[
https://issues.apache.org/jira/browse/PHOENIX-6796?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17610786#comment-17610786
]
Daniel Wong commented on PHOENIX-6796:
--------------------------------------
For implementation thoughts I had were 3 components. A query logger can be 1
class, a sampler controller (how often and what type of queries), a santizer,
and a publisher. Default might be always sample, publish to system.log and
don't sanitize.
> Pluggable query logger in Phoenix
> ---------------------------------
>
> Key: PHOENIX-6796
> URL: https://issues.apache.org/jira/browse/PHOENIX-6796
> Project: Phoenix
> Issue Type: Improvement
> Reporter: Daniel Wong
> Priority: Minor
>
>
> Requirements
> # Pluggable logger or log publisher
> ## , we don't want to use system.log for multiple reasons from auditing to
> load uncertainty, in addition we need 2 publishing implementations to handle
> future/past deployment patterns. (Think of this as taking the same info we
> would push in system.log but for example using log4j to publish)
> ## We need to sanitize personal information prior to logging for GDPR etc,
> (I have some ideas on how to do this easily using the syntax-tree on
> sanitizing everything by replacing constants with ? or similar)
> # We need some client side performance metrics embedded in the logged
> queries. Most of this is there today I believe but need to audit.
> # We need errors/exceptions with the query logging.
> # Ability to log every ddl/dml/dql.
> # Logging capability across phoenix jdbc/phoenix spark/phoenix MR/phoenix pig
> Some initial discussion on Apache slack
> [thread|https://the-asf.slack.com/archives/CPGHNKLSK/p1620026444027500]
>
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
