The /opt/openshift folder is specific to OpenShift and we want to make this image consumable outside OpenShift as well. We should use more generic folder (/usr/local?)

Make sure exactly which users get created during initialization, what their access rights are, and so on. This should be mainly done by looking into mysql.user table. We should also document as much as possible the behaviour of (not) setting MYSQL_ROOT_USER.

Ideally, we'd run the container with USER mysql, or at least drop the privileges as soon as possible.

Starting from the example template with persistent storage, where a root password is not set, trying set it triggers an error and deployment fails:

[vagrant@openshiftdev origin]$ oc env dc mysql --list                                                                 
# deploymentconfigs mysql, container mysql
MYSQL_USER=user0PG
MYSQL_PASSWORD=qwNhkOiCXe51w1wu
MYSQL_DATABASE=sampledb
[vagrant@openshiftdev origin]$ oc env dc mysql MYSQL_ROOT_PASSWORD=NewAdminPassword
deploymentconfigs/mysql
[vagrant@openshiftdev origin]$ oc get pods
NAME             READY     STATUS    RESTARTS   AGE
mysql-2-deploy   1/1       Running   0          2m
mysql-2-q7bst    0/1       Running   5          59s
[vagrant@openshiftdev origin]$ oc logs mysql-2-q7bst
Starting local mysqld server ...
Waiting for MySQL to start ...
150901 11:18:09 [Warning] One can only use the --user switch if running as root

150901 11:18:09 [Note] Plugin 'FEDERATED' is disabled.
150901 11:18:09 InnoDB: The InnoDB memory heap is disabled
150901 11:18:09 InnoDB: Mutexes and rw_locks use GCC atomic builtins
150901 11:18:09 InnoDB: Compressed tables use zlib 1.2.7
150901 11:18:09 InnoDB: Using Linux native AIO
150901 11:18:09 InnoDB: Initializing buffer pool, size = 128.0M
150901 11:18:09 InnoDB: Completed initialization of buffer pool
150901 11:18:09 InnoDB: highest supported file format is Barracuda.
150901 11:18:09  InnoDB: Waiting for the background threads to start
Waiting for MySQL to start ...
150901 11:18:10 InnoDB: 5.5.37 started; log sequence number 1599071
150901 11:18:10 [Warning] 'user' entry 'root@mysql-1-lke7j' ignored in --skip-name-resolve mode.
150901 11:18:10 [Warning] 'user' entry '@mysql-1-lke7j' ignored in --skip-name-resolve mode.
150901 11:18:10 [Warning] 'proxies_priv' entry '@ root@mysql-1-lke7j' ignored in --skip-name-resolve mode.
150901 11:18:10 [Note] Event Scheduler: Loaded 0 events
150901 11:18:10 [Note] /opt/rh/mysql55/root/usr/libexec/mysqld: ready for connections.
Version: '5.5.37'  socket: '/tmp/mysql.sock'  port: 0  MySQL Community Server (GPL)
ERROR 1133 (42000) at line 1: Can't find any matching row in the user table

We should take a look what changes got to the default mysql image on docker hub (https://github.com/docker-library/mysql), whether there is something worth take an inspiration from.

I was reading readme on https://github.com/sclorg/mysql-container/tree/master/8.0, but the registry URL mentioned here are not working

registry.access.redhat.com/rhscl/mysql-80-rhel7
https://hub.docker.com/r/centos/mysql-80-centos7/

We should cover docker alone and OpenShift.

Keep the volume only for storing MySQL databases. No need to have extra folder.

The use case I'd like to support is when somebody migrates database from one machine to this container. Then the user typically has an SQL dump and wants to do just this:

• run the container without anything being initialized
• import data as root

The solution may be to have MYSQL_DATABASE, MYSQL_USER and MYSQL_PASSWORD optional when MYSQL_ROOT_PASSWORD is set.

FYI @mnagy

When using root access I am not able to create multiple databases inside a single container. I get grant privileges error.

mysql> CREATE DATABASE demoDB;
Query OK, 1 row affected (0.00 sec)
mysql> GRANT ALL PRIVILEGES ON demoDB.* TO 'demo'@'%' IDENTIFIED BY 'demo';
ERROR 1044 (42000): Access denied for user 'root'@'%' to database 'demoDB'

To resolve this I have to:

mysql> UPDATE mysql.user SET Grant_priv='Y', Super_priv='Y' WHERE User='root';
mysql> FLUSH PRIVILEGES;
mysql> GRANT ALL ON *.* TO 'root'@'localhost';
Query OK, 0 rows affected (0.00 sec)
mysql> GRANT ALL PRIVILEGES ON demoDB.* TO 'demo'@'%' IDENTIFIED BY 'demo';
Query OK, 0 rows affected (0.00 sec)

By default, root user does not have grant and super privileges.

Any help is appreciated. I need to make sure the mysql 5.6 images works out of the box for me otherwise I will have to rebuild a new image.

when i create a database like this
"docker run -d --name mysql -e MYSQL_USER=user -e MYSQL_PASSWORD=pass -e MYSQL_DATABASE=db -e MYSQL_CHAR_SET=utf8 -e MYSQL_COLLATION=utf8_unicode_ci -p 3306:3306 rhscl/mysql-57-rhel7"

but the character_set_database and character_set_server is still latin1...
what should i do to solve this

Thanks for providing this useful container.

I am doing a research aiming at finding issues in configuration files. I have a question about one MySQL config: max_binlog_cache_size. It seems the official document says "The maximum recommended value is 4GB; this is due to the fact that MySQL currently cannot work with binary log positions greater than 4GB."

However, the default value is 18446744073709551615, which is much larger than the recommended value.

Shall we change the value to 4GB?
Thanks!

Current way of passing passwords using -e is not the best one, since even root (or whoever can talk to docker socket) should be able to read the passwords. However, with -e way, even if the variable is unset inside the container, one can read the value from docker inspect output.

This issue is meant to track an RFE to be able to specify password somehow differently, maybe in addition to the current -e option. This new way shouldn't allow to see the password even for docker admin/root.

Need to change the replication/binlog format to "ROW" instead of default STATEMENT.
Would like also more control over the my.cnf files.

PR on the way

Document the needed :Z option for volumes so SELinux works properly. This needs to be done for all our database images.

Is there a supported/existing way to add plugins to this MySQL container? I see a plugin-dir option in my-base.cnf.template for scl compatibility, but nothing seems to call --plugin-load or INSTALL PLUGIN.

https://dev.mysql.com/doc/refman/5.5/en/server-options.html#option_mysqld_plugin-load

I want to build an image that comes up with no extra initialization steps needed in docker-compose or openshift.

The my.cnf should be a template where we do env substitution to provide basic customization for the mysql server to users. For beginning we should have same set of ENV vars as we had in v2. Consider this for future.

I used mysql 5.7 to make images,it was success;but I got a error when I do docker run.
error :
': not a valid identifierpts/mysql/scl_enable: line 2: unset: ENV is probably not installed.-mysql57 /usr/share/container-scripts/mysql/common.sh: line 2: $'\r': command not found : No such file or directorys/mysql/common.sh: line 3: /usr/share/container-scripts/mysql/helpers.sh /usr/share/container-scripts/mysql/common.sh: line 4: $'\r': command not found /usr/share/container-scripts/mysql/common.sh: line 9: $'\r': command not found /usr/share/container-scripts/mysql/common.sh: line 12: $'\r': command not found /usr/share/container-scripts/mysql/common.sh: line 13: syntax error near unexpected token $'{\r''
'usr/share/container-scripts/mysql/common.sh: line 13: `function export_setting_variables() {
/usr/bin/run-mysqld: line 6: export_setting_variables: command not found

My host :
docker version
Client:
Version: 1.12.3
API version: 1.24
Go version: go1.6.3
Git commit: 6b644ec
Built:
OS/Arch: linux/amd64

Server:
Version: 1.12.3
API version: 1.24
Go version: go1.6.3
Git commit: 6b644ec
Built:
OS/Arch: linux/amd64

On OpenShift Online Service Catalog, the link to README.md points to a symlink instead the actual README.md file. The "View Documentation" link also points to this symlink.

The link is: https://github.com/sclorg/mysql-container/blob/master/5.7/README.md

The real file is located in: https://github.com/sclorg/mysql-container/blob/master/5.7/root/usr/share/container-scripts/mysql/README.md

The same goes to mariadb-container.

I lost track of the other issue, but mysql is still causing flakes in the push job:
https://ci.openshift.redhat.com/jenkins/view/Image%20Verification/job/push_images/18216/console

@mnagy @mfojtik
(if the other issue is still open, feel free to close this one).

Replication templates can be unified as was done for PostgreSQL recently. This'll also allow us to make extended tests nicer.

I'm trying out the replication example, and first thing I did was compare the source for the two images 5.5 and 5.6.

The scripts mysqld-master, mysqld-slave, run-mysqld, run-mysqld-master, run-mysqld-slave differ in apparently unnecessary and strange ways between 5.5 and 5.6.

Using a tool like meld to compare the dirs 5.5 and 5.6 reveals what I mean:

I'm confused by the diffs.

• Why does 5.5 export MYSQL_RUNNING_AS_MASTER=1 and 5.6 doesn't?
• Why log_volume_info $MYSQL_DATADIR lines appear in different places (5.5 != 5.6)?
• Why does the GRANT REPLICATION statements differ?
• Why does 5.5 has some extra code related to binlog that 5.6 doesn't have?

Debugging our database images has been a bit of a pain for me, especially when I do not have access to the machine where the container in question is being run (think QE and cooperation through issues/bugzilla bugs). I often find myself in a need to craft a custom image with debugging statements which is often tedious and can waste a lot of time (especially if OpenShift does not use the image I want).

It's hard to make sense from the container logs. I don't know if something is being printed out by the entrypoint script, database initialization script, or the final exec'd binary. In case of either MySQL or PostgreSQL (don't remember which), the last line runs along the lines of "server is shutting down". This is from the pre-exec'd process and it doesn't indicate failure. For a user who does not know how the image works, this must be extremely confusing.

Proposed solutions:

1. Let's have a unified way of printing out information about what we are doing and use that. Let's tell the user that we're starting a mysql/postgres daemon just to check/modify the database. Let's tell him we're calling the final exec. Let's tell him when and what failed. Let's make sure the output from entrypoint script stands out and that it's recognized who prints out what.
2. Let's have a debugging mode. It could at least do a set -x. This will be helpful for us. We're just going to tell our users to set a variable and send us output from that. We should also check out the volumes. We often have to deal with permission-related bugs or user mistakes (nfs). Putting couple of ls -la statements on volumes might provide useful information. As well as calling id.

@bparees @thesteve0 @rhcarvalho @hhorak @praiskup @eliskasl

• ReplicationController should probably be replaced with DeploymentConfig, since the template doesn't create any RC explicitly, but rather a DC.
• The example starts with 1 slave and not 3.

I found the following script in the file "root-common/usr/libexec/container-setup":

#  some config options are only valid for particular versions
if [ "`version2number $MYSQL_VERSION`" -ge "800" ] ; then
    rm -f "${CONTAINER_SCRIPTS_PATH}/cnf/10-mysql57.cnf"
fi

-------

cat 10-mysql57.cnf

[mysqld]
#  http://www.chriscalender.com/ignoring-the-lostfound-directory-in-your-datadir/
ignore-db-dir=lost+found

However: mysql8 on ocp/okd also needs this parameter, why is it deleted? I think this file 10-mysql57.cnf should be retained, just the file name leads to misunderstanding.

Let's allow users to setup separately the data directory. Something like that might be quite handy for users: run it once with docker run --rm -e MYSQL_PASSWORD=* ... run-mysqld-setup.sh, which configures the storage area, then run with docker run ...

We have experienced mysql pods crashing many times over the past few weeks due to OOM errors.

So to avoid this happening unexpectedly we started monitoring memory usage.

There is currently no notification available on OpenShift online for a memory threshold being reached.

So we have devised a custom solution using a script which periodically checks the RSS of the mysql process and enters a row in the database if it breaches a threshold (and we can then send an email to notify our support team).

However, after a pod restart, the script (which runs as a daemon) needs to be started manually.

We tried starting this daemon using the kubernetes hook infrastructure, but it did not work.

Would it be possible to add a custom hook into the mysql image so the daemon could start automatically on restart?

When running a 5.5 pod on openshift, what is the procedure for performing an upgrade to version 5.7? Basically I am looking at upgrading but just replacing the image with the 5.7 version will not work due to changes that need to be made to tables etc.

Could someone explain why the Replication Example is not "not meant to be used in any production"?

2017-05-04T11:49:23.523858Z 0 [Note] InnoDB: Retrying to lock the first data file
2017-05-04T11:49:24.525227Z 0 [ERROR] InnoDB: Unable to lock ./ibdata1 error: 11
2017-05-04T11:49:24.525305Z 0 [Note] InnoDB: Check that you do not already have another mysqld process using the same InnoDB data or log files.
2017-05-04T11:49:25.526448Z 0 [ERROR] InnoDB: Unable to lock ./ibdata1 error: 11

can we get build tags (i.e., centos/mysql-56-centos7:1.0) so we don't have to pull "latest" in our kubernetes rc/dc files?

A weird syntax error pops up while starting a MySQL container:

Version: '5.7.21'  socket: '/tmp/mysql.sock'  port: 0  MySQL Community Server (GPL)
---> 13:17:49     MySQL started successfully
/usr/bin/run-mysqld: line 24: [: missing `]'
=> sourcing 40-datadir-action.sh ...

Line 24 in run-mysqld says:

if [ -z "${MYSQL_ROOT_PASSWORD:-}" || is_allowing_connection_with_empty_password ]; then

while it should be:

if [ -z "${MYSQL_ROOT_PASSWORD:-}" ] || [ is_allowing_connection_with_empty_password ]; then

See https://github.com/sclorg/mysql-container/pull/215/files#r190365336

Should the example/replica template even have a PVC for the slave?
If so, how do we refactor the reference PR/commit to allow slave pod restart per the openshift/origin extended tests I referenced?

thanks

All of the current environment variables are mandatory even though we
no longer need them, e.g. when the database is already initialized.

The documentation under 5.6/README.md describes usage pretty well, but I miss some general description of the image in few sentences. Something like "This image is meant to be used for most common use cases, provides... etc.". We should also document how to extend the image for more specific use cases.

2017-04-27T06:11:16.439317Z 2 [Note] Access denied for user 'mysql'@'localhost' (using password: NO)
---> 06:11:16 MySQL started successfully
---> 06:11:16 Setting passwords ...
ERROR 1133 (42000) at line 1: Can't find any matching row in the user table

We have to make sure that all our database images works well with the NFS volume mount inside OpenShift.

@mnagy you have to sync with @markturansky about how to setup the Volume for MySQL pod to use the NFS server you have to run in Vagrant box. Then you should deploy the MySQL, make it create database and insert some records. Then you should re-deploy MySQL (you can just change some random Env var in PodSpec) and verify that the data you created are persisted. (you will need set ConfigChange trigger for that)

The documentation in 5.6/examples/replica/README.md is a copy-paste of 5.5/examples/replica/README.md, but the replication strategies are different.

5.5 uses binlog, and 5.6 uses GTID but that's not reflected in the documentation.

I see the mysql container restarting every 10-20seconds. I see this exit code in docker ps:

36b961f1d7f2        docker.io/centos/mysql-56-centos7@sha256:6803ba424dbebce4ea577882e6e37ecf368892cd4804162a7c90f7ef307bbdc1                     "container-entrypoint"   5 minutes ago       Exited (137) 3 minutes ago

There is nothing in container logs (at least nothing that smells). The mysql process just exit. From the docs, the 137 code means: No more records (read after end of file)... In some case, the 137 in docker means "container killed".

@mnagy can you please have a look at this? It is pretty annoying, because the DB get up and keep running for ~10s and then container restart, cutting all connections... then it come back again... annoying :/

No architecture dependencies should be hard coded into the Dockerfiles

assert_login_access always succeed.

That means that assertions like assert_login_access $USER "${PASS}_foo" false never fail.

Hi when i try to add the mysql image to any project that i create in the Openshift Web Console i see the following message:

Failed Scheduling | 0/1 nodes are available: 1 Insufficient memory.

And the mysql Pod stays in a pending state of deployment.

Any advice or suggestions as to how to resolve this would be greatly appreciated.

If any more information is required please let me know

At present the default latin1 charset is used for created databases. It would be great if we could adjust this through env variables like: DEFAULT_CHARSET and DEFAULT_COLLATION.

We would probably like to log queries for example.

For example the "extend-image" example can be tested like this:

 oc new-app mysql:5.7~https://github.com/sclorg/mysql-container.git \
   --context-dir=examples/extend-image \
   --name my-mysql-rhel7d \
   MYSQL_OPERATIONS_USER=opuser \
   MYSQL_OPERATIONS_PASSWORD=oppass \
   MYSQL_DATABASE=opdb \
   MYSQL_USER=user \
   MYSQL_PASSWORD=pass

After this, connecting using opuser:oppass should work.

Currently, running mysql command in a running container requires a lot of command line arguments, like hostname, username and password. I would be in favour of enabling the unix socket. This is the easiest way to do it and is the same way we do it with PostgreSQL.

Also, I'd like to create a bash script in standard path named "mysql" that will scl_source the collection and do an exec mysql. This way docker exec -ti mysql as well as oc exec will work out of box and will in my opinion greatly enhance the user experience. This I think we should also do for PostgreSQL and MongoDB images.

@mfojtik @bparees thoughts? cc @hhorak

for runing openshift tests and customize for this containers' use-case.

Example:
sclorg/nginx-container#35

And its diff for easier application:
https://patch-diff.githubusercontent.com/raw/sclorg/nginx-container/pull/35.diff

I have found during load testing of an application on OpenShift Online v3 that MySQL memory continues to grow until eventually the container is killed due to OOM.

I am using mysql57.

After doing some internet research, I found a link saying that mysql does not release memory with linux THP set to true and that if THP is disabled, this issue does not occur:

https://bugs.mysql.com/bug.php?id=84003
(final comment at bottom of page)

Please could someone look into this as it is potentially a very serious issue for anyone running mysql in OpenShift Online.

Thanks