm

  2. Thanks.
    Very usfull patch for query profiling.

    Большое спасибо.

    Comment :: October 5, 2006 @ 11:42 am

  3. I patch mysql source, but it doesn’t work.

    I can not set long_query_time=0 . The least value of long_query_time – 1. If I set long_query_time=0, “SHOW VARIABLES;” show me long_query_time=1.

    And I see always seconds in slow log: Query_time: 2 Lock_time: 0 .

    How to turn on the patch?

    Comment :: October 9, 2006 @ 6:03 am

  4. 4. Vadim

    Michael,

    It looks like you are still using unpatched mysqld.

    The procedure is:
    get sources
    patch with: patch -p1 < patch.slow-micro.4.1.21.diff
    ./configure –prefix=/usr/local/mysql-micro …other-options..
    make
    make install

    In /usr/local/mysql-micro you should get binary

    Comment :: October 9, 2006 @ 7:09 am

  5. I do it again, how you write.
    After I run slow query and get this:

    # Time: 061010 0:58:54
    # User@Host: root[root] @ localhost []
    # Query_time: 7 Lock_time: 0 Rows_sent: 1 Rows_examined: 227073
    use mindmix;
    SELECT SQL_CALC_FOUND_ROWS * FROM `topics` WHERE `forum_id` !=1 LIMIT 1;

    Do you have another ideas?

    Comment :: October 9, 2006 @ 2:00 pm

  6. I use MyISAM tables…

    Comment :: October 9, 2006 @ 2:01 pm

  7. Sorry, I find problem.
    I used old mysqld.

    Comment :: October 9, 2006 @ 2:05 pm

  8. 8. peter

    Michael,

    Good you’ve found the problem. Regarding query you have in your example it is slow because of SQL_CALC_FOUND_ROWS – it effectively negates the limit.

    Comment :: October 9, 2006 @ 3:00 pm

 

Subscribe without commenting

Trackbacks/Pingbacks