Hi Janus, Performance will be the main concern. In busy HDFS clusters admins are likely to use dfs.namenode.audit.log.debug.cmdlist.
Have you enabled read audit logging in your Ozone cluster? What is the number of requests/second? Thanks, Arpit > On Apr 3, 2022, at 7:58 PM, Janus Chow <yiyang0...@gmail.com> wrote: > > Hi Ozone dev, > > When checking the audit logs from Ozone components, we found that by > default Ozone only logs WRITE operations. In order to enable the audit log > for READ operations, we need to change the configurations in > audit-log4j2.properties. > That brings some confusion for users when comparing it to some other > storage systems, like HDFS, in which audit logs are enabled for both READ > and WRITE by default. > > We have a Jira ticket(https://issues.apache.org/jira/browse/HDDS-6532) and > PR (https://github.com/apache/ozone/pull/3255) about adding audit logs for > READ operations by default. > Could you help to check and comment if there are any specific concerns > not to enable READ audit logs? > > Yiyang > Thank you very much. --------------------------------------------------------------------- To unsubscribe, e-mail: dev-unsubscr...@ozone.apache.org For additional commands, e-mail: dev-h...@ozone.apache.org
