Github user karanmehta93 commented on the issue: https://github.com/apache/zookeeper/pull/307 Agreed @zodvik . The scope of this Jira is to limit it to a generic slow query log and identify the outliers if any. For that reason, I am printing the actual request to the log as well. An improvement would be definitely to separate based on request types as others have also suggested. If this gets in, I would like to work on the improvement part of using t-digest(https://github.com/tdunning/t-digest) to give out detailed insight and not have the problem of setting a fixed threshold.
---