TP+ configuration¶

This page lists the ems.yaml and systemParams.yaml fields used to configure TP+ behavior.

ems.yaml¶

The following fields from ems.yaml are used by TP+. Other fields are used by RT only.

Attribute	Definition	Example
`name`	Feed name	`feed1`
`taskset`	If supplied, the orchestrator applies the CPU taskset options to the TP+ tasks, by default it inherits RTO's taskset.	`0-7`
`external`	Flags the streams exposed to external publishers or subscribers so that bridging helpers and discovery registration only apply when needed.	`false`
`archiving`	Dictates how long TP+ logs are retained (`retentionDuration`), how large the merged TP log directory can grow (`size`), and what maximum disk usage percentage per volume is tolerated (`diskUsage`). RTO enforces all three heuristics.	`archiving:` `size: 20Gi` `retentionDuration: 18000` `diskUsage: 90`
`tp`	Dictates stream-specific TP+ parameters.	`tp:` `backup: true` `checkpointFreq: 10000` `flushFreq: 1000` `bufferSize: 1000` `bufferPurgeRatio: 0.2`

The following fields from systemParams.yaml are used by TP+.

Attribute	Definition	Example
`tpLogDir`	Root directory where TP+ stores its log files. Mirrors the RT log directory structure so data is compatible.	`${KXS_ROOT}/tpp`
`tpConnTO`	Timeout in milliseconds for `.tpsub` when dialing TP+ endpoints.	`1000`
`tpPubAck`	Whether `.tppub` waits for TP+ to confirm disk flush before clearing its buffer.	`true`
`tpPubBufferSize`	Depth of the publisher-facing circular buffer before sequence overflow is triggered.	`100000`
`rtoChkDiskFreq`	How often the orchestrator checks disk usage (in seconds). When non-null it is the primary watchdog for disk saturation.	`10`
`rtoDiskBufPct`	Safety buffer percentage of disk usage below configured archiving thresholds; exceeding this causes TP+/RT to halt for intervention.	`5`