This work is built on top of https://github.com/jeffbonhag/db2-docker. Kudos to jeffbonhag.
Note the Docker version used to create this was 1.3.0. The DB2 used was Express-C version 10.5 FP4 and DB2 Advanced Enterprise Server version 10.5 FP3.
The first thing is to Download DB2 Express-C. Put the downloaded file v10.5_linuxx64_expc.tar.gz
in the expc
sub-directory. You can also use it to create containers for DB2 server edition, of which the file name should be similar to DB2_Svr_10.5.0.3_Linux_x86-64.tar.gz
.
Note that only gzipped tar ball is supported (
*.tar.gz
or*.tgz
).
Run the build script
$ ./build.sh <db2 install package> [license fiels...]
For DB2 Express-C, it builds two images: bryantsai/db2-expc
and bryantsai/db2-expc:db2_inst_1
. For DB2 server editions, it builds two images: bryantsai/db2-server
and bryantsai/db2-server:db2_inst_1
. The first contains only DB2 base images while the second has the instance created. You should also see aanother "data-only" container named "db2_data_1
" created and is in Exited
status. Never delete container db2_data_1
as it contains the actual instance data files. More on this "data-only" container in the next section.
When you need to create a new DB2 instance, there's a separate shell script "create-inst.sh
" for creating additional DB2 containers:
$ ./create-inst.sh 2
This creates a new image bryantsai/db2-expc:db2_inst_2
(or bryantsai/db2-server:db2_inst_2
for DB2 server editions) as well as the "data-only" container "db2_data_2
".
Use the included shell script:
$ ./run-inst.sh [instance #] [bryantsai/db2-expc or bryantsai/db2-server]
The optional parameter indicates which DB2 instance to run (default is 1). Of course, you have to create the corresponding instance first (using build.sh
).
This run script issues db2start
for you and then puts you in the bash shell. Initially there is no database created yet. You can create one easily after entering the container's shell as "db2inst1
":
> db2 "create db sample"
After the database is created, you can safely logout of the container as the database's data files are stored in another "data-only" container. Run the launch script as many times as you want. You will always be able to connect to the same persisted database content as long as you don't manually remove the "data-only" container (see later section for more details).
When launching from the run script, you might see the following error message. It does not seem to cause any problem for DB2.
$ ./run-inst.sh
SQL1063N DB2START processing was successful.
bash: cannot set terminal process group (7): Inappropriate ioctl for device
bash: no job control in this shell
$
There's another script for starting DB2 contatiner and keep it running in the background:
$ ./run-inst-server.sh [instance #] [bryantsai/db2-expc or bryantsai/db2-server]
Of course, create the database (using run-inst.sh
to get into the container shell) first before you run this.
Also note that since DB2 is not meant to be running multiple server instances against the same set of database data files, you should only keep one DB2 instance running at all time. If you use the launch script to run DB2 instances, it takes care of that since instance containers are named like "db2_inst_1
".
To access the DB2 instance in the container from outside world, you need to lookup Docker's mapped public port (while an instance is running):
$ docker port <container> 50000
0.0.0.0:49178
50000 is the container private port (the one we used inside the container). In this example, 49178 is the port you use to connect to DB2 from external to Docker host.
There's a script included to help clean up everything related to a specific instance's images:
$ ./clean-inst.sh [instance #] [bryantsai/db2-expc or bryantsai/db2-server]
There are 2 problems (with the Docker version used at the time) running DB2 on Docker.
DB2 has a problem where it needs more shared memory than Docker originally provides. So to run DB2 on Docker, you need to run the container in privileged mode, that is:
$ docker run --privileged=true ...
This will increase the max shared memory size. Of course, the included run script already does that for you.
Because DB2 uses O_DIRECT system call, the default Docker AUFS does not work for DB2. There are 2 options to work-around it:
- Use volume.
- Use devicemapper storage mode for Docker. See Switching Docker from aufs to devicemapper.
The first option is used in this code, as the second option requires change of docker daemon options.
Note that we use Persistent volumes with Docker - Data-only container pattern Docker "data-only" container pattern. That's why you would see a container named like "db2_data_1
" using command docker ps -a
. You don't really need to worry about the "data-only" container as long as you don't delete it (it is not actually running).
See also Docker user guide.