Re: Tuning for TCP-W benchmark - Kenji HIROHAMA - com.mysql.lists.cluster

5 messages in com.mysql.lists.clusterRe: Tuning for TCP-W benchmark

From	Sent On	Attachments
Kenji HIROHAMA	15 Nov 2005 01:01
Martin Skold	15 Nov 2005 01:20
Kenji HIROHAMA	15 Nov 2005 01:37
Adam Dixon	15 Nov 2005 14:18
Kenji HIROHAMA	15 Nov 2005 18:05

Subject:	Re: Tuning for TCP-W benchmark
From:	Kenji HIROHAMA (kenj...@gmail.com)
Date:	11/15/2005 06:05:23 PM
List:	com.mysql.lists.cluster

Hi Adam,

Thanks for your suggestion. I'm not sure this is a bug or not, but I just reported on mysql.com.

Thanks, Kenji

On 11/16/05, Adam Dixon <adam...@gmail.com> wrote:

Kenji, If you think its a bug related to the size of your database, perhaps lodge a bug and see if the mysql guys can reproduce your error with another large database.

Adam

On 11/15/05, Kenji HIROHAMA <kenj...@gmail.com> wrote:

Hi,

Yes, and I'm testing with 10sec of that value.

(BTW, under my 6.7GB-data-environment, system recovery doesn't work, so I need to use ndb_restore to reflect confit.ini file.

In my environment, during system recovery, the error always happens like; ndb_mgm> Node 2: Forced node shutdown completed. Occured during startphase 4. Initiated by signal 0. Caused by error 2341: 'Internal program error (failed ndbrequire)(Internal error, programming error or missing error message, please report a bug). Temporary error

I would like to report a bug, but BACKUP file is several GB.)

Do you have any other parameters I should try?

Thanks,

On 11/15/05, Martin Skold < Mart...@mysql.com> wrote:

Hi!

You probably need to increase TransactionDeadlockDetectionTimeout.

BR -- Martin

Kenji HIROHAMA wrote: Hi,

I'm running TPC-W-like-benchmark (DBT-1 by OSDL) to MySQL

Cluster.

However, the result seems not to be good comparing InnoDB.

Here is

table definitions;

http://www.hirohama.biz/wp/wp-content/dbt1_create_tables.sql

Here

is the output of slow query log;

http://www.hirohama.biz/wp/wp-content/c1-slow.log

The

benchmark connects 100 connections under my env., but active connection is under 50.

I tried to set --ndb-use-exact-count=0, but I got; Got temporary

error 233 'Out of operation records in transaction coordinator (increase MaxNoOfConcurrentOperations)' from ndbcluster

Then I increased

MaxNoOfConcurrentOperations to 10 times than default value; Lock wait timeout exceeded; try restarting transaction

Anybody would suggest some

parameter I should try for small number of connections, and long queries?

Thanks,

-- kenj...@nospam.gmail.com Kenji Hirohama

Martin Sköld, Software Engineer MySQL AB, www.mysql.com Office: +46

(0)730 31 26 21