Convert DB2 to Aurora Postgres

Purpose

Demo converting DB2 database to postgresql using DB2 on an EC2 instance with SCT and DMS. Additional option to use Kinesis in between DMS and Aurora.

Outline

Overview
AWS Services Used
Technical Overview
Instructions
DB2 Setup
Windows Steps
DB2 with DMS
- DB2 Replication
- DB2 Drop Foreign Keys
Create DMS Resources
Cleaning up

Overview

Initially used a CloudFormation template from Data Migration Immersion Day. This template has been modified to include an IBM DB2 instance with DMS components. This readme documents using SCT/DMS to convert DB2 sample database to Aurora PostgreSQL.

AWS Services Used

Technical Overview

Bring up DMS/SCT environment using modified immersion days template
Review Security Group Settings
Setup VNC viewer on redhat instance. VNC is needed for using IBM DB2 Setup Wizard
A db2 installation response file is available for a silent install to avoid need for VNC
Install DB2 and create sample DB2 database
Use SCT and DMS to convert sample DB2 database to Aurora PostgreSQL
Additional DMS setup to use kinesis

Instructions

*IMPORTANT NOTE: Creating this demo application in your AWS account will create and consume AWS resources, which will cost money. Costing information is available at AWS DMS Pricing The template will cost money for the other resources as well.

ADDITIONAL NOTE: Latest IBM DB2 version supported as of January 2021 is 11.1 with all Fix Packs. I worked through the entire process using IBM DB2 v11.5 and everything works until the very last step to move the data with DMS. This step fails to find the DB2 tables and no data is moved. DB2 11.5 is not supported by DMS and there is not an ETA for that support. Free trial downloads of earlier DB2 version are not available. I successfully used DB2 11.1 FixPack 5 for this exercise. I am unclear on usage limitation with this downloadable content so use this at your own discression.

Create Environment

Start with AWS Account instructions
After reviewing "Introduction" and "Getting Started", follow the Regular AWS Account instructions.
Complete the "Login to the AWS Console" and "Create an EC2 Key Pair" steps
In the "Configure the Environment" step, use the provided ./templates/DB2Workshop.yaml file instead of the link. Choose SQL Server for the source database

Edit Security Group Settings

Additional ports need to be open to allow VNC connectivity to the redhat 8 instance to install DB2. Additionally, using additional agents for DMS, can require additional ports to be open

Find the security group. There are two security group created with the template. Click on the InstanceSecurityGroup (not the DMSSecurityGroup)
Tighten security on the RDP rule. Currently it is open to public
- two choices-can open the ports to amazon ip using the amazon VPN. problem vnc will not work while on amazon VPN. vnc is needed for DB2 install.
  - it is best to log out of amazon vpn and use the non-amazon ip address
  - get amazon ip address using amazon checkip URL
  - get ip address using a "get my ip" search
  - a separate option is to use tightvnc client on the windows VM (vnc viewer is flaky on windows)
- Click "Edit Inbound Rules"
- on the RDP inbound rule, remove "0.0.0.0/16" and put in the address obtained in checkip with a /32 e.g. "1.2.3.4/32"
- open all internal communication on private. Easy way is to change the inbound rule with Access Type of Oracle-RDS to All TCP.
- Click Add Rule and add a new rule for VNC access from the address found in checkip. Use port range of 5900-6000
- Click Add Rule and add a new rule for SSH access from the address found in checkip. Use inbound rule type "SSH"
- Click "save rules"

Setup VNC

VNC is needed to do the IBM install on the redhat instance unless a silent install is chosen with the response file. If using silent install, this section on VNC setup can be skipped. I have provided a response file in this github at db2 install response file This could eliminate the need for VNC. Additionally, there is db2_install script which may work without VNC. I have not verified either of these approaches so these instructions require VNC setup.

To login from client to redhat instance

ssh -i keypairFile ec2-user@redhatIPaddress

Install needed packages to redhat instance- NOTE: git is not needed for VNC but needed for later

sudo yum groupinstall 'Server with GUI'
sudo yum install pixman pixman-devel
sudo yum -y install tigervnc-server git

change ssd to allow password login with ssh

sudo vi /etc/ssh/sshd_config
Set the ChallengeResponseAuthentication yes
Set the PasswordAuthentication yes
# save the file
# restart sshd
sudo systemctl restart sshd.service

disable firewalld

sudo systemctl stop firewalld
sudo systemctl disable firewalld

define port 1 for ec2-user

sudo bash
echo ":1=ec2-user" >> /etc/tigervnc/vncserver.users

set up VNC password as ec2-user

vncpasswd
Password:
Verify:
Would you like to enter a view-only password (y/n)? n

set up an os password for ec2-user probably easier if you choose same password as VNC password (be logged in as ec2-user)

sudo passwd ec2-user

add a vnc config file as ec2-user

echo "session=gnome" >> ~/.vnc/config
echo "alwaysshared" >> ~/.vnc/config

ensure these two lines in /etc/tigervnc/vncserver-config-mandatory

session=gnome
alwaysshared

setup mandatory vncserver
change vi /etc/gdm/custom.conf under the default section WaylandEnable=false DefaultSession=gnome-xorg.desktop
start and enable tigervnc

sudo systemctl start vncserver@:1
sudo systemctl enable vncserver@:1
sudo systemctl set-default graphical

follow disable se linux URL

Troubleshoot VNC

Check the vnc process to ensure it is fully running

sudo systemctl status vncserver@:1
● vncserver@:1.service - Remote desktop service (VNC)
   Loaded: loaded (/usr/lib/systemd/system/vncserver@.service; enabled; vendor preset: disabled)
   Active: active (running) since Tue 2021-01-05 02:37:52 UTC; 26s ago
  Process: 2599 ExecStart=/usr/libexec/vncsession-start :1 (code=exited, status=0/SUCCESS)
 Main PID: 2605 (vncsession)
    Tasks: 1 (limit: 98872)
   Memory: 1.8M
   CGroup: /system.slice/system-vncserver.slice/vncserver@:1.service
           ‣ 2605 /usr/sbin/vncsession ec2-user :1

Jan 05 02:37:52 xxxxx.ec2.internal systemd[1]: Starting Remote desktop service (VNC)...
Jan 05 02:37:52 xxxxx.ec2.internal systemd[1]: Started Remote desktop service (VNC).

Redhat 8 PDF on gnmome and remote access
screen saver lock out says "Cloud User" this is actually ec2-user so enter the OS password for ec2-user
TightVNC on windows is easiest connection (VNC Viewer on windows did not work)
VNC Viewer on Mac seems to work but not with Amazon VPN so it becomes inconvenient especially with mac port used in security group

DB2 Setup

install prerequisite libraries

sudo bash
yum install -y libstdc++.i686
yum install -y pam.i686
yum install -y gcc-c++ cpp gcc kernel-devel make patch
yum install -y libaio

DB2 needs VNC to do the install with the setup wizard. There is also a silent option not needing vnc with a response file. In the github home the response file is db2response

Download DB2

NOTE: At the time, db2 11.1 fixpack 5 is most recent version supported by AWS DBS
Using the VNC connection to the IBM server, go to download page for DB2 software DB2 software
- an IBMID will be needed for this (it is free)
Scroll down to "DB2 Fix Packs" button and click on the version
Scroll down and click the download link for
Scroll down on this page to find the
Download the file. This is the name of the file: v11.1.4fp5_linuxx64_server_t.tar.gz
create holding file for DB2 Software (not necessary but makes later documentation easier)

sudo bash
cd /home
mkdir software
chmod 777 software
exit

Move the downloaded db2 installation file to the /home/software and decompress the file using tar -xvzf
rename directory so what was the server_t directory becomes ibm-db2

sudo bash
cd /home/softare
tar -xvzf "tar file name"
mv server_t ibm-db2

Install DB2

There are two ways to do this. Both methods use db2setup. If VNC is setup use the setup wizard by starting db2setup with any parameters. If VNC is not available, do a silent install using the provided db2 response file in the github

Do a root based install using ./db2setup
If VNC is available

sudo bash
cd /home/software/ibm-db2/
./db2setup

if VNC is not available
- copy db2server.rsp file to redhat server from the githbu

ssh -i "path to ssh key file" db2server.rsp ec2-user@"ip address for node":/home/software/ibm-db2

* run db2setup with response file

sudo bash
cd /home/software/ibm-db2/
./db2setup -r db2server.rsp
# must set a password for db2inst1 user
passwd db2inst1

db2 validation

sudo bash
su - db2inst1
db2val

Troubleshoot DB2 install

Obviously, skip this if everything is fine. These are steps in case of problems. If going well skip forward to DB2 Sample Database

this IBM link IBM troubleshoot is helpful. Odd that it says to ignore the 32 bit libraries.
here are some maybe overly complex install instructions complex install instructions
the DB2 install is hard to fix if there is a mistake. It is easier to re-install
To re-install DB2, make sure everything is deleted. These delete steps help but are not complete.

These are steps to delete DB2-THIS WILL REMOVE YOUR DB2!!!

sudo bash
su - db2inst1
# not all these steps may be needed but without drop can fail with sql1035N
db2 connect to sample
db2 unquiesce db
db2 connect reset
db2 drop database sample
exit
#  now back as root
# get DB2 scripts in the profile
. /home/db2inst1/sqllib/db2profile
cd /opt/ibm/db2/V11.1
db2stop force
db2 terminate
/opt/ibm/db2/V11.1/instance/db2idrop db2inst1
/opt/ibm/db2/V11.1/install/db2_deinstall -a 
# if you don't delete this user, db2 will create db2inst2
# if you don't remove the home directory, the files are left owned by previous UID causing problems
# this will fail if any processes are running like vnc for db2inst1
# to make sure, stop db2inst1 vncserver is running
systemctl stop vncserver@:2
userdel db2inst1
userdel db2fenc1
# check for any other db2 users and delete if found
grep db2 /etc/passwd
rm -rf /home/db2inst1
rm -rf /home/dbfenc1
rm -rf /opt/ibm/db2/V11.1

DB2 sample database

create sample database. For more information look at documentation sample database db2sampl

sudo bash
su - db2inst1
db2sampl

get familiar with DB2 and sample database using this document The install directions are dated but the rest is very good. Be careful what steps you run as some of these commands are "impactful"

Setup VNC for db2inst1

This is a very optional step but handy for using VNC type tools with db2inst1. Only do this if VNC was configured above

define port 2 for db2inst1

sudo bash
echo ":2=db2inst1" >> /etc/tigervnc/vncserver.users

set up VNC password as db2inst1 (be logged in as db2inst1)

vncpasswd
Password:
Verify:
Would you like to enter a view-only password (y/n)? n

start and enable tigervnc second session

sudo systemctl start vncserver@:2
sudo systemctl enable vncserver@:2
sudo systemctl set-default graphical

Windows Steps

Install Windows DB2 Client

download free trial download db2 trials
click on download for free
select "Microsoft Windows(x64) Download"
expand the software zip file
expand in software tree to find "setup.exe" and double click
find "Install Data Server Client". follow prompts to install

Install IBM Data Studio

On Windows machine, download IBM Data Studio client software Download Data Studio
Click on download link for Windows
After download completes, expand the zip file
Within the ibm_ds4130_win folder, Double-click the launchpad.exe and follow the prompts to install

Install git

Git is needed to pull this repository onto the windows machine. The repository has some needed scripts.

Got to the git download
Download the latest git release for windows
Double-click on the git exe file and follow the prompts to install
clone the git project using a windows DOC command line

cd "Desktop\DMS Workshop"
git clone https://github.com/jphaugla/awsDB2ToAuroraPostgres.git

SCT

Return back to the DMS and SCT steps using the SQL Server to Amazon Aurora PostgreSQL

Start back at this point in the guide
Perform the following Part 1 Schema Conversion Steps: "Connect to the EC2 Instance", "Install the AWS Schema Conversion Tool (AWS SCT)"
Restart the Windows Server (this seems to be important after the SCT install)
Create a New Project using New Project Wizard
Connect to DB2
- if requested for the db2 LUW driver point at db2jcc4.jar file (usually there are several of them after installing the drivers and the client drivers)
Accept the risks
Click on the "DB2INST1" instance on the left panel and click "Next" to generate the assessment report
Click Next and enter parameters for Aurora PostgreSQL connection
- To find the password, look back in the original templates/DMSWorkshop.yaml in the repository home
- Click "Finish"
Right click on the "DB2INST1" database in the left panel and select Convert Schema to generate the data definition language (DDL) statements for the target database.
Right click on the db2inst1 schema in the right-hand panel, and click Apply to database. click Yes
Can see the tables now in the target database. Use the pg4admin tool to verify the new schema exists but is not populated.

Troubleshooting Windows

if can't connect through ports
- Disable Windows Defender Disable Defender
- Restart the windows machine (yep, it is not 1995 but restarting a windows machine never hurts!)

DB2 with DMS

Several links for background on changes needed for DB2 setup with DMS

Link for using IBM DB2 as a source for DMS DB2 DMS
Link for changing logarchmeth1 on db2 11.5 Turn on Log
Link to change Configuration Parameters. Configuration Parameter Change
NOTE: LOGARCHMETH1 and LOGARCHMETH2 are for newer versions of DB2. Older versions use LOGRETAIN

DB2 Replication

Choose most comfortable method for making the configuration parameter. Following steps are using a command line unix connection

Turn on replication from unix connection to the redhat Db2 instance

su - db2inst1
db2 connect to sample
# turn replication on
db2 update db cfg for sample using logarchmeth1 logretain
# database must be deactivated and reactivated.  THis is the easiest method but less aggressive methods may also work
db2stop 
# if this does not work and this message appears "SQL1025N  The database manager was not stopped because databases are still active."
db2stop force
# restart
db2start
# backup must be taken once logretain is enabled or following error will be returned "SQL1116N A connection to or activation of the database cannot be made because of BACKUP PENDING."
db2 backup database sample

DB2 Drop Foreign Keys

Open the pgadmin tool by clicking on the elephant icon on the bottom of the windows server This is very similar to the immersion day Configure the Target DataBase step Configure Target but using the drop_constraints from the git repo

Using the tree structure on the left, find the AuroraDB database and then the db2inst1 schema and click on the schema
Use Tools->Query Tool to open the query tool
click on the "open folder" icons on far left of the icons for pagamin query tool and open the drop_constraints script
- the script is at C:\Users\Administrator\Desktop\DMS Workshop\awsDB2ToAuroraPostgres\scripts\drop_constraints.sql

Create DMS Resources

These resources are automatically created if the flag is set in the script but the task must be started manually

Some choices here. In addition, can add a separate Migration Task using the same DB2 source tables but with a target Kinesis endpoint. In this scenario, there are no mapping rules but a separate Kinesis stream is needed for each table. The streams are defined in the cloud formation template as well.

Create DMS Replication Instance

Follow steps from immersion training

Create DMS Endpoints

If the parameter "CreateDMSComponents" in the initial Cloudformation template was set to true, these components will already be created and need validation. Test the source DB2 endpoint. Verify aurora and kinesis endpoint tests was successful.

Follow these Steps
Use these parameters for the source
THe specific PostgreSQL endpoints, KinesisEndpoints and TargetKinesis roles are output in the cloudformation

Create IBM to Aurora Task

Similar to this create DMS Migration Task If the parameter "CreateDMSComponents" in the initial Cloudformation template was set to true, the Task will already be created. Just start it.

add a selection rule where schema name is like "DB2INS%"
DB2 uses upper case schema, table, and column names so these must all be converted in mapping rules
add 3 separate mapping rules for columns, tables and schema
using the wizard you need this. (instead can put JSON in directly-down below)

where schema name is like '%' and table name is like '%' and column name is like "%", convert-lowercase
where schema name is like '%' and table name is like '%', convert-lowercase
where schema name is like '%'  convert-lowercase

JSON to use directly

{"rules":[{"rule-type":"transformation","rule-id":"1","rule-name":"1","rule-target":"column","object-locator":{"schema-name":"%","table-name":"%","column-name":"%"},"rule-action":"convert-lowercase","value":null,"old-value":null},{"rule-type":"transformation","rule-id":"2","rule-name":"2","rule-target":"table","object-locator":{"schema-name":"%","table-name":"%"},"rule-action":"convert-lowercase","value":null,"old-value":null},{"rule-type":"transformation","rule-id":"3","rule-name":"3","rule-target":"schema","object-locator":{"schema-name":"%"},"rule-action":"convert-lowercase","value":null,"old-value":null},{"rule-type":"selection","rule-id":"4","rule-name":"4","object-locator":{"schema-name":"DB2INST%","table-name":"%"},"rule-action":"include","filters":[]}]}

Upgrading to current verson of DMS
This is the error if 11.5 is used

2021-01-08T01:36:14 [TASK_MANAGER ]E: No tables were found at task initialization. Either the selected table(s) no longer exist or no match was found for the table selection pattern(s). [1021707] (replicationtask.c:2107)

Create IBM to Kinesis Task

Add a kinesis Stream with a shard for each table planning to be moved to kinesis
Add a kinesis endpoint with these settings:
- MessageFormat: json
- ServiceAccessRoleArn "Kinesis Role ARN" see this link for role requirements
- StreamArn: use kinesis stream ARN from above
- Could not get this to work on the cloudformation but add these two parameters
  - IncludePartitionValue: true
  - PartitionIncludeSchemaTable: true
Create DMS Migration Task
- SourceEndpoint is DMSDB2Endpoint
- TargetEndpoint is DMSKinesisEndpoint
- JSON mapping string

{"rules":[{"rule-type":"selection","rule-id":"1","rule-name":"1","object-locator":{"schema-name":"DB2INST%","table-name":"EMPLOYEE"},"rule-action":"include","filters":[]},{"rule-type":"selection","rule-id":"2","rule-action":"include","object-locator":{"schema-name":"DB2INST%","table-name":"DEPARTMENT"},"rule-name":"2"}]}

Cleaning up

Remove all files from S3 to avoid accruing unnecessary charges

Name		Name	Last commit message	Last commit date
Latest commit History 23 Commits
README_PHOTOS		README_PHOTOS
scripts		scripts
templates		templates
DB2 LUW.pdf		DB2 LUW.pdf
README.md		README.md
db2server.rsp		db2server.rsp

jphaugla/awsDB2ToAuroraPostgres

Folders and files

Latest commit

History

Repository files navigation