10/05 … a very sad day for all apple fans…

2011
10/05
CATEGORY
Unix / Linux
Write comment

Nagios Error 127- the most useless error output in the world

I’m currently working on some optimization in our monitoring system and was struggling determining why the notification plugins failed with an error code 127 (we are using nagios in passive mode so no problem with the other plugins but I guess the issue might also happen if you’re using nagios with active checks)

[1316062525] Warning: Attempting to execute the command "/usr/bin/php -q /usr/local/nagios/addons/misc/nagios_mail.php" resulted in a return code of 127.  Make sure the script or binary you are trying to execute actually exists...

Problem with nagios is that this error sounds like a default error that means also any issues with the commands :
Can be

  • missconfigured commands.cfg
  • issue in the location
  • problem with the rights on the file
  • problem in you Lib path, …

Let’s use our best friend strace to find out what’s wrong with Mr Nagios :

strace -f -r -s 4096 -p  `pidod nagios` -o /var/tmp/strace-nagios.log
gollum@{locate}nagios001:/tmp 140 $ sudo grep php /var/tmp/strace-nagios.log 
972   execve("/bin/sh", ["sh", "-c", "/usr/bin/php -q /usr/local/nagios/addons/misc/nagios_mail.php"], [/* 207 vars */]) = -1 E2BIG (Argument list too long)
29787 write(9, "[1316063729] Warning: Attempting to execute the command \"/usr/bin/php -q /usr/local/nagios/addons/misc/nagios_mail.php\" resulted in a return code of 127.  Make sure the script or binary you are trying to execute actually exists...\n", 248) = 248

The interesting part is : E2BIG (Argument list too long)

Looks like when your environment grows, nagios becomes less handy so you need to make some concession.
Nagios can make your life really easier with the global variables set that allows you to generate some funky reports when you send the notification based on the macro list : Nagios : macrolist.html

But when you reach a certain number of services, you’ll have to pass the variables to your script and get rid of the default full export of the environment macros :
Nagios enable_environment_macros

Anyway it’s still good to see that Nagios can scale on more than 10000 hosts, but as a baby grows, it still needs some attention to ensure that you’re not missing a notification :-)

2011
09/14
CATEGORY
Unix / Linux
monitoring
TAGS



Write comment

Setup mysql master – master replication + netscaler Vips

This is just a quick reminder on my setup of the replication + load balacing :

I’m using a basic master – slave replication for the moment from where I’m doing my mysqldump :

mysqldump -p  nagios_mgr_dB > nagios_manager.dump
mysqldump -p  nagios_dB > nagiosdB.dump

1- Setup the new volume group for the database

lvcreate -L 300G -n database v01
mke2fs -j  /dev/mapper/v01-database
mkdir /database
sudo vim /etc/fstab
/dev/v01/database    /database    ext3    defaults        1       2

2- Install Mysql on both servers

sudo rpm -ivh MySQL-server-5.5.11-1.rhel5.x86_64.rpm
sudo rpm -ivh MySQL-client-5.5.11-1.rhel5.x86_64.rpm

2- Setup the folder for mysql :

mkdir -p /database/databases/innodb
mkdir -p /database/databases/redo
mkdir -p /database/databases/data
mkdir -p /database/databases/tmp
cp -a /var/lib/mysql/mysql/ /database/databases/data
cd /database && chown -R mysql:mysql databases/

3- Start your mysql server & check

sudo /etc/init.d/mysqld start
mysqladmin -u root password rootpassword

4- Setup the master master replication :

4-1. edit the my.cnf

On master 1 (lop-mastermysql001)

log-bin=mysql-bin
expire_logs_days=5
server-id=10
replicate-same-server-id = 0
auto-increment-increment = 2
auto-increment-offset = 2

On master 2 (lop-mastermysql002)

log-bin=mysql-bin
expire_logs_days=5
server-id=11
replicate-same-server-id = 0
auto-increment-increment = 2
auto-increment-offset = 1

(you can also reduce the dB to replicate by using but I’ll write more later about mysql replication) :
/binlog-ignore-db or replicate-do-db/

4-2. Grant the users for replication & set the replication

On Master 1 “lop-mastermysql001″ 172.18.1.1

GRANT REPLICATION SLAVE ON *.* TO 'replication'@172.18.1.2 IDENTIFIED BY 'replica';
CHANGE MASTER TO MASTER_HOST='lop-mastermysql002',  MASTER_USER='replication', MASTER_PASSWORD='replica';

On Master 2 “lop-mastermysql002″ 172.18.1.1

GRANT REPLICATION SLAVE ON *.* TO 'replication'@172.18.1.1 IDENTIFIED BY 'replica';
CHANGE MASTER TO MASTER_HOST='lop-mastermysql002',  MASTER_USER='replication', MASTER_PASSWORD='replica';

and restart your mysqld servers

/etc/init.d/mysqld restart

4-3. – Start the slave

START SLAVE;

and check the status

SHOW SLAVE STATUS\G

4-4. from Master1

CREATE DATABASE nagios_mgr_dB;
CREATE DATABASE nagios_dB;

Re-inject you dump in Master1

mysql -u root -p nagios_mgr_dB < nagios_manager.dump
mysql -u root -p nagios_dB < nagiosdB.dump

You should see the database in both servers and the insert should not breack the replica ~ just take some time to test it :-)

5- Set up the Netscaler for the Load Balancing :

5-1 Add the server

add server "lop-mastermysql001" 172.18.1.1
add server "lop-mastermysql002" 172.18.1.2

5-2 Create your services

add service mastermysql-svc-001 "lop-mastermysql001" TCP 3306 -gslb NONE -maxClient 0 -maxReq 0 -cip DISABLED -usip NO -cltTimeout 9000 -svrTimeout 9000 -CKA NO -TCPB NO -CMP NO
add service MySQL_Master2 "lop-mastermysql002" TCP 3306 -gslb NONE -maxClient 0 -maxReq 0 -cip DISABLED -usip NO -cltTimeout 9000 -svrTimeout 9000 -CKA NO -TCPB NO -CMP NO

5-3 Create you vip

add lb vserver "nagiosdbmaster-vip" TCP 172.22.1.20 3306  -persistenceType NONE -cltTimeout 180

5-4 Bind your servers to the vip

 bind lb vserver "nagiosdbmaster-vip" "lop-mastermysql001"
 bind lb vserver "nagiosdbmaster-vip" "lop-mastermysql002"

5-5 Check your vip

sh lbv nagiosdbmaster-vip
	nagiosdb-master-lb (172.22.1.20:3306) - TCP	Type: ADDRESS 
	State: UP
	Effective State: UP
	Client Idle Timeout: 180 sec
	Down state flush: ENABLED
	Configured Method: LEASTCONNECTION
	Current Method: Round Robin, Reason: Bound service's state changed to UP
	Mode: IP
	Persistence: NONE
	Connection Failover: DISABLED
1) lop-mastermysql001 (172.18.1.1: 3306) - TCP State: UP	Weight: 1
2) lop-mastermysql002 (172.18.1.2: 3306) - TCP State: UP	Weight: 1
 Done

Bonus // my.cnf file

[client]
socket = /database/databases/mysql.sock

[mysqld]

log-bin=mysql-bin
expire_logs_days=1

server-id=10
replicate-same-server-id = 0
auto-increment-increment = 2
auto-increment-offset = 2

socket = /database/databases/mysql.sock
open_files_limit = 64000
user = mysql
datadir = /database/databases/data
max_tmp_tables = 5000
net_read_timeout = 1000
net_write_timeout = 1000

old-passwords

tmpdir = /database/databases/tmp

key_buffer=200M
max_allowed_packet=8M

sort_buffer=8M
sort_buffer_size=8M
read_buffer_size=8M
net_buffer_length=8K
myisam_sort_buffer_size=256M
myisam_max_sort_file_size = 512M

read_buffer_size = 4M
sort_buffer_size = 8M

max_connections=16384
key_buffer_size = 512M

table_cache=16384
thread_cache=64

query_cache_type=2
query_cache_size=256M
query_cache_limit=1M

max_connect_errors=10000

innodb_open_files = 1000

innodb_data_home_dir = /database/databases/innodb
innodb_data_file_path = ibdata1:100M:autoextend:max:48G
innodb_file_per_table

innodb_buffer_pool_size=4G
innodb_additional_mem_pool_size=32M
innodb_flush_method=O_DIRECT

innodb_log_files_in_group = 2
innodb_log_group_home_dir = /database/databases/redo
innodb_log_file_size=512M
innodb_log_buffer_size=8M

innodb_flush_log_at_trx_commit=2
innodb_autoextend_increment=32
innodb_thread_concurrency=8
innodb_file_io_threads=8

innodb_support_xa=0
innodb_lock_wait_timeout=50

[mysqldump]
quick
max_allowed_packet=16M

[mysql]
no-auto-rehash

[myisamchk]
key_buffer=512M
sort_buffer=32M
read_buffer=32M
write_buffer=32M

[mysqlhotcopy]
interactive-timeout

2011
09/08
CATEGORY
Unix / Linux
Write comment

Log Changes in your dB using Triggers in MySQL

Most of the time when you allow the users to manage a database through some command line/gui tools, you need keep a trace, log, all the changes per user to know at least who to contact if something’s missing.

Instead of coding a double insert in the application which is not really the purpose of the application, you can use integrated triggers to keep a trace of the changes in your Database.
The good thing is that you will keep the integrity of the history table despite the changes in your application.

Let’s start with the exemple :

We have a table called Host with some information that I want to keep (ip/name) if someone try to change them
select column_name,DATA_TYPE from information_schema.columns where table_name=’Host’;

+-------------------+-----------+
| column_name              | DATA_TYPE   |
+-------------------+-----------+
| Host_id                         | int                |
| Host_name                    | varchar        |
| Host_ip                         | varchar         |
| Host_timestamp            | timestamp   |
| Host_lastuser                | varchar         |
+-------------------+-----------+

Here is the History table

mysql> desc History_table;
+-------------------------+--------------+------+-----+-------------------+-----------------------------+
| Field                   | Type         | Null | Key | Default           | Extra                       |
+-------------------------+--------------+------+-----+-------------------+-----------------------------+
| History_table_id        | int(11)      | NO   | PRI | NULL              | auto_increment              |
| History_table_name      | varchar(255) | YES  |     | NULL              |                             |
| History_table_field     | varchar(255) | YES  |     | NULL              |                             |
| History_table_id_field  | int(11)      | YES  |     | NULL              |                             |
| History_table_old_value | varchar(255) | YES  |     | NULL              |                             |
| History_table_new_value | varchar(255) | YES  |     | NULL              |                             |
| History_table_user      | varchar(255) | YES  |     | NULL              |                             |
| History_table_timestamp | timestamp    | NO   |     | CURRENT_TIMESTAMP | on update CURRENT_TIMESTAMP |
+-------------------------+--------------+------+-----+-------------------+-----------------------------+

Let ‘s create a trigger that will copy an old value to the history table if it is changed when someone edits a row.
I’m gonna keep the username of the last user that change it ‘Host_lastuser’ in ‘History_table_user’
The values that have been changed (only if they have been changed) : ‘History_table_old_value’/'History_table_new_value’
You can do that for different table so I will put the name of the table in ‘History_table_name’
the name pof the field in ‘History_table_field’ and keep the primary keep value of the Host table if everything in the table has been changed ‘History_table_id_field’

DELIMITER $$
 
DROP TRIGGER IF EXISTS insert_host$$
DROP TRIGGER IF EXISTS update_host$$
DROP TRIGGER IF EXISTS delete_host$$
 
CREATE TRIGGER insert_host AFTER INSERT on Host
FOR EACH ROW
BEGIN
        INSERT INTO History_table (
                        History_table_id,
                        History_table_name,
                        History_table_field,
                        History_table_id_field,
                        History_table_old_value,
                        History_table_new_value,
                        History_table_user,
                        History_table_timestamp
                )
                VALUES
                ('','Host','Host_add',NEW.Host_id, NEW.Host_name, NEW.Host_id,NEW.Host_lastuser,NOW());
END$$
CREATE TRIGGER delete_host BEFORE DELETE on Host
FOR EACH ROW
BEGIN
        INSERT INTO History_table (
                        History_table_id,
                        History_table_name,
                        History_table_field,
                        History_table_id_field,
                        History_table_old_value,
                        History_table_new_value,
                        History_table_user,
                        History_table_timestamp
                )
                VALUES
                ('','Host','Host_delete',OLD.Host_id,OLD.Host_name,OLD.Host_ip,OLD.Host_lastuser,NOW());
END$$
CREATE TRIGGER update_host AFTER UPDATE on Host
FOR EACH ROW
BEGIN
        IF (NEW.Host_name != OLD.Host_name) THEN
                INSERT INTO History_table (
                        History_table_id,
                        History_table_name,
                        History_table_field,
                        History_table_id_field,
                        History_table_old_value,
                        History_table_new_value,
                        History_table_user,
                        History_table_timestamp
                )
                VALUES
                ('','Host','Host_name',NEW.Host_id, OLD.Host_name, NEW.Host_name,NEW.Host_lastuser,NOW());
        END IF;
        IF (NEW.Host_ip != OLD.Host_ip) THEN
                INSERT INTO History_table (
                        History_table_id,
                        History_table_name,
                        History_table_field,
                        History_table_id_field,
                        History_table_old_value,
                        History_table_new_value,
                        History_table_user,
                        History_table_timestamp
                )
                VALUES
                ('','Host','Host_ip',NEW.Host_id, OLD.Host_ip, NEW.Host_ip,NEW.Host_lastuser,NOW());
        END IF;
END$$

More information about this triggers

{OLD|NEW}

are mysql keywords to get the value before and after the commit.

{AFTER INSERT | BEFORE DELETE}

are used to keep the initial and final values of the entry.

The DELIMITER Option needs to be set because ‘;’ is already used for the statements inside the trigger, $$ is commonly used for that….

FYI The mysql in the app looks like

"UPDATE Host SET 
                                                Host_name='#{@_hostinfo['name_host']}',
                                                Host_ip='#{@_hostinfo['ip_host']}',
                                                Host_timestamp='#{@_hostinfo['timestamp_host']}',
                                                Host_lastuser='#{$app_username}'
                                WHERE Host_id='#{@_hostinfo['id_host']}'
                                "

with

@_hostinfo['timestamp_host']=Time.now.strftime("%Y-%m-%d %H:%M:%S").to_s
$app_username=Etc.getlogin

Here is a sample result :

mysql> select * from History_table ORDER BY History_table_id DESC LIMIT 2\G
*************************** 1. row ***************************
       History_table_id: 21867
     History_table_name: Host
    History_table_field: Host_ip
 History_table_id_field: 5977
History_table_old_value: 10.10.10.1
History_table_new_value: 10.10.10.2
     History_table_user: ronan
History_table_timestamp: 2011-06-28 18:14:02
*************************** 2. row ***************************
       History_table_id: 21866
     History_table_name: Host
    History_table_field: Host_ip
 History_table_id_field: 5976
History_table_old_value: 10.10.11.1
History_table_new_value: 10.10.11.2
     History_table_user: ronan
History_table_timestamp: 2011-06-28 18:14:02
2 rows in set (0.00 sec)
2011
06/28
CATEGORY
Unix / Linux
Write comment