Versions Compared

Key

  • This line was added.
  • This line was removed.
  • Formatting was changed.

Agenda

  • In the process of promoting Kevin Smokowski as one of committers for the logging project, voting ends on  6pm EST
  • El Alto was officially released on  
  • The future of logging project
  • Frankfurt status
    • M1 is targeted for  , logging is currently behind  
    • update to version 3.x of Python
    • update to Java 11 Dave Williamsonto reach out Kevin Smokowski to evaluate the effort required
  • Draft Frankfurt spec ONAP Application Logging Specification v1.3 (Frankfurt) - Continuation
    • Is there any components still write log to /var/log/ONAP_EELF? Luke ParkerAny updates?
    • We owe Steve Smokowski an answer for his question on August 30, 2018 (Response Code and Response Description) and September 4, 2018 (Invoke and Invoke-return).
      - Sanjay and Dave Williamson to talk with the AT&T Ops team for the use of Response Code and Response description.
      Prudence Auto reply to Steve on invoke and invoke-return and add a JIRA ( LOG-1174 - Update reference library to include a sample invoke-return marker OPEN )to update the sample implementation to include invoke-return marker 
    • What is needed for Michael's TODO on November 21, 2018 regarding X-FromAppId?
      Prudence Au to include X-FromAppId as one of the option for PartnerName
    • Still need to address the issue with 2 test environments and only 1 log instance?  Dave Williamson reported it on December 6, 2018
      Sanjay and Dave Williamsonto investigate.  This might be resolved by the tenant isolation feature.
    • The description for p_message should be updated to "standard attribute - defined in logback.xml - Message - used for %msg%" as suggested by Lorraine A. Welch on  July 3, 2019?
      Agreed to update the description as described by Lorraine.
    • Removing "VirtualServerName" as mentioned by Dave Williamsonon July 23, 2019? 
      Agreed
    • Still need a description for ServiceInstanceID, any taker?
      Suggestion: An unique identifier of a newly instantiated network service instance.
      ----------------------------------------------------------------------------------------
    • Any feedback on Kevin Smokowski's comment on August 6, 2019 regarding the use of User-Agent as PartnerName?
    • Any feedback on Kevin Smokowski's comment on August 5, 2019 regarding the value of ServerIPAddress in the case of docker & kubernetes deployment?
    • Any updates needed for TargetEntity and PartnerName?  Mentioned on August 15, 2019 (a conversation between Dave and Kevin)
    • Any feedback on composite RequestID or NOT composite RequestID (from October 2, 2019)?

Attendees

Notes

...

Compliance Report

Where:

  • Files whose path suggest they're for testing are skipped. 
  • XML is sanitized first: 
    • To remove XML comments. 
    • To fix XML that has bad data ahead of <?xml, including a lot of copyright headers. (Java may fare better than Python, but they're technically invalid).
  • A few documents (properties and XML both) are unparseable for other reasons, so they're reported as such in Notes.
  • Analysis is pattern-based and fairly superficial: 
    • There are 300+ files. 
    • Even within EELF, there's a lot of variation. 
    • For Logback, EELF is identified by characteristic MDC keys, and compliant logging by telltale MDC serialization, markers and escaping (all of which seem to be non-existent in non-compliant logging). 
    • For Log4j, likewise, but slightly more forgiving when sniffing for EELF, and there are no instances of compliant logging for Log4J, so nothing to look for. 
  • Summarizing: 
    • Log provider configuration is in shambles. 
    • Ad hoc configurations predominate. 
    • EELF remains ubiquitous. 
    • Log4j, too, and note the number of Log4j properties files (→ Log4j v1, which does not support escaping, and therefore cannot emit logs that can be reliably indexed). 
    • Compliant logging:
      • Is rare in component configuration (i.e. inside projects).
      • Is slightly more common in OOM overrides (which generally take precedence where they exist). 

...