EditAttachPrintable
r1 - 2014-10-30 - 21:40:39 - TimFeeneyYou are here: TWiki >  Deployment Web > OnlineBackupCLM > OnlineBackupCLMDB2

Transitioning to online backup of CLM databases using IBM DB2 todo.png

Authors: HariVetsa
Build basis: v4.0.x and 5.0 Jazz applications in CLM

Summary

IBM DB2 is among the supported database servers that can be used to store repository data for Rational Team Concert deployments. The RTC product is designed for Enterprise application deployment for teams to develop products across geographies. It becomes quite apparent that the availability of the Team Concert server is as critical as the data that it manages. IBM DB2, as other DBMSs, provides a facility to backup the critical data managed by the RTC server without compromising availability (uptime). In fact, by implementing proper processes, normal maintenance can be accomplished without interrupting the functioning of the RTC server. The following sections give an outline of the entire online backup and recovery process for IBM DB2, which can also be used to restore the latest production data available on staging/test servers.

Assumptions

  • The DB2 backend server is running either DB2 V9.1 or DB2 V9.5.
  • The directions are provided for DB2 running on a UNIX or Linux system. The commands on windows system will be identical with the exception of path names.
  • The Administrator is comfortable with UNIX shell environment.
  • Two destinations (directories) on File System are allocated, one for storing backups and one for storing Archive logs.
  • The user has basic knowledge of using DB2 command line on Windows and UNIX machines.

Preparing database for online backups

DB2 databases are created by default with Circular logging.

Logging is a concept in database systems where all modifications are written to transaction logs as they happen when the actual data is modified. In essence, all modifications to the database (insert, delete and update) are stored twice; once in the data files and once, in the transaction logs.

Circular logging is a method where the old transaction logs are overwritten when there is a need to allocate a new transaction log file, thus erasing the sequence of transactions happened to the database.

Archive logging is a method where the transaction logs are overwritten when they are needed, after creating a backup copy of the transaction log. This backup copy of the transaction log is called the ARCHIVE LOG and is numbered sequentially to preserve the transaction sequence.

Setting Archive logging is a prerequisite of taking online backups for a DB2 database. Setting Archive logging for the database requires a directory that is writable for DB2 processes. We recommend creating a directory that has the same ownership as the DB2 instance directory.

When defining a Archive log, you must also create a destination.Here is how to create an Archive log destination:

$ mkdir /backup/ArchiveDest
$ chown db2inst1:db2iadm1 /backup/ArchiveDest

Once the destination is ready, setup the database for Archive logging by updating the database configuration using the following command:

$ db2 update database configuration for JAZZDB using LOGARCHMETH1 'DISK:/backup/ArchiveDest'

NOTE: An immediate offline backup of the database is required before a new connection can be established to the database. When archive logs are generated, they are stored in the directory:

/backup/ArchiveDest/<instanceName>/<databaseName>/<NodeID>/<DBImageCopy>

The disk size consumed by archive logs in the destination directory is directly proportional to the database activity. They continually accumulate to the Archive destination directory and we recommend monitoring the directory and purging the Archive log files that are not required for restoring database, or else pushing them to a tape backup. We recommend making sure all the Archive logs are available on the disk is as recent as the earliest online database backup kept on the disk.

Since the change in the database configuration requires that you take an offline backup before establishing new connections, you have to take an offline database backup before the database can be used by Rational Team Concert. We recommend storing the offline backup to the same target location configured for online backups. This can be accomplished using the following command:

$ db2 backup database JAZZDB to /backup/OnlineBackups

This will be your base backup to which archive logs can be applied to get the database to a consistent state. Though you never need to take any offline backups of the database after this configuration change, it is a best practice to take offline backups before any major configuration changes to environment like software patches, version upgrades etc. An example would be to take an offline backup just before upgrading the RTC Server from Beta3 to Version 1.0.

Taking online backups

Once all the prerequisites of setting the database configuration values are completed, the DB2 database is ready for online backups. Online backups can be taken using the following command:

$ db2 backup database JAZZDB ONLINE to /backup/OnlineBackups COMPRESS INCLUDE LOGS

Please note the two differences between the previous backup command and this one. The keywords are explained below:

  • ONLINE: Instructs the database server to backup the database while transactions are in progress. There is no interruption to any activity while this operation is executing.
  • COMPRESS: This a very nifty feature of DB2 which compresses the backup image on the fly while backing up the database. For those who used to take backups and compress them to save disk space and tape, this comes in really handy.
  • INCLUDE LOGS: This two-word command gives DB2 an instruction to include any archive logs that are required to recover the database to a consistent image. This will be discussed further in the recovery section later in this article.

If you are planning to take nightly backups, it doesn't take long to accumulate numerous backup images and use up your allocated disk space. Even though most installations require only the most recent backup, we recommend keeping the three recent database backup copies on the disk.

The recovery process

No backup is useful unless it can be used to recover the data that was backed up. This section gives an overview of how to recover data from an online backup.

Scenario:

  1. Most recent online backup taken @ 2:00 AM
  2. Most recent Archive log spooled @ 10:15 AM
  3. You want to create a copy of the production database on test system as of 10:00 AM

Recovery steps:

1. Copy the most recent backup to the directory /testsystem/FromProd/ on the test system. To restore the database to test system issue the command:

$ db2 restore database JAZZDB from /testsystem/FromProd

2. If you have more than one copy of the database stored in the directory, you need to specify which backup copy to use for the restore by using the keyword "TAKEN AT". (E. G.: TAKEN AT 20080603155403. Links to DB2 documentation of related commands are provided in Section 7). At this point the database is restored, but you still need to apply the archive logs to make the database consistent. To query the rollforward status of the database issue the command

$ db2 rollforward database JAZZDB  query status

3. Next you need to extract the database archive logs that are bundled into the backup image. To do so, create a temporary directory /testsystem/FromProd/temparchivelogs and extract archive logs using the command:

$ db2 restore database JAZZDB LOGS from /testsystem/FromProd LOGTARGET /testsystem/FromProd/temparchivelogs

Upon completion of this command all the logs required are extracted to the directory provided by the key word LOGTARGET.

4. The final step is to apply all the transactions stored in these log files to the restored database using the following command:

$ db2 "rollforward database JAZZDB to end of logs overflow log path (/testsystem/FromProd/temparchivelogs)"

Upon successful completion of this command the database is consistent to the earliest point (This includes all the changes written to the database while the online backup is in progress). This point will coincide the time of the last committed transaction before the online backup is completed. It can be recovered using this backup. The status should show "DB Working".

If you want to continue restoration by applying more logs, you can copy all the archive logs that are more recent than the ones restored to LOGTARGET and issue the above rollforward command again.

5. Once you have restored the database to the desired state, the rollforward process can be completed by issuing the command:

$ db2 rollforward database test complete

You have successfully restored a DB2 database using an online backup.

Miscellaneous Notes

Quiesce Database: Quiescing database is a feature provided by the DB2 server to put a particular database into isolation mode. While the database is in this state, only Administrators can connect to the database. You do not need to quiesce the database while taking online backups, in fact, for Rational Team Concert to function normally; one should never quiesce the database.

Size and number of LOGs files settings: Log file settings refer to the setting used to specify the number and size of active transaction logs. In DB2, all the transaction log files will have the same size. The configuration of the LOGFILSIZ parameter is represented in 4KB pages, so the value of 1024 for the transaction log size represents 4MB of size. We recommend tuning the size to produce an archive log anytime between 10 to 30 minutes. Greater frequency impacts the database performance and less frequency reduces the granularity of recovery, so it is important to reach a configuration that provides the granularity you desire without negative impact to the database performance.

The number of transaction logs can be configured using the LOGPRIMARY database setting, will determine the size of the largest transaction. A good starting point for the LOGPRIMARY configuration value is 10. The value represented by LOGSECOND is used to allocate additional transaction logs when all PRIMARY LOGS are in use. If you want to have an infinite transaction size set LOGSECOND to a value of "-1".

References

Related topics: Deployment web home, Deployment web home

Edit | Attach | Printable | Raw View | Backlinks: Web, All Webs | History: r5 | r4 < r3 < r2 < r1 | More topic actions...
 
This site is powered by the TWiki collaboration platformCopyright © by IBM and non-IBM contributing authors. All material on this collaboration platform is the property of the contributing authors.
Contributions are governed by our Terms of Use. Please read the following disclaimer.
Dashboards and work items are no longer publicly available, so some links may be invalid. We now provide similar information through other means. Learn more here.