Cloud Cruiser 3.5 Release Notes

OpenStack Ceilometer enhancements

The following enhancements are made to the OpenStack Ceilometer Collector in this release:

• The collector can now use a data source that requires an SSL connection without you explicitly importing and trusting public certificates.
• You can collect more resource metadata by setting the new includeMeta property to true.
• The feed configuration now has the following date and time defaults, so you no longer need to explicitly set them:

  • <property name="dateTimeZone" value="UTC" />

  • <property name="dateTimePattern" value="yyyy-MM-dd'T'HH:mm:ss" />

  • <property name="datePattern" value="yyyy-MM-dd'T'HH:mm:ss" />

OpenStack AMQP enhancements

The following enhancements are made to the OpenStack AMQP Collector in this release:

• The collector can now parse data in multiple timestamp formats when you collect mixed event types. The new alternateTimestampPatterns property of AMQPConsumerTasklet, enables you to specify a list of event types, each with its own pattern.
  For a usage example of the new property, see the sample job <install_dir>/job_samples/nova_event_listener.xml or the JavaBean Reference documentation at <install_dir>/docs/api/batch/com/cloudcruiser/batch/collect/amqp/AmqpConsumerTasklet.html.
• You can now create a managed AMQP data source instead of maintaining credentials in multiple, unsecure job XML files.

Read JSON lists

The JSON Collector can now map data from lists of both objects and intrinsics using standard JSONPath expressions. For information about this, see Mapping data to Cloud Cruiser.

Filter records when collecting customers

When importing customer records using CustomerCCRecordCollector, you can now use a standard filter property to specify which CC Records should be imported.

ConvertEventsToAllocations

There are two enhancements in this release:

• Added support for audit events. These are control events that report resources observed within some time period. In the case of Nova, it is the audit period which may be configure to hourly or daily. This is different from previous control data that represented a point in time where the resources are known to exist.
• The new boolean properties syncInsertEnabled and syncDeleteEnabled tell the processor whether to trust the control feed over the records being processed. When ​syncInsertEnabled is true, the processor creates a new CC Record when a record exists in the control feed without a corresponding record in the processed set. When syncDeleteEnabled​ is true, the processor deletes a CC Record when no corresponding record exists in the control feed.

CreateIdentifierFromTable

The new property defaultToSourceOnException controls what happens when no match is found and exception processing is off. When true, the value of the source identifier is written to the target identifier, the previous behavior of the processor. When false, the processor does nothing to the record.

RestrictIdentifiers

This new processor drops all identifiers except those specified.

DivertRecordToFile 

This processor now creates an empty output file instead of no output file when no records are diverted.

AdjustEndTime

This new processor subtracts one second from the end time. This is needed when data contains an exclusive end time because we expect inclusive end times.

ConvertToEndEvent

This new processor modifies a record to have only an end time by dropping the start time. If no end time exists, the start time is copied to the end time before being dropped.

ConvertToStartEvent

This new processor modifies a record to have only a start time by dropping the end time. If no start time exists, the end time is copied to the start time before being dropped.

Transform exception files are reset for each run

The exception file for a transform step using CCRecordTransformTasklet is now initialized before record processing begins. This ensures exceptions from a previous run are cleared out if the job is executed again for same date.

Sort uses start time by default

When CCRecordSortTasklet compares timestamps to sort records where all other identifiers match, where both start and end times exist the comparison now uses the start time by default. Previously the end time was the default. You can compare end times by setting the new preferEndTime property to true.

This change was to resolve issues with event correlation because true chronological order is expected there. For information about CCRecordSortTasklet, see the JavaBean Reference documentation at <install_dir>/docs/api/batch/com/cloudcruiser/batch/CCRecordSortTasklet.html.

Aggregation can combine records with distinct resource values

The defaultFunction property of CCRecordAggregationTasklet now accepts a value of DISTINCT. This function combines only records where all resource values match. These values are untouched: aggregation only merges usage periods and mutable identifiers. You can also use DISTINCT when using AggregateResource to specify the aggregation function for an individual resource.

For information about CCRecordAggregationTasklet, see the JavaBean Reference documentation at <install_dir>/docs/api/batch/com/cloudcruiser/batch/CCRecordAggregationTasklet.html.