Learn how to safely change primary keys in MySQL using pt-online-schema-change. This blog highlights tips to avoid data loss during online alter of primary key.

The post How to avoid data loss in MySQL Primary Key change first appeared on Change Is Inevitable.

Sure, wait event times in any database will include some CPU time, though ideally it should be minimal. Waits should be instrumented as close to the off CPU wait event as possible, i.e. if I’m doing an IO, then I want the wait to cover just the read call:

start_wait('my_read_wait');
pread( ... );
end_wait('my_read_wait');

So my function that sets up the read will do some CPU that could get caught under the cover of the wait itself and the call to pread will use some CPU but most of the time should be waiting for the actual I/O if the I/O is coming from storage.

start_wait('my_lock');
semtimedop(...);
end_wait('my_lock');

On MySQL I’ve noticed that average active sessions on CPU are often well below what the host reports as CPU utilization. For example if I have one connection on CPU a 100% of the time on a 2 vCPU host then CPU Utilization should be around 50%.  There might be …

Just  quick post to document a troublesome error when running Jmeter on Mysql

JDBC Request: Cannot load JDBC driver class’com.mysql.jdbc.Driver’ solution

Should be pretty simple, it’s just a question of putting the mysql-connector-java jar file, available at https://dev.mysql.com/downloads/connector/j/, into the classpath for Jmeter. The easiest way for me is to put this into the jmx file:

      <stringProp name=“TestPlan.user_define_classpath”> /Users/kyle.hailey/jars/mysql-connector-java-8.0.29.jar,/home/ec2-user/jars/mysql-connector-java-8.0.29.jar</stringProp>

Notice there is a space ” ” before the first jar file.  Apparently the space  was enough to cause the error. After removing the space , jmeter connected successfully to …

Anyone familiar with Oracle performance tuning knows about ASH , aka active session history. ASH saves a history of every session that is active every collection, which by default is once per second. Once per second is fast enough to give a clear view of database load.

The ASH data comes from v$session which is a structure that contains information on every connection and shows which connections are active, what SQL they are running (SQL_ID) and if they are runnable on CPU or waiting on a wait_event for some resource like an I/O to complete or a lock or latch. It’s a pretty simple select query on v$session to select who is active,  their wait_event and SQL_ID.

On Oracle that gives us a query like:

select sid||':'||serial# session_id,
       username,
       s.sql_id||':'||sql_child_number sqlid,
       decode(state, 'WAITING', wait_class||':'||event, 'CPU') event
from v$session s
where   (( s.wait_time != 0  /* on CPU  */ …

How does one use  NESTING_EVENT_ID?

Below I’ve selected from  events_waits_history_long for a single thread, but none of the  NESTING_EVENT_ID values show up in EVENT_ID or END_EVENT_ID.

To simplify things , I’m just showing the data for one THREAD_ID listed in order of TIMER_START.

mysql>  select 
            THREAD_ID,
            EVENT_ID,
            END_EVENT_ID,
            NESTING_EVENT_ID,
            NESTING_EVENT_TYPE,
            EVENT_NAME,
            SOURCE,
            TIMER_WAIT,
            SPINS,
            OBJECT_SCHEMA,
            OBJECT_NAME,
            INDEX_NAME,
            OBJECT_TYPE,
            OBJECT_TYPE,
            OBJECT_INSTANCE_BEGIN,
            OPERATION,
            NUMBER_OF_BYTES,
            FLAGS 
 from 
           performance_schema.events_waits_history_long 
 where 
            THREAD_ID in ( select max(thread_id) from performance_schema.events_waits_history_long where …

Question: how do I know that one user is waiting for a lock and another  user is burning CPU on MySQL?

i.e. how can I create active session history on MySQL like on Oracle?

Two simple cases I’m looking at

1. session waiting on lock
   1. The wait event is ” wait/io/table/sql/handler” which happens for a number of potential reasons
   2. -> I can join to information_schema views INNODB_LOCK_WAITS & INNODB_TRX but this is heavy, and I’d rather only access those after I know there is a lock waiter.
2. session burning CPU
   1. How do I know from performance schema views that the connection is active and burning CPU?
   2. It has a wait event that shows up wait/synch/mutex/sql/THD::LOCK_thd_data
   3. on Oracle to determine CPU, we …

We have planned for archiving the data to improve the DB performance and to reclaim the space. We were evaluating Compression in InnoDB and TokuDB. To find out the best compression method. We started benchmarking the compression ratio between InnoDB and TokuDB.

Everything goes well for some time, but after a few hours got an error message that can’t able to insert the data into the TokuDB table due to storage being full. It is so strange and the host has enough free space.

Table structure:-

mysql> show create table mydbops.tokudb\G
*************************** 1. row ***************************
       Table: tokudb
Create Table: CREATE TABLE `tokudb` (
  `ID` int DEFAULT NULL,
  `Name` longtext,
  `Image` blob
) ENGINE=TokuDB DEFAULT CHARSET=utf8mb4 COLLATE=utf8mb4_0900_ai_ci
1 row in set (2.18 sec)

mysql> show create table mydbops.innodb\G
*************************** 1. row …

To copy the data of the particular column of the table to another table/server, We have an option to export the data as CSV and import the data back to a different table. But when the table size is large and we need to copy the data only for the required data to the target table will cause the load in the server since the table scanning is huge.

To overcome this, we have the pt-archiver copy the data from the source table to the destination as a whole or only for required columns. And also we can do this in a controlled manner as well. So there will be no performance impact even on the production time.

Source table structure :

mysql> show create table source\G
*************************** 1. row ***************************
       Table: source
Create Table: CREATE TABLE `source` (
  `id` int unsigned NOT NULL …

If you haven’t already upgraded to MySQL 5.6, NOW IS THE TIME!!

Starting, August 3rd 2021, RDS will automatically upgrade MySQL 5.6 instances to version 5.7 within the earliest scheduled maintenance window that follows.
Starting, September 1st 2021, RDS will any remaining MySQL 5.6 instances to version 5.7 whether or not they are in a maintenance window.

AWS RDS

If you’re thinking to yourself, “It’s an automatic update, so why should I care?” the answer is no.

Whilst, the upgrade seems to be automatic but sometimes the aftermath is catastrophic. Listed below are the significant impact which few organization face after the automatic upgrade:

Data loss
Application getting choked
Data inconsistency flow
Performance degradation
Insert failure
…

After downloading and installing MySQL 8.0.24 yesterday, I opened a command shell. In the command shell, I could access the MySQL Shell (mysqlsh.exe) but not the MySQL Client (mysql.exe). Typing in the following:

C:\WINDOWS\system32>mysql

It returned:

'mysql' is not recognized as an internal or external command,
operable program or batch file.

The MySQL Client (mysql.exe) was installed because MySQL Workbench relies on it. However, the MySQL Microsoft Software Installer (MSI) does not put the mysql.exe file’s directory in the common Windows %PATH% environment variable. You can find the required %PATH% directory variable by opening the File Manager and searching for the mysql.exe file.

You should return several directories and programs but the directory you want is:

C:\Program …