You can view or change settings for the transaction service.
For example, you can change the location or default file size of the
transaction log files, change transaction timeout properties, or change
heuristic-related properties.
About this task
The transaction service is a server runtime component
that can coordinate updates to multiple resource managers to ensure
atomic updates of data. Transactions are started and ended by applications
or the container in which the applications are deployed.
You
might perform this task when you want to move the transaction logs
to a different storage device, or when you need to change the transaction
service settings. You must restart the application server to make
configuration changes take effect.
- In the administrative console, click . The properties of the application
server, server_name, are displayed in the content
pane.
- Click . The Transaction Service settings
page is displayed.
- Ensure that the Configuration tab is displayed.
Optional: To change
the directory in which transaction logs are written, type the full
path name of the directory in the Transaction log directory field.
You can check the current runtime value of Transaction
log directory, by clicking the Runtime tab.If you do not enter a value for the Transaction
log directory, the application server assumes a default
location in the appropriate profile directory.
When you use WebSphere® Application
Server without high availability support, you do not need to set the
recovery log configuration for persistent services such as the transaction
service. The application server assumes a default location in the
appropriate profile directory. When high availability support is enabled,
this default might not be visible from all servers in the cluster
(for example, if the servers are in different profiles or physical
nodes.) Because of this behavior, configure the recovery log location
for each server in the cluster before enabling high availability.
Ensure that each server in a cluster has a unique transaction log
directory, so that multiple servers do not attempt to access the same
log file. Also, ensure that each server in a cluster can access the
transaction log directories of the other servers in the cluster.
In a high availability (HA)
environment, both the transaction log and the compensation log directory
for each server in a cluster must be unique.
Note: If
you change the transaction log directory, apply the change and restart
the application server as soon as possible, to minimize the risk of
problems occurring before the application server is restarted. For
example, if there is problem and a server fails with in-flight transactions,
when the server restarts, it uses the new log directory and cannot
automatically resolve in-flight transactions that were recorded in
the old log directory.
You can also specify a size for the
transaction logs, as described in the following step.
- Optional: To change the size of transaction
log files, modify the Transaction log directory field
to include a file size setting. Use one of the following formats,
where directory_name is the name of the transaction
log directory and file_size is the disk space allocation
for the transaction log files, specified in kilobytes (nK)
or megabytes (nM). The minimum transaction log
file size that you can specify is 64K. If you specify a value that
is less than 64K, or you do not specify a value for the file size,
the default value of 1M is used.
;file_size <!-- This format keeps the default directory -->
directory_name;file_size
dir://directory_name/directory_name;file_size
/directory_name/directory_name;file_size
![[AIX HP-UX Linux Solaris Windows]](../../dist.gif)
For example, for a Windows system, the following entry
specifies that transaction log files are created in the directory
c:\tranlogs with
a size of 2 megabytes.
c:\tranlogs;2M
For a z/OS system, you might want to reduce the size
of the transaction logs to ensure that they do not exceed the maximum
amount of data space storage is allocated for memory mappings (the
MAXMMAPAREA parameter).
In a non-production environment, you
can turn transaction logging off by entering ;0 in
the Transaction log directory field (do not
enter a directory name). Do not turn transaction logging off in a
production environment because this prevents recovery after a system
failure, and therefore data integrity cannot be guaranteed.
For more information about transaction log sizes,
see Managing transaction logging for optimum server availability.
Optional: Increase the storage
space that is used for memory mapping for transaction log files.
You can modify the MAXMMAPAREA parameter, which specifies the
maximum amount of data space storage, in pages, that can be allocated
for memory mappings of the transaction log files. There are two transaction
log files, which are named log1 and log2, and each file is allocated
1 MB. Therefore, each server needs 512 pages by default. The
following example shows how to calculate the value for the OMVS parameter,
if you use the default size for log files:
MAXMMAPAREA = 512 x number_of_servers + (resources needed outside the application server)
where
number_of_servers is
the number of controllers that are running simultaneously, including
application servers and the deployment manager, but not the node agent.
Optional: Disable MMAP functionality.
Set the com.ibm.ws.recoverylog.spi.NoMemoryMappedFiles property.
Avoid trouble: With this option set, the behavior is the same
as it was before z/OS Version 1.9. You do not need to adjust MAXMMAPAREA.
gotcha
- From the administrative console, select .
- Click .
- Click New. Enter the information
for the com.ibm.ws.recoverylog.spi.NoMemoryMappedFiles property.
Table 1. com.ibm.ws.recoverylog.spi.NoMemoryMappedFiles
property
Name |
Value |
com.ibm.ws.recoverylog.spi.NoMemoryMappedFiles |
true |
- Click OK.
- Save your configuration and synchronize the changes.
- Optional: Review or change the value of transaction
timeout properties:
- Total transaction lifetime timeout
- Type the number of seconds to allow for a transaction that is
started on this server, before the transaction service initiates timeout
completion. If a transaction does not begin completion processing
before this timeout occurs, it is rolled back. A value of 0 (zero)
indicates that this timeout does not apply, and therefore the maximum
transaction timeout is used instead. Application components can override
the total transaction lifetime timeout for their transactions by setting
their own timeout value.
- Maximum transaction timeout
- Type the number of seconds a transaction that is propagated into
this application server can remain inactive before it is ended by
the transaction service. This value also applies to transactions that
are started in this server, if their associated applications do not
set a transaction timeout and the total transaction lifetime timeout
is set to 0.
This value should be equal to or greater than the total
transaction lifetime timeout. A value of 0 (zero) indicates that this
timeout does not apply. In this case, transactions that are affected
by this timeout never time out.
- Client inactivity timeout
- Type the number of seconds after which a client is considered
inactive and the transaction service ends any transactions associated
with that client. A value of 0 (zero) indicates that there is no timeout
limit.
- Optional: Review or change heuristic-related
properties:
- Heuristic retry limit
- The number of times that the application server retries a completion
signal, such as commit or rollback. Retries occur after a transient
exception from a resource manager or remote partner, or if the configured
asynchronous response timeout expires before all Web Services Atomic
Transaction (WS-AT) partners have responded.
- Heuristic retry wait
- The number of seconds that the application server waits before
trying again to send a completion signal, such as commit or rollback,
after a transient exception from a resource manager or remote partner.
- Enable logging for heuristic reporting
- Select this property to enable the application server to log "about
to commit one-phase resource" events from transactions that involve
a one-phase commit resource and two-phase commit resources.
- Heuristic completion direction
- Select the direction used to complete a transaction that has a
heuristic outcome; either the application server commits or rolls
back the transaction, or depends on manual completion by the administrator.
The
heuristic completion direction property specifies how a transaction
is completed in the following situations:
- The transaction manager reports a heuristic outcome for a last
participant support (LPS) resource.
- The heuristic retry limit is exceeded during the recovery of a
subordinate server in a distributed transaction.
- The transaction is imported from a Java EE Connector Architecture
(JCA) provider.
This property applies only to transactions that are in
the situations just described.
- Review or change other configuration properties, to suit
your requirements. For more information about the properties
of the transaction service, see Transaction service settings.
- Click OK and save.
- Stop, then restart, the application server.
If
you change the transaction log directory configuration property to
an incorrect directory name, the application server restarts but cannot
open the transaction logs. Change the configuration property to a
valid directory name, then restart the application server.
If
you are running the application server as non-root, modify the permissions
on the new transaction log location. If you want to use peer recovery
of transactions on a shared device with non-root users, make sure
that your non-root users and groups have matching identification numbers
across machines.