[
https://issues.apache.org/jira/browse/PHOENIX-3684?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15872395#comment-15872395
]
Hudson commented on PHOENIX-3684:
---------------------------------
SUCCESS: Integrated in Jenkins build Phoenix-master #1557 (See
[https://builds.apache.org/job/Phoenix-master/1557/])
PHOENIX-3684 Handle _HOST in principal specified in JDBC url (elserj: rev
130f29dc5ca012868f4951849ca1af83577c4ccb)
* (edit)
phoenix-core/src/main/java/org/apache/phoenix/jdbc/PhoenixEmbeddedDriver.java
* (edit)
phoenix-core/src/test/java/org/apache/phoenix/jdbc/PhoenixEmbeddedDriverTest.java
* (edit)
phoenix-core/src/test/java/org/apache/phoenix/jdbc/SecureUserConnectionsTest.java
> ConnectionQueryServices connection leak on principal with "_HOST"
> -----------------------------------------------------------------
>
> Key: PHOENIX-3684
> URL: https://issues.apache.org/jira/browse/PHOENIX-3684
> Project: Phoenix
> Issue Type: Bug
> Reporter: Arpit Gupta
> Assignee: Josh Elser
> Priority: Blocker
> Fix For: 4.9.1, 4.10.0
>
> Attachments: PHOENIX-3684.001.patch
>
>
> Through some internal testing, we found that Ambari's use of Phoenix to host
> metrics data was leaking ConnectionQueryServices (and thus HConnections and
> ZK connections), ultimately running into ZK's rate limiting maxClientCnxns.
> After a bit of digging around (and revisiting the old issues around this
> topic PHOENIX-3607, PHOENIX-3611, etc), I finally realized that the logic in
> ConnectionInfo was simply not correctly handling the {{_HOST}} special string
> in the principal (that UGI will replace with the FQDN for the current host).
> This resulted in Phoenix repeatedly re-logging in the user when they created
> a new Connection instead of using the UGI current user, leaking another set
> of connections.
--
This message was sent by Atlassian JIRA
(v6.3.15#6346)
