Related
The Environment
We have a staging/test WordPress site on a CentOS server running MariaDB (version 10.3.8). We've been experimenting with a plugin called GeoDirectory (https://wpgeodirectory.com/) The plugin creates a variety of tables in the database. The database also is using the innodb_file_per_table set to ON, so the database has generated IBD files for each tablespace in the database.
The Issue
The staging server had 80 GB of storage. After making a particular change to the settings of the plugin (updating the Place Settings), we noticed that about ten minutes later the staging site was timing out when we tried to access it via the browser.
Logging in via SSH, we noticed that the machine was completely out of space. Looking for the largest files in /var, I noticed that one file was now taking up 70 GB of space (wp_g1a4rar7xx_geodir_gd_place_detail.ibd). This is the IBD file that corresponded to the table where the settings were updated.
Logging into SQL and running a count(*) on the number of records in that table, there were only 6,000 records.
Because the table had blown up to take up the entirety of the disk, even trying OPTIMIZE TABLE wouldn't work because of the lack of space.
The Question
What could have happened to make that file balloon so large so quickly? How can we avoid this happening in the future?
Reckless Speculation
Based on searches, we believe it might have something to do with the rollback/logging that InnoDB performs, and that we might be able to avoid this by making some changes in my.cnf to limiting logging. Again, it could be something completely different.
Here's the current my.cnf from /etc/my.cnf:
[mysqld]
bind-address = 0.0.0.0
# bind-address = ::ffff:127.0.0.1
local-infile=0
datadir=/var/lib/mysql
socket=/var/lib/mysql/mysql.sock
# Disabling symbolic-links is recommended to prevent assorted security
risks
symbolic-links=0
# Settings user and group are ignored when systemd is used.
# If you need to run mysqld under a different user or group,
# customize your systemd unit file for mariadb according to the
# instructions in http://fedoraproject.org/wiki/Systemd
[mysqld_safe]
log-error=/var/log/mariadb/mariadb.log
pid-file=/var/run/mariadb/mariadb.pid
#
# include all files from the config directory
#
!includedir /etc/my.cnf.d
Thanks in advance for your help!
Based on the name of the plugin, I am guessing that the issue could be related to updates of indexed geometry columns. MariaDB Server 10.2 imported the InnoDB support for SPATIAL INDEX from MySQL 5.7, and it looks like the purge of version history is sometimes being skipped, depending on concurrent DML activity.
I have mentioned this peculiar design choice in MDEV-15284, which reports that SELECT on a SPATIAL INDEX can return inconsistent results if a concurrent ROLLBACK is being executed.
I tried to import a large sql file through phpMyAdmin...But it kept showing error
'MySql server has gone away'
What to do?
As stated here:
Two most common reasons (and fixes) for the MySQL server has gone away
(error 2006) are:
Server timed out and closed the connection. How to fix:
check that wait_timeout variable in your mysqld’s my.cnf configuration file is large enough. On Debian: sudo nano
/etc/mysql/my.cnf, set wait_timeout = 600 seconds (you can
tweak/decrease this value when error 2006 is gone), then sudo
/etc/init.d/mysql restart. I didn't check, but the default value for
wait_timeout might be around 28800 seconds (8 hours).
Server dropped an incorrect or too large packet. If mysqld gets a packet that is too large or incorrect, it assumes that something has
gone wrong with the client and closes the connection. You can increase
the maximal packet size limit by increasing the value of
max_allowed_packet in my.cnf file. On Debian: sudo nano
/etc/mysql/my.cnf, set max_allowed_packet = 64M (you can
tweak/decrease this value when error 2006 is gone), then sudo
/etc/init.d/mysql restart.
Edit:
Notice that MySQL option files do not have their commands already available as comments (like in php.ini for instance). So you must type any change/tweak in my.cnf or my.ini and place them in mysql/data directory or in any of the other paths, under the proper group of options such as [client], [myslqd], etc. For example:
[mysqld]
wait_timeout = 600
max_allowed_packet = 64M
Then restart the server. To get their values, type in the mysql client:
> select ##wait_timeout;
> select ##max_allowed_packet;
For me this solution didn't work out so I executed
SET GLOBAL max_allowed_packet=1073741824;
in my SQL client.
If not able to change this with MYSql service running, you should stop the service and change the variable in "my.ini" file.
For example:
max_allowed_packet=20M
If you are working on XAMPP then you can fix the MySQL Server has gone away issue with following changes..
open your my.ini file
my.ini location is (D:\xampp\mysql\bin\my.ini)
change the following variable values
max_allowed_packet = 64M
innodb_lock_wait_timeout = 500
If you are running with default values then you have a lot of room to optimize your mysql configuration.
The first step I recommend is to increase the max_allowed_packet to 128M.
Then download the MySQL Tuning Primer script and run it. It will provide recommendations to several facets of your config for better performance.
Also look into adjusting your timeout values both in MySQL and PHP.
How big (file size) is the file you are importing and are you able to import the file using the mysql command line client instead of PHPMyAdmin?
If you are using MAMP on OS X, you will need to change the max_allowed_packet value in the template for MySQL.
You can find it at: File > Edit template > MySQL my.cnf
Then just search for max_allowed_packet, change the value and
save.
I had this error and other related ones, when I imported at 16 GB SQL file. For me, editing my.ini and setting the following (based on several different posts) in the [mysqld] section:
max_allowed_packet = 110M
innodb_buffer_pool_size=511M
innodb_log_file_size=500M
innodb_log_buffer_size = 800M
net_read_timeout = 600
net_write_timeout = 600
If you are running under Windows, go to the control panel, services, and look at the details for MySQL and you will see where my.ini is. Then after you edit and save my.ini, restart the mysql service (or restart the computer).
If you are using HeidiSQL, you can also set some or all of these using that.
I solved my issue with this short /etc/mysql/my.cnf file :
[mysqld]
wait_timeout = 600
max_allowed_packet = 100M
The other reason this can happen is running out of memory. Check /var/log/messages and make sure that your my.cnf is not set up to cause mysqld to allocate more memory than your machine has.
Your mysqld process can actually be killed by the kernel and then re-started by the "safe_mysqld" process without you realizing it.
Use top and watch the memory allocation while it's running to see what your headroom is.
make a backup of my.cnf before changing it.
I got same issue with
$image_base64 = base64_encode(file_get_contents($_FILES['file']['tmp_name']) );
$image = 'data:image/jpeg;base64,'.$image_base64;
$query = "insert into images(image) values('".$image."')";
mysqli_query($con,$query);
In \xampp\mysql\bin\my.ini file of phpmyadmin we get only
[mysqldump]
max_allowed_packet=110M
which is just for mysqldump -u root -p dbname . I resolved my issue by replacing above code with
max_allowed_packet=110M
[mysqldump]
max_allowed_packet=110M
I updated "max_allowed_packet" to 1024M, but it still wasn't working. It turns out my deployment script was running:
mysql --max_allowed_packet=512M --database=mydb -u root < .\db\db.sql
Be sure to explicitly specify a bigger number from the command line if you are donig it this way.
If your data includes BLOB data:
Note that an import of data from the command line seems to choke on BLOB data, resulting in the 'MySQL server has gone away' error.
To avoid this, re-create the mysqldump but with the --hex-blob flag:
http://dev.mysql.com/doc/refman/5.7/en/mysqldump.html#option_mysqldump_hex-blob
which will write out the data file with hex values rather than binary amongst other text.
PhpMyAdmin also has the option "Dump binary columns in hexadecimal notation (for example, "abc" becomes 0x616263)" which works nicely.
Note that there is a long-standing bug (as of December 2015) which means that GEOM columns are not converted:
Back up a table with a GEOMETRY column using mysqldump?
so using a program like PhpMyAdmin seems to be the only workaround (the option noted above does correctly convert GEOM columns).
If it takes a long time to fail, then enlarge the wait_timeout variable.
If it fails right away, enlarge the max_allowed_packet variable; it it still doesn't work, make sure the command is valid SQL. Mine had unescaped quotes which screwed everything up.
Also, if feasible, consider limiting the number of inserts of a single SQL command to, say, 1000. You can create a script that creates multiple statements out of a single one by reintroducing the INSERT... part every n inserts.
i got a similar error.. to solve this just open my.ini file..here at line no 36 change the value of maximum allowed packet size ie. max_allowed_packet = 20M
Make sure mysqld process does not restart because of service managers like systemd.
I had this problem in vagrant with centos 7. Configuration tweaks didn't help. Turned out it was systemd which killed mysqld service every time when it took too much memory.
I had similar error today when duplicating database (MySQL server has gone away...), but when I tried to restart mysql.server restart I got error
ERROR! The server quit without updating PID ...
This is how I solved it:
I opened up Applications/Utilities/ and ran Activity Monitor
quit mysqld
then was able to solve the error problem with
mysql.server restart
I am doing some large calculations which involves the mysql connection to stay long time and with heavy data. i was facing this "Mysql go away issue". So i tried t optimize the queries but that doen't helped me then i increased the mysql variables limit which is set to a lower value by default.
wait_timeout
max_allowed_packet
To the limit what ever suits to you it should be the Any Number * 1024(Bytes). you can login to terminal using 'mysql -u username - p' command and can check and change for these variable limits.
For GoDaddy shared hosting
On GoDaddy shared hosting accounts, it is tricky to tweak the PHP.ini etc files. However, there is another way and it just worked perfectly for me. (I just successfully uploaded a 3.8Mb .sql text file, containing 3100 rows and 145 cols. Using the IMPORT command in phpMyAdmin, I was getting the dreaded MySQL server has gone away error, and no further information.)
I found that Matt Butcher had the right answer. Like Matt, I had tried all kinds of tricks, from exporting MySQL databases in bite-sized chunks, to writing scripts that break large imports into smaller ones. But here is what worked:
(1) CPANEL ---> FILES (group) ---> BACKUP
(2a) Under "Partial Backups" heading...
(2b) Under "Download a MySQL Database Backup"
(2c) Choose your database and download a backup (this step optional, but wise)
(3a) Directly to the right of 2b, under heading "Restore a MySQL Database Backup"
(3b) Choose the .SQL import file from your local drive
(3c) True happiness will be yours (shortly....) Mine took about 5 seconds
I was able to use this method to import a single table. Nothing else in my database was affected -- but that is what step (2) above is intended to protect against.
Notes:
a. If you are unsure how to create a .SQL import file, use phpMyAdmin to export a table and modify that file structure.
SOURCE:
Matt Butcher 2010 Article
If increasing max_allowed_packet doesn't help.
I was getting the same error as you when importing a .sql file into my database via Sequel Pro.
The error still persisted after upping the max_allowed_packet to 512M so I ran the import in the command line instead with:
mysql --verbose -u root -p DatabaseName < MySQL.sql
It gave the following error:
ASCII '\0' appeared in the statement, but this is not allowed unless option --binary-mode is enabled
I found a couple helpful StackOverflow questions:
Enable binary mode while restoring a Database from an SQL dump
Mysql ERROR: ASCII '\0' while importing sql file on linux server
In my case, my .sql file was a little corrupt or something. The MySQL dump we get comes in two zip files that need to be concatenated together and then unzipped. I think the unzipping was interrupted initially, leaving the file with some odd characters and encodings. Getting a fresh MySQL dump and unzipping it properly worked for me.
Just wanted to add this here in case others find that increasing the max_allowed_packet variable was not helping.
None of the solutions regarding packet size or timeouts made any difference for me. I needed to disable ssl
mysql -u -p -hmyhost.com --disable-ssl db < file.sql
https://dev.mysql.com/doc/refman/5.7/en/encrypted-connections.html
How do I enable the Binary log in mySQL. I have tried:
1) Entering this SQL code on phpmyadmin
SET GLOBAL log_bin ='ON'
This gave me an error message that bin_log is a read-only file;
2) Inserting this into my.ini file below [mysqld]
--log-bin= C:\mySqlbinlog
This displays that I cannot access the database when I try to get to the webpage.
Any idea how I can enable the Binary Log through phpmyadmin or in the my.ini file? Any help appreciated.
Just Put log_bin parameter after mysqld. Please give path for binlog where you have large amount of space because binlog eats lots of disk space.
log_bin = "path where want to store bin files.extension"
You can also use some other parameters like max_binlog_size, max_binlog_cache_size,
sync_binlog for well performance
For More Here's [a link] (http://dev.mysql.com/doc/refman/5.5/en/replication-options-binary-log.html)
According to the documentation, the option-file format is log-bin, so you should put this in your my.ini (under the [mysqld] heading):
log-bin=C:\mySqlbinlog
Don't forget to restart the MySQL server process afterward!
I currently use wamp with MySQL ver 5.6.17 so your milage may vary.
Open your my.ini file and scroll through it. You're looking for the [mysqld] section. Add log-bin=C:\Your_Dir_Here to the [mysqld] section. Restart the MySQL services. Once the restart has finished, login to MySQL. Run the command SHOW BINARY LOGS;
You should then see an output like this:
+------------------+-----------+
| Log_name | File_size |
+------------------+-----------+
| MySQL_Log.000001 | 120 |
+------------------+-----------+
1 row in set (0.00 sec)
NOTE
I don't know exactly how log-bin works but I would recommend an absolute path so you know exactly where your log files will be. Also, your MySQL installation won't read the my.ini file until it's services restart.
Further reading
http://dev.mysql.com/doc/refman/5.6/en/binary-log.html
This link covers a great deal about the binary logs. It literally covers too much for me to even gloss over.
If I've left something out or if I was vague on something, leave a comment and I'll try to reply back.
I tried to import a large sql file through phpMyAdmin...But it kept showing error
'MySql server has gone away'
What to do?
As stated here:
Two most common reasons (and fixes) for the MySQL server has gone away
(error 2006) are:
Server timed out and closed the connection. How to fix:
check that wait_timeout variable in your mysqld’s my.cnf configuration file is large enough. On Debian: sudo nano
/etc/mysql/my.cnf, set wait_timeout = 600 seconds (you can
tweak/decrease this value when error 2006 is gone), then sudo
/etc/init.d/mysql restart. I didn't check, but the default value for
wait_timeout might be around 28800 seconds (8 hours).
Server dropped an incorrect or too large packet. If mysqld gets a packet that is too large or incorrect, it assumes that something has
gone wrong with the client and closes the connection. You can increase
the maximal packet size limit by increasing the value of
max_allowed_packet in my.cnf file. On Debian: sudo nano
/etc/mysql/my.cnf, set max_allowed_packet = 64M (you can
tweak/decrease this value when error 2006 is gone), then sudo
/etc/init.d/mysql restart.
Edit:
Notice that MySQL option files do not have their commands already available as comments (like in php.ini for instance). So you must type any change/tweak in my.cnf or my.ini and place them in mysql/data directory or in any of the other paths, under the proper group of options such as [client], [myslqd], etc. For example:
[mysqld]
wait_timeout = 600
max_allowed_packet = 64M
Then restart the server. To get their values, type in the mysql client:
> select ##wait_timeout;
> select ##max_allowed_packet;
For me this solution didn't work out so I executed
SET GLOBAL max_allowed_packet=1073741824;
in my SQL client.
If not able to change this with MYSql service running, you should stop the service and change the variable in "my.ini" file.
For example:
max_allowed_packet=20M
If you are working on XAMPP then you can fix the MySQL Server has gone away issue with following changes..
open your my.ini file
my.ini location is (D:\xampp\mysql\bin\my.ini)
change the following variable values
max_allowed_packet = 64M
innodb_lock_wait_timeout = 500
If you are running with default values then you have a lot of room to optimize your mysql configuration.
The first step I recommend is to increase the max_allowed_packet to 128M.
Then download the MySQL Tuning Primer script and run it. It will provide recommendations to several facets of your config for better performance.
Also look into adjusting your timeout values both in MySQL and PHP.
How big (file size) is the file you are importing and are you able to import the file using the mysql command line client instead of PHPMyAdmin?
If you are using MAMP on OS X, you will need to change the max_allowed_packet value in the template for MySQL.
You can find it at: File > Edit template > MySQL my.cnf
Then just search for max_allowed_packet, change the value and
save.
I had this error and other related ones, when I imported at 16 GB SQL file. For me, editing my.ini and setting the following (based on several different posts) in the [mysqld] section:
max_allowed_packet = 110M
innodb_buffer_pool_size=511M
innodb_log_file_size=500M
innodb_log_buffer_size = 800M
net_read_timeout = 600
net_write_timeout = 600
If you are running under Windows, go to the control panel, services, and look at the details for MySQL and you will see where my.ini is. Then after you edit and save my.ini, restart the mysql service (or restart the computer).
If you are using HeidiSQL, you can also set some or all of these using that.
I solved my issue with this short /etc/mysql/my.cnf file :
[mysqld]
wait_timeout = 600
max_allowed_packet = 100M
The other reason this can happen is running out of memory. Check /var/log/messages and make sure that your my.cnf is not set up to cause mysqld to allocate more memory than your machine has.
Your mysqld process can actually be killed by the kernel and then re-started by the "safe_mysqld" process without you realizing it.
Use top and watch the memory allocation while it's running to see what your headroom is.
make a backup of my.cnf before changing it.
I got same issue with
$image_base64 = base64_encode(file_get_contents($_FILES['file']['tmp_name']) );
$image = 'data:image/jpeg;base64,'.$image_base64;
$query = "insert into images(image) values('".$image."')";
mysqli_query($con,$query);
In \xampp\mysql\bin\my.ini file of phpmyadmin we get only
[mysqldump]
max_allowed_packet=110M
which is just for mysqldump -u root -p dbname . I resolved my issue by replacing above code with
max_allowed_packet=110M
[mysqldump]
max_allowed_packet=110M
I updated "max_allowed_packet" to 1024M, but it still wasn't working. It turns out my deployment script was running:
mysql --max_allowed_packet=512M --database=mydb -u root < .\db\db.sql
Be sure to explicitly specify a bigger number from the command line if you are donig it this way.
If your data includes BLOB data:
Note that an import of data from the command line seems to choke on BLOB data, resulting in the 'MySQL server has gone away' error.
To avoid this, re-create the mysqldump but with the --hex-blob flag:
http://dev.mysql.com/doc/refman/5.7/en/mysqldump.html#option_mysqldump_hex-blob
which will write out the data file with hex values rather than binary amongst other text.
PhpMyAdmin also has the option "Dump binary columns in hexadecimal notation (for example, "abc" becomes 0x616263)" which works nicely.
Note that there is a long-standing bug (as of December 2015) which means that GEOM columns are not converted:
Back up a table with a GEOMETRY column using mysqldump?
so using a program like PhpMyAdmin seems to be the only workaround (the option noted above does correctly convert GEOM columns).
If it takes a long time to fail, then enlarge the wait_timeout variable.
If it fails right away, enlarge the max_allowed_packet variable; it it still doesn't work, make sure the command is valid SQL. Mine had unescaped quotes which screwed everything up.
Also, if feasible, consider limiting the number of inserts of a single SQL command to, say, 1000. You can create a script that creates multiple statements out of a single one by reintroducing the INSERT... part every n inserts.
i got a similar error.. to solve this just open my.ini file..here at line no 36 change the value of maximum allowed packet size ie. max_allowed_packet = 20M
Make sure mysqld process does not restart because of service managers like systemd.
I had this problem in vagrant with centos 7. Configuration tweaks didn't help. Turned out it was systemd which killed mysqld service every time when it took too much memory.
I had similar error today when duplicating database (MySQL server has gone away...), but when I tried to restart mysql.server restart I got error
ERROR! The server quit without updating PID ...
This is how I solved it:
I opened up Applications/Utilities/ and ran Activity Monitor
quit mysqld
then was able to solve the error problem with
mysql.server restart
I am doing some large calculations which involves the mysql connection to stay long time and with heavy data. i was facing this "Mysql go away issue". So i tried t optimize the queries but that doen't helped me then i increased the mysql variables limit which is set to a lower value by default.
wait_timeout
max_allowed_packet
To the limit what ever suits to you it should be the Any Number * 1024(Bytes). you can login to terminal using 'mysql -u username - p' command and can check and change for these variable limits.
For GoDaddy shared hosting
On GoDaddy shared hosting accounts, it is tricky to tweak the PHP.ini etc files. However, there is another way and it just worked perfectly for me. (I just successfully uploaded a 3.8Mb .sql text file, containing 3100 rows and 145 cols. Using the IMPORT command in phpMyAdmin, I was getting the dreaded MySQL server has gone away error, and no further information.)
I found that Matt Butcher had the right answer. Like Matt, I had tried all kinds of tricks, from exporting MySQL databases in bite-sized chunks, to writing scripts that break large imports into smaller ones. But here is what worked:
(1) CPANEL ---> FILES (group) ---> BACKUP
(2a) Under "Partial Backups" heading...
(2b) Under "Download a MySQL Database Backup"
(2c) Choose your database and download a backup (this step optional, but wise)
(3a) Directly to the right of 2b, under heading "Restore a MySQL Database Backup"
(3b) Choose the .SQL import file from your local drive
(3c) True happiness will be yours (shortly....) Mine took about 5 seconds
I was able to use this method to import a single table. Nothing else in my database was affected -- but that is what step (2) above is intended to protect against.
Notes:
a. If you are unsure how to create a .SQL import file, use phpMyAdmin to export a table and modify that file structure.
SOURCE:
Matt Butcher 2010 Article
If increasing max_allowed_packet doesn't help.
I was getting the same error as you when importing a .sql file into my database via Sequel Pro.
The error still persisted after upping the max_allowed_packet to 512M so I ran the import in the command line instead with:
mysql --verbose -u root -p DatabaseName < MySQL.sql
It gave the following error:
ASCII '\0' appeared in the statement, but this is not allowed unless option --binary-mode is enabled
I found a couple helpful StackOverflow questions:
Enable binary mode while restoring a Database from an SQL dump
Mysql ERROR: ASCII '\0' while importing sql file on linux server
In my case, my .sql file was a little corrupt or something. The MySQL dump we get comes in two zip files that need to be concatenated together and then unzipped. I think the unzipping was interrupted initially, leaving the file with some odd characters and encodings. Getting a fresh MySQL dump and unzipping it properly worked for me.
Just wanted to add this here in case others find that increasing the max_allowed_packet variable was not helping.
None of the solutions regarding packet size or timeouts made any difference for me. I needed to disable ssl
mysql -u -p -hmyhost.com --disable-ssl db < file.sql
https://dev.mysql.com/doc/refman/5.7/en/encrypted-connections.html
How do I profile a MySQL database. I want to see all the SQL being run against a database.
I know you can do this:
set profiling=1;
Run your slow query (eg SELECT * FROM messages WHERE fromaddress='xxx';
SHOW PROFILES;
But this seem to only apply to stuff run on the command line, I want to see the results from running a website.
You want the query log - but obviously doing this on a heavy production server could be... unwise.
That worked for me on Ubuntu.
Find and open your MySQL configuration file, usually /etc/mysql/my.cnf on Ubuntu. Look for the section that says “Logging and Replication”
# * Logging and Replication
# Both location gets rotated by the cronjob.
# Be aware that this log type is a performance killer.
log = /var/log/mysql/mysql.log
or in newer versions of mysql, comment OUT this lines of codes
general_log_file = /var/log/mysql/mysql.log
general_log = 1
log_error = /var/log/mysql/error.log
Just uncomment the log variable to turn on logging. Restart MySQL with this command:
sudo /etc/init.d/mysql restart
Now we’re ready to start monitoring the queries as they come in. Open up a new terminal and run this command to scroll the log file, adjusting the path if necessary.
tail -f /var/log/mysql/mysql.log
There is a commercial product
http://www.webyog.com/en/
You can simply parse the query log in real time. If on linux, you can use tail -f to see the log live
Also, you can try some free software from these guys:
http://hackmysql.com/mysqlsla
MySqlAdministrator have some useful build in features (including logs view), but for logs it have to be run on same machine as database