Can the following command on a live server

mysqldump -h XXXX -u root -pXXXXX --skip-opt --flush-privileges 
    --flush-logs --routines --triggers --allow-keywords 
    --create-options --add-locks --quick --single-transaction 
    --extended-insert --all-databases

cause a web application (which relies on databases of server) to hang?

Some possibilities:

• MyISAM tables only support table-level locking, so while your mysqldump is reading the table, any writes will be queued up until it’s finished. If a write gets queued, any other reads will then be queued up behind the write, effectively locking the application. There’s nothing special about mysqldump in this sense, it’s just a single SELECT * FROM table that takes long enough for a write to happen and get queued.

• InnoDB tables support row-level locking, meaning that reads and writes can be interleaved as long as they don’t touch the same rows. There are some query types that do lock the entire table (such as ALTER table ...) but if you were running these frequently they would have caused troubles long before mysqldump did.

• A mysqldump can also cause a lot more disk access than usual. It is effectively reading in your entire database and writing it back out again, probably to the same physical disk. Reading from and writing to the same physical disk is going to cause extra latency (unless they are SSDs) because the parts of the disk that you are reading from and writing to have to be different and hence there will be more seeks, which take time. This will raise the load average on your box and will cause any other operations that require disk access to go slower.

• Most databases store more data than they have RAM. Your mysqldump may well have flushed your filesystem cache and filled it with mostly useless data, meaning that other disk access requests will have to come from the actual disks (which have excess latency at the time) rather than coming from the much faster cache. It may have also flushed caches within MySQL, depending on the limits you have set up for each of them.

• Depending on your MySQL config, the mysqldump that touches every table may have cause MySQL to use more RAM than you have available. This would cause some of your active RAM to be swapped out to disk, further increasing the disk utilisation and having an enormous impact on performance.

Run mysqltuner.pl on your system to help tune your config and consider switching to InnoDB rather than MyISAM if you are currently using MyISAM. Try streaming your backups straight off to your backup server rather than writing them to the same disk. (This can be achieved many ways but this works: mysqldump [OPTIONS] | gzip | ssh "cat - > database_backup.$(date +%F).sql.gz" <BACKUPHOST>) If (or once) you are on InnoDB, you can use Percona’s XtraBackup as a better backup tool that is roughly as efficient as a pure file copy and can be used to initialise replication. XtraBackup can stream directly to another server as well.

Check more discussion of this question.

Sometimes on my VPS I get the error

Warning: mysql_pconnect(): Can't connect to local MySQL server through socket '/var/run/mysqld/mysqld.sock' (111) in /var/www/mysql.php on line 8

It can be solved by restarting MySQL several times, restating the sever or sometimes using mysqld command.

Using the mysqld command I get the following:

root@web:~# mysqld
130109 10:47:23 [Note] Plugin 'FEDERATED' is disabled.
130109 10:47:23 InnoDB: The InnoDB memory heap is disabled
130109 10:47:23 InnoDB: Mutexes and rw_locks use GCC atomic builtins
130109 10:47:23 InnoDB: Compressed tables use zlib 1.2.3.4
130109 10:47:23 InnoDB: Initializing buffer pool, size = 128.0M
130109 10:47:23 InnoDB: Completed initialization of buffer pool
130109 10:47:23 InnoDB: highest supported file format is Barracuda.
InnoDB: The log sequence number in ibdata files does not match
InnoDB: the log sequence number in the ib_logfiles!
130109 10:47:23  InnoDB: Database was not shut down normally!
InnoDB: Starting crash recovery.
InnoDB: Reading tablespace information from the .ibd files...
InnoDB: Restoring possible half-written data pages from the doublewrite
InnoDB: buffer...
130109 10:47:24  InnoDB: Waiting for the background threads to start
130109 10:47:25 InnoDB: 1.1.8 started; log sequence number 43428438
130109 10:47:25 [Note] Server hostname (bind-address): '0.0.0.0'; port: 3306
130109 10:47:25 [Note]   - '0.0.0.0' resolves to '0.0.0.0';
130109 10:47:25 [Note] Server socket created on IP: '0.0.0.0'.
130109 10:47:25 [Note] Event Scheduler: Loaded 0 events
130109 10:47:25 [Note] mysqld: ready for connections.
Version: '5.5.28-0ubuntu0.12.04.3'  socket: '/var/run/mysqld/mysqld.sock'  port: 3306  (Ubuntu)
130109 10:47:25 [ERROR] mysqld: Table './sever/table' is marked as crashed and should be repaired
130109 10:47:25 [ERROR] mysqld: Table './sever/table' is marked as crashed and should be repaired
130109 10:47:25 [ERROR] mysqld: Table './sever/table' is marked as crashed and should be repaired
130109 10:47:25 [Warning] Checking table:   './sever/table'

My get only 50 visitors a day and uses more than 3000 queries per hour.
My VPS is only 1 core with 256MB ram and runs on Ubuntu 64bit.

MySQL downtimes directly affect my website and I can’t find a answer anywhere.

Thanks in Advance

I just upgraded to 512MB of Ram and it stopped. I think it is the OOM-killer mentioned by Ladadadada. you can see a Anseer for the OOM-Killer here.

Check more discussion of this question.

I need to count impressions of a .php script that is accessed via web. Right now I do this in a 1 line MySQL table and just increment the value. This works fine usually but when the server has traffic bursts (hundreds of requests/second) everything gets delayed and the increment SQL statements start to pop up in MySQL’s slow queries log.
I tried to count via APC but it “loses” values, that is, when I run a simple ab test command like

ab -n 100 -c 10 http://example.com/script.php

only ~50 new impressions are counted while on MySQL 100 are counted.
I also tried to count impressions via memcache, it works better (that is, its able to keep up with more simultaneous connections) but it also fails to count everything after 50 or 100 simultaneous connections.

I was wondering if there is anybody aware of a good solution to this problem?

In the end I chose to use a MEMORY table instead the usual INNODB one. While the solution isn’t perfect it works much better than just using INNODB for this. I also applied this solution on a number of other tables where data integrity wasn’t crucial and the results were positive.

Check more discussion of this question.

I just installed MySQL 5.5 and right after install I was asked to set root password by executing those two commands:

/usr/bin/mysqladmin -u root password 'new-password'
/usr/bin/mysqladmin -u root -h linux-opxd password 'new-password'

or

/usr/bin/mysql_secure_installation

I executed first command — OK. I executed second command this time with error:

error: ‘Access denied for user ‘root’@'localhost’ (using password:
NO)’

I re-executed the first one (for checking) — error as above.

So, I executed mysql_secure_installation. Everything went fine and my password was recognized. So I executed the first two commands again — and error again.

As another check:

mysql --user=root

The same error as above.

I tried resetting the password with mysql authentication, also I tried /usr/bin/mysql_upgrade -u root -p — still the error while trying to launch mysql console as root.

So finally I turned authentication off, connected to console and executed this:

SELECT user();

The output — root@. If I connect like this:

mysql --user='root@localhost'

I will get root@localhost@.

So for now it seems it is not the problem with privileges, or lack of password or anything like this, but no recognizing the user which was given at command line — mysql appends @ at the end and this seems to be the problem (at least for my eye).

The question is: how to make MySQL recognize the given user?

The first error you have received:

error: ‘Access denied for user ‘root’@'localhost’ (using password: NO)’

indicates that you probably did not use -p option in mysql command.

Once you have created mysql root password, you should be able to access MySql using:

mysql -u root -p

Where you will be prompted to enter MySql root password.

If you want to enable access from some other location (different from the localhost) you can grant privileges using SQL GRANT commands similar to this:

mysql> CREATE USER 'username'@'localhost' IDENTIFIED BY 'some_pass';
mysql> GRANT ALL PRIVILEGES ON databaseName.* TO 'username'@'localhost' WITH GRANT OPTION;
mysql> FLUSH PRIVILEGES;

On a similar fashion, you can enable MySql DB access from some other hostname/IP address or any location by changing ‘username’@'localhost’ to: ‘username’@'someFQDN’, ‘username’@'IP_address’, ‘username’@'%’.

For additional information I’d recommend: http://dev.mysql.com/doc/refman/5.1/en/adding-users.html.

Cheers!

Check more discussion of this question.

I’ve moved all the stuff in /var/lib/mysql to /500gb/mysql, and created a symbolic link from one to the other. I then did a chown mysql:mysql and chmod 777 (with both -h and -R as necessary) to both the /var/lib/mysql symbolic link, and also /500gb/mysql.

ls -al /var/lib gives me:

lrwxrwxrwx  1 mysql         mysql         15 2011-07-27 12:55 mysql -> /500gb/mysql

When I try to start MySQL (service mysql start), I get the following in /var/log/mysql/error.log:

110727 12:48:32 [Note] Plugin 'FEDERATED' is disabled.
/usr/sbin/mysqld: Can't find file: './mysql/plugin.frm' (errno: 13)
110727 12:48:32 [ERROR] Can't open the mysql.plugin table. Please run mysql_upgrade to create it.
110727 12:48:32  InnoDB: Operating system error number 13 in a file operation.
InnoDB: The error means mysqld does not have the access rights to
InnoDB: the directory.
InnoDB: File name ./ibdata1
InnoDB: File operation call: 'open'.
InnoDB: Cannot continue operation.

Both /500gb/mysql/ibdata1 and /500gb/mysql/mysql/plugin.frm exist and are owned by mysql. What’s the deal?

Also, MySQL seems to occasionally clobber my symbolic link. It disappears and becomes a regular directory after a number of failed starts.

I’ve also tried editing my /etc/mysql/my.cnf and setting datadir = /500gb/mysql with no luck.

Any ideas? Thanks.

It turns out that Apparmor was screwing things up. We had to edit this file: /etc/apparmor.d/usr.sbin.mysql.

Check more discussion of this question.

For some reason my /var/run/ dir on debian6 had it’s chown and chmod changed. Problem is from what? Outcome is that mysql fails. I’m guessing the easiest way to fix is to reinstall MySQL. (Previously this LAMP was running fine.)

However, I have established databases and tables and don’t want to lose them. Since mysql is not running I can’t backup (correct?).

I just want to confirm that:

1). if I reinstall, I won’t loose my databases.

2). apt-get install mysql-server mysql-client is the correct command to use.

Thanks.

EDIT:

/var/log/syslog output when running /etc/init.d/mysql start:

mysqld_safe: Starting mysqld daemon with databases from /var/lib/mysql
mysqld: 130109 10:44:10 [Note] Plugin 'FEDERATED' is disabled.
mysqld: #007/usr/sbin/mysqld: Can't find file: './mysql/plugin.frm' (errno: 13)
mysqld: 130109 10:44:10 [ERROR] Can't open the mysql.plugin table. Please run mysql_upgrade to create it.
mysqld: 130109 10:44:10 InnoDB: The InnoDB memory heap is disabled
mysqld: 130109 10:44:10 InnoDB: Mutexes and rw_locks use GCC atomic builtins
mysqld: 130109 10:44:10 InnoDB: Compressed tables use zlib 1.2.3.4
mysqld: 130109 10:44:10 InnoDB: Using Linux native AIO
mysqld: 130109 10:44:10 InnoDB: Initializing buffer pool, size = 128.0M
mysqld: 130109 10:44:10 InnoDB: Completed initialization of buffer pool
mysqld: 130109 10:44:10  InnoDB: Operating system error number 13 in a file operation.
mysqld: InnoDB: The error means mysqld does not have the access rights to
mysqld: InnoDB: the directory.
mysqld: InnoDB: File name ./ibdata1
mysqld: InnoDB: File operation call: 'create'.
mysqld: InnoDB: Cannot continue operation.
mysqld_safe: mysqld from pid file /var/run/mysqld/mysqld.pid ended
/etc/init.d/mysql[4531]: 0 processes alive and '/usr/bin/mysqladmin --defaults-file=/etc/mysql/debian.cnf ping' resulted in
/etc/init.d/mysql[4531]: #007/usr/bin/mysqladmin: connect to server at 'localhost' failed
/etc/init.d/mysql[4531]: error: 'Can't connect to local MySQL server through socket '/var/run/mysqld/mysqld.sock' (2)'
/etc/init.d/mysql[4531]: Check that mysqld is running and that the socket: '/var/run/mysqld/mysqld.sock' exists!
/etc/init.d/mysql[4531]: 

it would be good to find out what really has happened; you can take a look at any other computer with the same OS and mysql installed and use chmod and chown to restore the expected rights and ownership.

quick look at a random debian server tells me:

root@serv:/var/run# ls -la /var/run/mysqld/*
total 12
drwxr-xr-x  2 mysql root  4096 Oct 24 13:29 .
drwxr-xr-x 14 root  root  4096 Jan  9 01:28 ..
-rw-rw----  1 mysql mysql    5 Oct 24 13:29 mysqld.pid
srwxrwxrwx  1 mysql mysql    0 Oct 24 13:29 mysqld.sock

so try first:

chown mysql:mysql /var/run/mysqld/*
chown mysql:root /var/run/mysqld 

and appropriate chmods if needed

if you decide to re-install – be on the safe side:

• try to shut down cleanly mysql – if needed using regular kill and waiting
• copy all the data files [nut just few - eg not just the directory of one database; it might not be enough for innodb]
• copy all the config files

only then try to uninstall + reinstall

Check more discussion of this question.

I want to start mysql with the command argument “–log=log_file_name”

What is the proper way to do that when starting it with /etc/init.d?

Would it be like this?
/etc/init.d/mysql start –log=log_file_name

To make it simple you can create another entry in init.d to start the mysql with that logdir path option. Make a script like /etc/init.d/mysql-log and put following entries in it:

 #!/bin/sh -e
 set -e
 COMMAND=$1
 LOG="--log=/tmp/mysql.log"
 case $COMMAND in
 start)
      /etc/init.d/mysql $COMMAND $LOG
      ;;
 stop)
      /etc/init.d/mysql $COMMAND
      ;;
 restart)
      /etc/init.d/mysql stop
      /etc/init.d/mysql start $LOG
      ;;
 *)
      exit 1
 esac

Set the log file location in the above script as per your needs and start the mysql with the following command:

/etc/init.d/mysql-log start 

This way you can use different scripts for different occasions.

Check more discussion of this question.

I want to use another server to store my databases, so how can I redirect MySQL queries to another server?

You can use something like mysql proxy link1 link2 for it. But it’s not very well supported. Additionally, decrease in performance is somewhat significant. If possible, I would suggest you to simply tell your application to connect to the actual mysql server with a choice. For example, with PHP, there is mysqlnd_ms which allows multiple mysql servers to be added and it load balances among them.

Check more discussion of this question.

I have a MySQL (actually mariaDB) install that I just started replicating to another server. I want to ensure that I’ve setup SSL correctly. The SSL portion of SHOW SLAVE STATUS looks like:

Master_SSL_Allowed: Yes
Master_SSL_CA_File: /etc/mysql/newcerts/ca-cert.pem
Master_SSL_CA_Path: /etc/mysql/newcerts/
Master_SSL_Cert: /etc/mysql/newcerts/client-cert.pem
Master_SSL_Cipher:
Master_SSL_Key: /etc/mysql/newcerts/client-key.pem
Seconds_Behind_Master: 0
Master_SSL_Verify_Server_Cert: No

Is this typical or show that replication is using SSL? Is there something else I can check to ensure it’s using SSL? The slave is successfully replicating.

Inspect the relevant traffic with tcpdump/wireshark.

Check more discussion of this question.

Okay let me get straight to the point, I am a complete noob, and have never done stuff like this at all, I have read tutorial after tuorial but I cant get anything to work. When I tried to install the rpm file I got this error

rpm -Uvh ftp://ftp.pbone.net/mirror/rpms.famillecollet.com/enterprise/6/test/x86_64/php-mysqlnd-5.4.9-1.el6.remi.x86_64.rpm
Retrieving ftp://ftp.pbone.net/mirror/rpms.famillecollet.com/enterprise/6/test/x86_64/php-mysqlnd-5.4.9-1.el6.remi.x86_64.rpm
warning: /var/tmp/rpm-tmp.ez4vvd: Header V3 DSA/SHA1 Signature, key ID 00f97f56: NOKEY
error: Failed dependencies:
        php-pdo(x86-64) = 5.4.9-1.el6.remi is needed by php-mysqlnd-5.4.9-1.el6.remi.x86_64

so I tried installing that rpm file and got this error

rpm -ivh ftp://ftp.pbone.net/mirror/rrpms.famillecollet.com/enterprise/6/test/x86_64/php-pdo-5.4.6-1.el6.remi.x86_64.rpm
Retrieving ftp://ftp.pbone.net/mirror/rrpms.famillecollet.com/enterprise/6/test/x86_64/php-pdo-5.4.6-1.el6.remi.x86_64.rpm
curl: (9) Server denied you to change to the given directory
error: skipping ftp://ftp.pbone.net/mirror/rrpms.famillecollet.com/enterprise/6/test/x86_64/php-pdo-5.4.6-1.el6.remi.x86_64.rpm - transfer failed

I used the ftp links because I have no idea how else to get them to the server. I think I am getting overly frustrated with this, but I have to get this driver installed for any of my scripts to function correctly.

Any help would be greatly appreciated!

Try to add the Remi repository to yum and install it that way. Yum will automatically deal with all the dependencies for you, so you don’t have to follow the breadcrumb path all the way to the final dependency. Saves you lot of time and frustration

As root user, run these commands:

wget http://rpms.famillecollet.com/enterprise/remi.repo -O /etc/yum.repos.d/remi.repo
yum --enablerepo=remi-test install php-mysqlnd

The first command will retrieve the repository configuration file for Remi’s repo. The second command will install the php-mysqlnd package and will also use the Remi repo to install it.

Check more discussion of this question.