The tuning-config.xml file allows you to modify some of the internal settings in BMC Atrium Orchestrator / TrueSight Orchestration (TSO). The file does not exist by default, so TSO uses its default settings for all tunable variables. You can change these settings by creating the file [TSO_HOME]/config/tuning-config.xml and adding settings as in the example below.
You must restart a TSO peer before it will accept your changes. You must have the same tuning-config.xml file on all peers or the peers may fail to communicate. In some cases this includes TSO Development Studio.
Before you make any changes to the tuning-config.xml file, please contact BMC Customer Support. In many cases, you can modify the following settings within certain limits:
<config>
<job-manager>
<max-active-jobs>256</max-active-jobs>
<minimum-completion-age-for-gc-eligibility>300000</minimum-completion-age-for-gc-eligibility>
<lock-stripe-count>256</lock-stripe-count>
</job-manager>
<health>
<stats-dump-period>300000</stats-dump-period>
</health>
<adapter-manager>
<adapter-call-timeout>3600000</adapter-call-timeout>
</adapter-manager>
<grid-framework>
<network-tolerance-level>LOWEST</network-tolerance-level>
<messaging>
<inbound-message-processor-pool>
<core-size>32</core-size>
<maximum-size>32</maximum-size>
<maximum-backlog>512</maximum-backlog>
<reject-tasks>true</reject-tasks>
</inbound-message-processor-pool>
</messaging>
<logging>
<include-grid-framework>false</include-grid-framework>
<truncate-message-bytes>1024</truncate-message-bytes>
</logging>
</grid-framework>
<foundation>
<disable-modulesync>true</disable-modulesync>
</foundation>
</config>
- max-active-jobs: The maximum number of concurrent active jobs allowed in the system. An active job is defined as any job that is not in a final state, is not in the READY state, and is not on the pending jobs queue. After the maximum number of jobs is reached, any further jobs remain in the pending jobs queue until the number of active jobs is below the maximum number.
- Default: 256
- Minimum safe value: 32
- Maximum safe value: 512
- minimum-completion-age-for-gc-eligibility: The minimum amount of time, in milliseconds, that a job can be in the system before it becomes eligible for removal. After a job has been in a final state for this long, TSO will remove the job from its list the next time it reviews the job list for completed jobs. Decreasing this time allows the grid to remove jobs from memory faster, and increasing it allows completed jobs to remain longer in the list of processes in Grid Manager but increases memory consumption.
- Default: 300000 (five minutes)
- Minimum safe value: 60000 (one minute)
- Maximum safe value: 1800000 (thirty minutes)
- lock-stripe-count: The number of shared resource stripes used for job information synchronization. A shared resource stripe is a stripe decided to processing shared resource requests. A shared resource is a resource that is shared among peers on the grid, like the current jobs list, as opposed to a resource that is unique to a given peer, like a threads list. Reducing this value causes the peers to spend fewer resources processing each other's shared resource requests, which means that shared resource requests have a lower impact on the overall performance of a peer.
- Default: 1024
- Minimum safe value: 256
- Maximum safe value: 2048
- adapter-call-timeout: The amount of time a peer will wait for a response from an adapter request running on another peer, in milliseconds. If the remote peer takes longer than this to respond, the local peer will assume the remote peer is offline and cancel its workflow rather than waiting indefinitely.
- Default: 3600000 (3600 seconds, or one hour)
- Minimum safe value: 300000 (five minutes)
- Maximum safe value: 36000000 (ten hours)
- stats-dump-period: The frequency with which health information is written to the grid.log, in milliseconds.
- Default: 300000 (300 Seconds)
- Minimum safe value: 60000 (60 Seconds)
- Maximum safe value: None
- network-tolerance-level: TSO's sensitivity to network issues. This parameter covers a number of other parameters, including how many updates a peer is allowed to miss before other peers assume it is offline. The lower this value, the faster TSO will acknowledge a peer going offline, but higher values give TSO more leeway to handle a busy grid, network slowdown, or other factors that would make peers communicate more slowly. The default is LOWEST; other valid values are LOW, MEDIUM, and HIGH.
- Valid values: LOWEST (default), LOW, MEDIUM, HIGH, HIGHEST
- inbound-message-processor-pool: Settings that control the pool a peer uses to evaluation incoming requests from external systems and from other peers. A peer on a grid with many other peers or a high load may need a larger pool to process more events simultaneously. A large queue with a small pool minimizes resource utilization but may limit grid throughput.
- core-size: The default number of threads in the pool.
- Default: 32
- Minimum safe value: 32
- Maximum safe value: 128
- maximum-size: The maximum number of threads the pool can use during peak traffic.
- Default: 32
- Minimum safe value: 32
- Maximum safe value: 128
- maximum-backlog: The number of requests the peer will queue waiting for an available thread.
- Default: 512
- Minimum safe value: 256
- Maximum safe value: 1024
- reject-tasks: Whether the peer should reject inbound requests that exceed its queue capacity
- Valid values: true (default), false
- include-grid-framework: Indicates whether certain Grid Framework messages are written to the grid.log file. This includes health tests and peer advertisements. Setting this value to true will tell the peer to write more messages to the grid.log, which can provide detailed health information but on busy or large grids can cause the logs to roll over quickly.
- Valid values: true, false (default)
- truncate-message-bytes: Defines the maximum number of message bytes to be included in a Grid Framework log entry. Many Grid Framework messages are large, and to prevent them from filling the log this parameter caps the length of a given Grid Framework logging line at 1024 bytes. Increasing this value will tell the grid to write more information in each line before truncating it.
- Default: 1024
- Minimum safe value: 64
- Maximum safe value: 10240
- disable-modulesync: Indicates whether a peer synchronizes modules and adapters with the primary Configuration Distribution peer at startup. If a grid has a large amount of modules and adapters, this operation can time out before the peer downloads everything it needs. Setting this to "true" skips this synchronization, which can shorten peer startup times and prevent timeouts, and lets peers synchronize during normal operation.
- Value values: true, false (default)