Skip to content

DAVE-6.4.5 release notes

This document is a summary of content, changes, limitations and fixes of this release as compared to DAVE-6.4.4 release.

Refer to the the Punchplatform official site if you look for the complete documentations and resources.

The documentation for this release can be found inside the deployment archives (standalone and deployer versions), and at https://doc.punchplatform.com/6.4.5/index.html.

Note about upgrades: Please carefully review the documentation for this release before upgrading your clusters. The upgrade notes (e.g. upgrade from 6.3 to 6.4) discuss every critical information about compatibility issues, breaking changes, performance changes, and any other configuration changes that might impact your Punchplatform.

Changes

Mutual Lumberjack and Syslog TLS authentication

The Lumberjack and Syslog output nodes can be configured with mutual exclusions. A bug prevented however the optional private key set on the client side, resulting in a handshake failure. This has been corrected (#1905).

Punch light runtime

Topologies can be configured to automatically slow down if the tuple traffic completely fails. This typically happens on a remote log collector that cannot reach its destination(s) for some time. By slowing down the internal input_node polling rate, the CPU usage quickly fall. input_node activity will resume quickly as soon as some traffic is again acknowledged.

To activate this behavior use the following topology level settings:

settings:
  topology.enable.slowfails: true

This feature is not available in Storm runtime and the COTS does not offer such feature. If you need to slow down a topology in Storm, see next change on kafka_input node.

Kafka input node

The kafka_input node has the new option sleep in fail_action parameter. That makes the node sleep whenever it encounters a failure. The sleep time can be adjusted with fail_sleep_ms (50ms by default). When the tuple traffic completely fails, this has the same effect as the topology.enable.slowfails. You can use this parameter if you want to slow down a topology with kafka_input in Storm runtime.

dag:
  - type: kafka_input
    settings:
      fail_action: sleep
      fail_sleep_ms: 10

With this configuration, the kafka_input will sleep for 10ms when it encounters a tuple failure.

More details in Kafka Input Node documentation

Breaking changes

Refer to the Migration Guide.

Improvements

  • #1875 change patch message from warn to info in patched storm punchlines
  • #1880 improve the observability logs and events
  • #1880 added slow rate in case of failures mechanism
  • #1895 add schema support for elastic input node
  • #1901 add option to slow down kafka_input on tuple failure
  • #1903 manage date fields for elastic output spark node
  • #1907 add parameter to limit storm supervisor cache size
  • #0000 doc expectation setting in generator input
  • #0001 change misleading error log when failing tuple in file input node
  • #0001 improve parser development documentation
  • #0000 adding punch templating documentation

Bug Fixes

  • #1518 compression option in cmd line not working for lumberjack log injector
  • #1851 fixed the out of sync and submitted state that last forever in channelctl
  • #1867 high cpu usage when output bolt completely fail
  • #1868 remove supervisord refs in doc
  • #1878 update ssl certificates on standalone
  • #1879 configure command cannot generate additional files
  • #1884 missing ack and wrong ip calculation in filter node
  • #1887 double quote message makes elastic output fail
  • #1888 allow documentation plugin to deploy on the selected instances
  • #1889 remove unwanted warning messages on kibana
  • #1890 update modsecurity rules to restrict index resolving when creating index patterns
  • #1891 path modification to right documentation folder
  • #1892 case sensitive parameter not working in filter node
  • #1893 archive housekeeping delete all archives instead of deleting only archives older than the specified date
  • #1894 planctl cannot run manually on a shiva server
  • #1896 Elasticsearch writes its logs to multiple location
  • #1902 Wrong elasticsearch address when deploying metricbeat
  • #1904 manage ilm indices for elastic housekeeping
  • #1905 take into account the documented ssl_private_key settings on the client side.
  • #1906 misleading error message when a punchlet is not found.
  • #1908 shiva out of memory because of reporter library leak
  • #1909 misleading error message when using wildcards in channelctl status and stop commands
  • #0001 ssl options in cmd line not working for lumberjack log injector

Security Fixes

Internal tasks

  • #1866 automate release with cicd
  • #0000 remove punch plugin references in doc
  • #0000 add dedicated page for punchlang javadoc

Limitations

Known bugs