134

My overall upgrade went well, but I am left with a critically lingering issue of mysql-server not being able to install itself, and nothing I am trying gets it to work.

This is the error I see when trying to install/reinstall it:

Setting up mysql-server-5.7 (5.7.11-0ubuntu6) ...
Job for mysql.service failed because the control process exited with error code. See "systemctl status mysql.service" and "journalctl -xe" for details.
invoke-rc.d: initscript mysql, action "start" failed.
dpkg: error processing package mysql-server-5.7 (--configure):
 subprocess installed post-installation script returned error exit status 1
dpkg: dependency problems prevent configuration of mysql-server:
 mysql-server depends on mysql-server-5.7; however:
  Package mysql-server-5.7 is not configured yet.

dpkg: error processing package mysql-server (--configure):
 dependency problems - leaving unconfigured
No apport report written because the error message indicates its a followup error from a previous failure.
                   Processing triggers for systemd (229-4ubuntu4) ...
Processing triggers for ureadahead (0.100.0-19) ...
Errors were encountered while processing:
 mysql-server-5.7
 mysql-server
E: Sub-process /usr/bin/dpkg returned an error code (1)

I've tried completely removing it, though doing so attempts to install MariaDB because of dependencies (?). Any suggestions on what I can do to fix this would be most welcome.

EDIT: Looks like I'm not the only one: https://bugs.launchpad.net/ubuntu/+source/mysql-5.7/+bug/1573279

Improve this question
5
  • I have same problem here.
    – KernelPanic
    Apr 23, 2016 at 7:46
  • 1
    Had to uninstall phpmyadmin, which was causing the dependency problems somehow. After reinstalling, everything is working fine again.
    – Hinrich
    Apr 23, 2017 at 17:39
  • 1
    do-release-upgrade from Ubuntu 14 to 16 performs an unsupported upgrade from MySQL 5.5 to 5.7, so it is expected MySQL is broken after, as reported here: bugs.launchpad.net/ubuntu/+source/ubuntu-release-upgrader/+bug/…
    – Marco Marsala
    Oct 24, 2017 at 5:24
  • 1
    @MarcoMarsala I don't know if we can say it is expected for MySQL to be broken after the Ubuntu 14 to 16 upgrade, though I suppose it depends on your perspective. To the average user, it is certainly not expected to have an LTS upgrade break something like MySQL. It is surprising that was not caught during testing, though.
    – TheGremlyn
    Oct 24, 2017 at 12:51
  • apt install phpmyadmin --reinstall fixed the above with no need to touch mysql after... go figure
    – Andy
    Apr 9, 2019 at 14:44

10 Answers 10

Reset to default
129

The instructions @andrew-beerman posted are on the right track, though they aren't quite clear to me and seem to recommend more than is necessary. I pieced together the answer from the above and a helpful post in the bug thread.

These are the steps I took to correct this:

  1. Back up your my.cnf file in /etc/mysql and remove or rename it

    sudo mv /etc/mysql/my.cnf /etc/mysql/my.cnf.bak

  2. Remove the folder /etc/mysql/mysql.conf.d/ using 

    sudo rm -r /etc/mysql/mysql.conf.d/

  3. Verify you don't have a my.cnf file stashed somewhere else (I did in my home dir!) or in /etc/alternatives/my.cnf use 

    sudo find / -name my.cnf

  4. Backup and remove /etc/mysql/debian.cnf files (not sure if needed, but just in case)

    sudo mv /etc/mysql/debian.cnf /etc/mysql/debian.cnf.bak
sudo apt purge mysql-server mysql-server-5.7 mysql-server-core-5.7
sudo apt install mysql-server

  5. In case your syslog shows an error like "mysqld: Can't read dir of '/etc/mysql/conf.d/'" create a symbolic link:

    sudo ln -s /etc/mysql/mysql.conf.d /etc/mysql/conf.d

    Then the service should be able to start with sudo service mysql start.

That got it working!

Improve this answer
20
  • 20
    after all these operations (I did it before) I still have same error issues...
    – Serge P
    Apr 29, 2016 at 14:05
  • 5
    You have to remove every my.cnf.* from the /etc/mysql directory. Look for my.cnf.backup, my.cnf.fallback and my.cnf.migrated - all of these files have to be removed too (backup first!)
    – hitzi
    May 3, 2016 at 8:51
  • 8
    @SergiiP sudo find / -name "my.cnf" might come in handy.
    – starbeamrainbowlabs
    May 14, 2016 at 11:31
  • 1
    mysql-server-core- should be mysql-server-core-5.7. Otherwise worked like a charm. Thank you!
    – David Tay
    May 20, 2016 at 12:56
  • 1
    +1. You've made my day! Also morning and evening! I was stuck.
    – Max Yudin
    Jun 7, 2016 at 17:44
63

Today I got the same problem, after trying many solutions i found that the problem was the command sudo systemctl disable mysql.service that i used to disable MySQL auto starting, so to get it working i re-enabled again MySQL server using the command sudo systemctl enable mysql.service and run again the upgrade process and it terminated perfectly.

Improve this answer
10
  • 7
    I ran in to this same problem and the same fix worked for me.
    – Allen
    Apr 26, 2016 at 15:13
  • 6
    Worked for me. Did every solution here, mysql worked after this one.
    – Błażej Michalik
    Apr 29, 2016 at 14:51
  • 7
    Same here - this should be top post.
    – a1phanumeric
    Sep 15, 2016 at 14:32
  • 5
    Worked for me too! Thank you so much. Just to be explicit, run: sudo systemctl enable mysql.service and after that run sudo apt install -f.
    – Fernando Paladini
    Jan 25, 2017 at 0:03
  • This is all I had to do. I had disabled autostart like described by @naruto. I enabled autostart and run "sudo apt upgrade". Now fixed. Should be top answer. This is a much simpler, more standard and less trouble prone solution.
    – Jack Holt
    Aug 7, 2017 at 16:00
21

Your error message contains this line:

subprocess installed post-installation script returned error exit status 1

However, this installed post-installation script is not mentioned by name. After much tinkering, I found out that its name is (in my case) /var/lib/dpkg/info/mysql-server-5.7.postinst.

Open this file with sudo vi /var/lib/dpkg/info/mysql-server-5.7.postinst, or your preferred editor.

At the top, change line 3 (or so): set -e to set -x, save the file. (option -e is "exit on errors", -x means "explicitly show command executed", presumably)

Run sudo dpkg --configure -a --log /tmp/dpkg.log (the --log option is optional). You can also simply run apt upgrade if you know it'll be the only package that will be upgraded.

Now you get verbose output of the mysql-server-5.7.postinst bash script, and you can figure out what's wrong.

In my case it unsuccessfully tried to (re-)run mysql_upgrade, but that was not needed for my customized mysql installation. I was sure I've run it manually before, successfully, and all was well.

So I commmented out line 321 (for older mysqld releases try line 281),

#mysql_upgrade --defaults-file=/etc/mysql/debian.cnf || result=$?

and the command that has failed before, sudo apt upgrade (run it again), finished successfully, and dpkg removed the error status for this package.

Now you can set back the set -x to set -e (mentioned above). And optionally uncomment the mysql-upgrade line.

Extra work might be required if you have moved your mysql data partition to a nonstandard location. I moved mine from /var/lib/mysql/data to a different drive via symlink. Then you might have to remove the symbolic link temporarily, before the postinst script manipulation. Then re-create it after running the package upgrade.

After the next minor version upgrade of the mysqld debian package, this problem with the /var/lib/dpkg/info/mysql-server-5.7.postinst script can show up again.

Improve this answer
4
  • I ended up purging everything and then downloading the community mysql-server version and manually installing from here: dev.mysql.com/downloads/mysql
    – RyanNerd
    Sep 1, 2016 at 19:34
  • Exactly the same thing happened to me, and your steps solved it to. Bur for me my mysql_upgrade call was on line 320. Can you explain why it return a non-zero value when called from postinst script?
    – emiliopedrollo
    Sep 24, 2016 at 18:28
  • @emiliopedrollo No I can't explain here. But I think the line number is now 320 because recently the package maintainers have augmented the postinstall script, I have observed the same thing the other day, during the last run of the software-updater (which included a new mysql-deb-package).
    – knb
    Sep 24, 2016 at 19:25
  • Thanks! With set -e I was able to work out the exact problem - MySQL didn't have access for the ubuntu system user specified in /etc/mysql/debian.cnf . So I added this user to MySQL and granted privileges, ran dpkg again and it worked!
    – Allen Hamilton
    Apr 24, 2017 at 5:17
18

The instructions here fixed it on my server: https://bugs.mysql.com/bug.php?id=72722

I can understand the pain of having your system in inconsistent state but lets not worry about the whole situation and take it step by step to get the system clean.

First lets see the current state of all the mysql packages on machine using: dpkg -l | grep mysql (Please paste the output excluding last column)

The first column denotes the current status of the package. Here are the possible options:

ii) Installed rc) Removed config-files kept (This should be the state of all the packages you have removed with 'apt-get remove' that does not remove config-files under /etc)

For this to work, you will need to run 'apt-get purge <pkg-name>' till you do not see any packages in the above list.

Please remember that some non-mysql-server packages like python-mysql.connector and python-mysqldb, if installed, need not be removed as they do not have any affect on this situation but if removed might cause trouble to applications using them.

We will definitely try to re-visit our docs to see how can we safeguard users from getting into this trouble. Thanks for sharing your feedback in detail with us.

Improve this answer
2
  • 2
    Thanks for dpkg -l | grep mysql . It helped to understand the direction.
    – Max Yudin
    Jun 7, 2016 at 17:47
  • command was wrong for purge it must be sudo apt-get purge <pkg-name> I fixed it with an edit but: I used unicode characters for the < because otherwise it wouldnt display. Be warned copy pasting this command
    – Toskan
    Jul 27, 2019 at 6:37
3

In my case, with strace, I saw that /var/run/mysqld/ didn't exist and mysqld can't create the file mysqld.sock.

These commands solved my problem:

mkdir /var/run/mysqld
chown mysql.mysql /var/run/mysqld
chmod 700 /var/run/mysqld

Now: 

systemctl start mysql

And mysql works again :)

Improve this answer
1
  • I suggest you add these lines to /usr/share/mysql/mysql-systemd-start starting at line 25, then you won't have to recreate this directory after every reboot (sorry the line returns don't work in this comment): if [ ! -d /run/msyqld ]; then mkdir -p -m0755 /run/mysqld || { echo "Unable to create /run/mysqld"; exit 1; } chown mysql:mysql /run/mysqld || { echo "Unable to chown /run/mysqld"; exit 1; } fi
    – scoobydoo
    Jun 30, 2016 at 14:44
3

In my case I could solv the problem by adding

# Allow log file access
/home/system/var/log/mysql.err rw,
/home/system/var/log/mysql.log rw,
/home/system/var/log/mysql/ r,
/home/system/var/log/mysql/** rw,

to /etc/apparmor.d/local/usr.sbin.mysqld

For more details take a look at my answer (by ChristophS) at stackoverflow.

Improve this answer
1

I had the same issue. I tried to re-install mysql several times, but had no success.

I figured out that the problem for me was, that another mysql process was already running.

In details:

After I read carefully logs in /var/log/mysql/error.log, and found:

[ERROR] Can't start server: Bind on TCP/IP port: Address already in use

[ERROR] Do you already have another mysqld server running on port: 3306 ?

Looks like another application was using already the port.

I checked it using ps -aux | grep 3306:

$ ps -aux | grep 3306
milkovs+  6802  0.0  0.0  16336  1084 pts/19   S+   21:39   0:00 grep --color=auto 3306
mysql    14706  0.0  0.3 1270192 13916 pts/2   Sl   Aug19   0:29 /usr/sbin/mysqld --basedir=/usr --datadir=/var/lib/mysql --plugin-dir=/usr/lib/mysql/plugin --user=mysql --skip-grant-tables --log-error=/var/log/mysql/error.log --pid-file=/var/run/mysqld/mysqld.pid --socket=/var/run/mysqld/mysqld.sock --port=3306 --log-syslog=1 --log-syslog-facility=daemon --log-syslog-tag=

And I killed the running process sudo kill -15 14706

Then I started mysql: /etc/init.d/mysql start

Finally mysql works for me! I hope it helps somebody.

Improve this answer
1

For WSl Ubuntu 16.04 version I found this helpful article here

Steps to follow to resolved

  1. sudo vi /var/lib/dpkg/info/mysql-server-5.7.postinst
  2. Look for this line mysql_upgrade --defaults-file=/etc/mysql/debian.cnf || result=$?
  3. Add this service mysql restart directly above line in number 2.
  4. Save and run this sudo dpkg --configure -a

it should look like

service mysql restart
mysql_upgrade --defaults-file=/etc/mysql/debian.cnf || result=$?

Hope this help

Improve this answer
0

None of the answers on this page worked for me.

I ended up going to the Oracle downloads page, downloading mysql-apt-config_0.8.8-1_all.deb, and installing MySQL from Oracle repo:

sudo dpkg -i mysql-apt-config_0.8.8-1_all.deb
sudo apt update
sudo apt install mysql-server
Improve this answer
3
  • 2
    isntall -> install typo. Stupid SO does not allow me to correct one character.
    – Csaba Toth
    Jul 11, 2018 at 16:54
  • Why did you go to the oracle to download mysql
    – Sinscary
    Aug 7, 2019 at 8:40
  • @Sinscary This answer is from awhile back, but it was probably because they had a newer version than the system package manager?
    – ostrokach
    Aug 7, 2019 at 15:42
0

I have had the issue on a few servers now The fix was to run apt install phpmyadmin --reinstall

which resolved the above (with no need to touch mysql afterwards)

Improve this answer

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .