Slow innodb replication

Mysqlians, Very Simple, Slave A is all MyISAM tables and replicates ~15writes/sec for ~200k/s with a top load of 0.00. I change 1 table(100mb data 1.5writes/sec) to innodb and the top load changes to .3 . I have set innodb_trx_commit=0 and set noatime on the filesystem ( ext3

Loading the .myd into memory

Mysqlians, Greetings. Besides the query cache is their a buffer which holds the data portion of MyISAM tables. All the buffers seem to hold key information or query processing information. Thanks, Trevor

Table lock statistics by TABLE not just server,db

Mysqlians, I am trying to determine in some of my database tables should be converted from MyISAM to Innodb. I have read that a collusion on 10-20% is the threshold at which one should convert( current # around .3%). However the table_locks_waited and table_locks_immediate

Mysql on Solaris 8/9 with ultrasparc

Mysqlians, Greetings .. I am at my wits end trying to find a performance problem with 4.0.16 on solaris 9 with 4 ultrasparc III process (sunfire v880). I would be much obliged if the Mame's and Sir's out their with this setup(or similar) would share with me their show variables or

Cardinality Bug ?

Hello, I was able to duplicate the following sequence in both 4.0.14-max and 4.0.16. It happens in all my tables. I do not understand why the cardinality for the date_created field does not = 223284. mysql select count(distinct date_created) from POSTING ;

2 Fulltext index's are better than 1? No

Hello, I recently ugraded to 4-14-Max from 4.13-Max to take advantage of the optimizer fulltext bug fix. However a certain query using the fulltext indexies is MUCH SLOWER then using no idex. Here is the query select distinct p.USER_CREATED, p.ID, p.DATE_CREATED,

Ugrade from 4.01.3-max to 4.0.16, now slower

Hello, I recently upgraded from 4.0.13-max to 4.0.16 and now the same queries/sec cause a higher load on the database server. I upgraded to 4.01.6 since the optimizer was not using some of my fulltext indicies. All our tables are MyIsam so I figure there is no reason to use the

Disk io wait during select,

Greetings.. Is their a way to list a time breakdown of a select? Amount of query waiting on disk-io, in memory, etc... I have tried the debug-info option with mysql client but it lists several memory settings. I have a 1.2 ultra-sparcIII that is twice as slow as a xeon 2.4

MySQL on Solaris 9 sparc with Perl

Greetings, I can not get the DBD::mysql module for perl to install correctly. It continues to complain about the mysql.so file. What worries me is the solaris dist does not include a libmysqlclient.so file which I do believe the linux DBD:mysql needs to install. Any help is most

Fulltext index not being used

Fellow Mysqlians, Can anyone tell me why mysql refuses to use the fulltext index in the second query? Query 1; mysql explain select straight_join distinct i.ID from INTEREST i use index (COMMENT) inner join PERSON p on p.ID=i.PERSON_ID where match(i.COMMENT)

MyODBC errors

Mysqlians, I am getting the following errors in an application log for 4.0.16 mysql and 3.51 myODBC SQL Error: SQLSTATE=S0002 Native error=1146 '[unixODBC][MySQL][ODBC 3.51 Driver][mysqld-4.0.16-Max-log]Table 'TRIBE.NETWORK_VECTOR' doesn't exist' SQL Error: SQLSTATE=S1000

Replication Error 1053

Mysqlians, The sql_thread on our slave slave server has been stopping with error 1053 ERROR: 1053 Server shutdown in progress 031203 16:05:01 Slave: error 'Server shutdown in progress' on query 'INSERT INTO INTEREST ( ID, PERSON_ID, COMMENT, DATE_CREATED, INTEREST_ID )

RAID Strip size

Greetings Mysqlians, Please comment on the validity of my logic: In setting the RAID(10/2disks) strip size everything I read says you must benchmark your particular system. Since that is not an option, my current logic is to have a large strip size (1024) with the reasoning

Ignor writes no block them.

Greetings, In converting a master into a slave, I would like to IGNORE all writes instead of blocking them. If I block writes with LOCK TABLES then once the server is converted into a slave and the tables are unlocked all of the blocked writes in the interrum period will be

Avg Queries per second...... per second

Greetings, From what I can gather the Queries per second average quoted by status is a pure division of Questions by Uptime in show status. Is there a way to flush these figures periodically? I want to be able to set the bin interval for this average, otherwise fluctuations get smoothed out.