MOD_CONTINUE Record Recovery Mode

In MOD_CONTINUE mode, record recovery accepts input of a list of partition offsets or time stamps. After configuration is applied, production pods are reset to the offset or time stamp and files are generated as normal.

MOD_CONTINUE Configuration Properties describes the configuration properties used to specify a range of metadata files by filename.

Table 1. MOD_CONTINUE Configuration Properties
Property Description
sftp.internal.recovery.modContinue.topics A list of topics.
topic.topic Topic name.
topic.partitions A list of partitions in topic.
topic.partitions.partition.partition The partition number of partition.
topic.partitions.partition.offset The partition offset. If no time stamp is specified, this is required.
topic.partitions.partition.timestamp The partition time stamp. If no offset is specified, this is required.

This mode runs as configured in the Helm values file of an otherwise typical ASN.1 sFTP Sink configuration. The following sba-5gc-events.yaml example is configured for MOD_CONTINUE mode:

# Configuration for the ASN.1 sFTP Sink
asn1-sftp-sink:
  # Enabled
  enabled: true
  # The logging level that the ASN.1 sFTP Sink should use (debug, error, warn, info, off)
  logging:
    level: debug
  recovery:
    enabled: true
  configuration:
    configmap:
      asn1-sftp-sink.yaml: |-
        sftp:
          internal:
            recovery:
              modContinue:
                topics:
                  - topic: cdr-route-1
                    partitions:
                        - partition: 0
                          offset: 10
                          timestamp:
                        - partition: 1
                          offset:
                          timestamp: 1658839435466

MOD_CONTINUE mode does the following:

  1. Collects start offsets for each specified topic and partition using the offsetsForTimes function. This is ignored if the offset is missing.
  2. Collects start and end offsets for each topic and partition.
  3. Resets offsets on Kafka using the specified recoveryGroupID.
  4. Records are reprocessed from collected partitions, between offsets using the subscribe function. The records are generated using configured file closure conditions. For more information, see the discussion about file closure control properties and conditions.
  5. Record recovery continues to run until the configuration is changed and re-applied using helm upgrade or similar.