Hot Upgrade from Release 6.3, 6.4, or 6.5 to Release 7.1


	Corporate Info \| News \| Solutions \| Products \| Partners \| Services \| Events \| Download \| How To Buy
	http://www.oracle.com/technology/documentation/index.html \| Site Map \| Search \| PDF Files \| Contact \| Glossary

Tuxedo Doc Home \| Installation \| Topic List \| Previous \| Next \| Contents

Installing the BEA Tuxedo System

During the upgrade, you must shut down the backup and master machines serially; the remainder of the configuration is unaffected.

Preliminary Notes

This section is for customers running Release 6.3, 6.4, or 6.5.

Once you have finished migrating the master to Release 7.1 (via a hot upgrade), you will not be able to migrate it back to the acting backup until you have upgraded the backup to Release 7.1. In other words, migrating the master is an irreversible procedure; once you have completed it, you cannot change the master back to the release of the BEA Tuxedo system on which it was running previously.

When you run tmadmin(1) on a Release 7.1 machine with a Release 6.3, 6.4, or 6.5 master, the UPGRADE file (as shown in Example) must reside in the current directory. To make sure that the files are in the current directory, you might, for example, run tmadmin from APPDIR.

During migration, keep in mind the following guidelines regarding the BEA Tuxedo data storage areas:

You will be instructed to remove the old TUXCONFIG file. The TUXCONFIG file is propagated automatically when the backup machine is run; therefore, you do not need to do anything.
Create the TLOG when you load the software. If you already have a TLOG, remove it.
Do not do anything with application queue spaces and queues.
If you have a BDMCONFIG file, leave it as is.

To minimize down time, we recommend installing Release 7.1 on the master and backup, and rebuilding all clients and servers in the directory on each machine that is defined by APPDIR, in parallel with the directory in which the Release 6.3, 6.4, or 6.5 clients and servers were created. (In other words, both a Release 6.3, 6.4, or 6.5 application and a Release 7.1 application should be available on each machine.) This is not a requirement, but a recommendation.

When migrating from Release 6.3, 6.4, or 6.5, you must rebuild your clients and servers. Additionally, you must always rebuild if you are running a BEA Tuxedo application on an HP platform. The HP-UX shared library loader does not, by default, use the paths specified in the SHLIB_PATH environment variable when trying to locate the shared libraries needed by an executable at run time. The HP-UX shared library loader must be turned on. (This procedure is not necessary on any other platform supported by the BEA Tuxedo system.) As a result, once you have upgraded from Release 6.3, 6.4, or 6.5 to Release7.1, your application programs will not be able to find the Release 7.1 BEA Tuxedo libraries. These programs must be recompiled with the following setting.

CFLAGS=Wl,+s

This setting causes the compiler to pass the +s option to the link editor when the application is recompiled. It changes the program's default behavior during future upgrades.

Note: The buildserver and buildclient commands include the +s option of the link editor; you do not need to specify +s explicitly. If you are using a different build procedure, however, you must specify the +s option.

Procedure

Make sure your environment is set up as follows:
- The configuration must have a backup machine.
- The MODEL option must be set to MP.
- The MIGRATE option must be set.
Shut down the backup machine.
Reboot the backup machine, using Release 7.1.
Migrate the master running Release 6.3, 6.4 or 6.5 to the backup running Release 7.1.
Shut down the machine that is now acting as the backup.
Using Release 7.1, reboot the machine that you shut down in Step 5.

Example

Suppose you have the following on a UNIX platform:

A machine called MACH1 that is configured as the master and that runs Release 6.3, 6.4 or 6.5
- A corresponding application directory called $APPDIR1
- A corresponding configuration file called $TUXCONFIG1
A machine called MACH2 that is configured as the backup
- A corresponding application directory called $APPDIR2
- A corresponding configuration file called $TUXCONFIG2

Refer to the following instruction sequence to perform a hot upgrade for your application.

MACH1> tmshutdown -B MACH2 -l MACH2
 
MACH1> Run tmconfig to reset the APPDIR, TUXCONFIG, TUXDIR for MACH2
 
       MACH2> Kill tlisten process
 
       MACH2> Load System 7.1 if it is not already loaded
 
       MACH2> cd $APPDIR2
 
       MACH2> rm $TUXCONFIG2
 
       MACH2> >UPGRADE    # Indicator that upgrade being done
 
       MACH2> Reset PATH, LD_LIBRARY_PATH, TUXCONFIG, APPDIR
 
       MACH2> Rebuild clients and servers, create TLOG
 
       MACH2> Start System 7.1 tlisten
 
MACH1> >UPGRADE    # Indicator that upgrade being done
 
MACH1> tmboot -B MACH2 -l MACH2
 
       MACH2> tmadmin
 
              master
 
              y
 
              psr
 
              q
 
       MACH2> tmshutdown -B MACH1 -l MACH1
 
       MACH2> Run tmconfig to reset the APPDIR, TUXCONFIG, TUXDIR for MACH1
 
MACH1> Kill tlisten process
 
MACH1> Load System 7.1 if it is not already loaded
 
MACH1> cd $APPDIR1
 
MACH1> rm $TUXCONFIG1
 
MACH1> Reset PATH, LD_LIBRARY_PATH, TUXCONFIG, APPDIR
 
MACH1> Rebuild clients and servers, create TLOG
 
MACH1> Start System 7.1 tlisten
 
MACH1> rm UPGRADE  # Remove indicator that upgrade being done
 
       MACH2> rm UPGRADE  # Remove indicator that upgrade being done
 
       MACH2> tmboot -B MACH1 -l MACH1

Note: If you are performing the upgrade illustrated in the previous example on a Windows NT platform, be sure to stop the BEA Tuxedo IPC Helper Service after killing the tlisten process and to start the BEA Tuxedo IPC Helper Service before starting the tlisten process.