Refiner configuration

Here are the definitions of Refiner’s configuration properties.

Modifying a property

To modify a property, add its new value to the file: <Framework-root>/global_config/overrides/RefinerService/Transformer/etc/refiner.properties (<Framework-root> is the topmost directory of the Deployment Framework.)

The format of a property is a dot-separated property name with an associated value, like this:

property.name1[.name2][.name3]...=value

For example:

batch.time=100
log.file.count=3

Refiner properties

The Refiner configuration properties are:

batch.time

Default: 0

The duration in milliseconds for which Caplin Refiner waits before publishing any changes it has received since the last time it published.

The default value causes Caplin Refiner to immediately publish any new updates as soon as it has finished publishing the previous batch.

If batch.time is negative or not a number, the default value is used.

Run Caplin Refiner initially with the default setting for batch.time (no batching of updates). If, in your system, the rate at which Caplin Refiner has to publish updates causes performance problems with Caplin Refiner itself or for clients, increase batch.time as required to give acceptable performance.

Also see, max.update.size.

filtering.algorithm.filter-name

Default: none

Defines a custom filter comparator.

filter-name is the name of the custom filter comparator.

You can define multiple custom filter comparators; each filtering.algorithm.<filter-name> must have a unique <filter-name>. The property value is the fully qualified class name of the custom filter comparator class.

For example: filtering.algorithm.BondRating=com.example.BondRatingComparator

log.file.count

Default: 1 (< v7.1.4), 5 (>= v7.1.4)

When log.file.count = 0, logging is disabled.

When log.file.count = 1, log messages are written to refiner.log and the following cycling rules apply:

  • If log.file.limit = 0, then refiner.log is not cycled and grows without limit.

  • If log.file.limit > 0, then refiner.log is cycled when it exceeds log.file.limit bytes in size and the archived log is discarded.

When log.file.count > 1, log messages are written to refiner.log.0 and the following cycling rules apply:

  • If log.file.limit = 0, refiner.log.0 is not cycled and grows without limit.

  • If log.file.limit > 0, then the refiner.log.0 and any archived logs are cycled when refiner.log.0 exceeds log.file.limit bytes in size. Log files are cycled by incrementing their filename suffix (refiner.log.n) by 1, beginning with the oldest log file. When all files have been renamed, a new refiner.log.0 file is created. If the number of log files now exceeds log.file.count, then the oldest log file is deleted.

log.file.level

Default: CONFIG

Defines the log level for messages written to the Caplin Refiner log file. The log levels are the standard Java ones:

  • SEVERE (highest level)

  • WARNING

  • INFO

  • CONFIG

  • FINE

  • FINER

  • FINEST (lowest level)

log.file.limit

Default: 0 (< v7.1.4), 268435456 (>= v7.1.4)

The maximum size of a Caplin Refiner log file in bytes.

When log-file-limit = 0, the current log file is not cycled and grows without limit.

When log-file-limit > 0, the current log file is cycled when it exceeds log-file-limit bytes in size. We recommend that this setting is used in conjunction with log-file-count set to 2 or greater.

max.update.size

Default: 1000

As Transformer receives updates to the records in a container, it creates (via Caplin Refiner) individual container record removal and record insertion messages according to how the updates match the filter and sort criteria. If the resulting number of record insertions is high, this can load the Liberator excessively.

The max.update.size property helps to reduce such loading on the Liberator. It defines the maximum number of container record inserts that Caplin Refiner sends in a single batch period (as defined by batch.time). If the number of updates exceeds max.update.size, Transformer sends the changes as a single container image instead.

If max.update.size is negative, 0 or not a number, this feature is turned off and all updates are sent as inserts.

nodata.records.rerequest.timer

Default: 60000

The period in milliseconds between re-requests made by Caplin Refiner for any records that were not previously successfully subscribed to on a supplying DataSource.

If nodata.records.rerequest.timer is negative, 0, or not a number, this feature is turned off.

placeholder.timeout

Default: 1000

How long in milliseconds Caplin Refiner waits for results before removing the placeholder record.

The placeholder record has a special name that allows the client to display an indicator showing that it’s waiting for data. When the placeholder is removed, the client should remove the indicator.

If placeholder.timeout is negative or not a number, the default value is used.

For more about container placeholders, see The container placeholder in How can I…​ Filter and Sort in Refiner.

sorting.algorithm.sort-name

Default: none

Defines a custom sort list or custom sort comparator.

sort-name is the name of the custom sort.

You can define multiple custom sort lists and sort comparators; each sorting.algorithm.<sort-name> must have a unique <sort-name>. For custom sort lists, the property value is a list of sort attributes of the form list: <item1>,<item2>,…​ in ascending sort order.

For example: sorting.algorithm.TenorSort=list:SPOT,ON,TN,SN,1W,2W,1M,2M,6M,1Y

For custom sort comparators, the property value is of the form class: <comparator-class> where <comparator-class> is the fully qualified class name of the custom comparator class.

For example: sorting.algorithm.BondRating=class:com.example.BondRatingComparator.


See also: