Our apache cluster was not able to serve all the requests so we need to raise MaxClients paramenter on each node. Unfortunatly the new value was exceeding the ServerLimit value so we had to raise it accordingly. Despite this on restart apache was again complaining that MaxClients was exceeding ServerLimit value:

# /etc/init.d/apache2 restart
* Forcing reload of web server (apache2)...
WARNING: MaxClients of 1000 exceeds ServerLimit value of 256 servers,lowering MaxClients to 256. To increase, please see the ServerLimitdirective.

These are the interesting lines in /etc/apache2/apache2.conf.

Can you see something wrong here? It was hard for me to figure out how to fix the problem!

  StartServers            10
  MinSpareServers         10
  MaxSpareServers         20
  MaxClients              1000
  ServerLimit             1000
  MaxRequestsPerChild     0
  

Here the right one.
The point is: ServerLimit must be put before MaxClients directive!

StartServers            10
MinSpareServers         10
MaxSpareServers         20
ServerLimit             1000
MaxClients              1000
MaxRequestsPerChild     0

Hope this help you!

Load2Test by DBA InfoPower is a the best for MySQL load testing - they can record actual MySQL production traffic and replay it in staging multiple ways, like "as is", "load parallelization", "think time reduction", etc and generate performance issues root cause analysis on a spot.

I used it with number of my customers and it takes no time to do load testing and get analytics and root cause analysis back (like gathering slow-log data without any impact on a system and more), compare runs, etc

They are small startup that operates "under the radar". I got introduced to this product through the person who works there - and I really like the tool!!!

Load2Test official page - load2test.com

Enter null values

Problem: You want to enter a null value into a field using the "Open Table" view.

Solution: Press Ctrl + 0 (zero) with the field highlighted.

Works in Management Studio and Enterprise Manager.

SQL Server 2005: Simple Cursor Example

DECLARE @id INT
DECLARE db_cursor CURSOR FOR
SELECT id From myTable
OPEN db_cursor
FETCH NEXT FROM db_cursor INTO @id

WHILE @@FETCH_STATUS = 0
BEGIN
-- do something
FETCH NEXT FROM db_cursor INTO @id
END

CLOSE db_cursor
DEALLOCATE db_cursor

Show Users
USE MASTER
SELECT * FROM sysprocesses WHERE dbid = DB_ID('dbname')

Disconnect Users

USE master
GO
DECLARE @dbname varchar(30), @spid varchar(10), @start datetime
SELECT @start = current_timestamp, @dbname = 'dbname'
-- Timeout after 5 mts
while(exists(Select * FROM sysprocesses WHERE dbid = db_id(@dbname)) AND
datediff(mi, @start, current_timestamp) < 5)
begin
DECLARE spids CURSOR FOR
SELECT convert(varchar, spid) FROM sysprocesses
WHERE dbid = db_id(@dbname)
OPEN spids
while(1=1)
BEGIN
FETCH spids INTO @spid
IF @@fetch_status < 0 BREAK
exec('kill ' + @spid)
END
DEALLOCATE spids
END
GO

To find the key for your installed product perform the following steps:

1. Click Run
2. Type regedit and click enter
3. Navigate to the following, depending on your version:
CD KEY for SQL 2000.

HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\Microsoft SQL Server\80\Registration

2005 comes "pre-licensed", but you might take a look at HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\Microsoft SQL Server\90\Tools\Setup

Basic procedure for a hot backup

The following PL/SQL demonstrates how a hot backup is taken on Oracle:


for ts in tablespaces
alter tablespace $ts begin backup
for df in datafiles of $ts
cp $df $SAVE/$df
alter database datafile '$df' end backup
end
alter tablespace end backup
end
alter system switch logfile
alter database backup controlfile to trace



The second last step:
alter system switch logfile

Forces a checkpoint which in turn updates the datafile headers with a new SCN that is the same for all datafile headers.
Note: Online redo log files should never be backed up!

Differences while hot backup and normal db operation
If the database is hot backuped, the operation is different from normal db operation in only two ways:

When a database block is modified the first time since the backup started, the entire block is written to the online redo logs.
In normal operation, only the changed bytes are written.
The datafile headers are not updated with the SCN when a checkpoint is performed.

Prerequisites:
Two prerequisites must be met for hot backups to make sense:

1) The database must be in archive log mode
2) All archived redo logs since the last backup are available when a database must be recovered.

Small query to list Oracle tablespaces, that are in backup mode:

SELECT d.tablespace_name, b.time
FROM dba_data_files d, v$­backup b
WHERE
d.file_id = b.FILE#
AND b.STATUS = 'ACTIVE';

#sqlplus /nolog

SQL*Plus: Release 10.2.0.4.0 - Production on Tue May 20 05:08:20 2008
Copyright (c) 1982, 2007, Oracle. All Rights Reserved.

SQL> connect / as sysdba
Connected.

SQL> SELECT d.tablespace_name, b.time
2 FROM dba_data_files d, v$­backup b
3 WHERE
4 d.file_id = b.FILE#A
5 ND b.STATUS = 'ACTIVE';

TABLESPACE_NAME TIME
------------------------------ ---------
USERS 02-JAN-08
SYSAUX 02-JAN-08
UNDOTBS1 02-JAN-08
SYSTEM 02-JAN-08
EXAMPLE 02-JAN-08

SQL>

List Oracle Listeners

pgrep -fl -u oracle 'tnslsnr'

List Oracle Databases

pgrep -fl -u oracle 'ora_pmon'

List Shadow Processes

pgrep -fl -u oracle 'LOCAL'