phoenix-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "ASF GitHub Bot (JIRA)" <>
Subject [jira] [Commented] (PHOENIX-3189) HBase/ZooKeeper connection leaks when providing principal/keytab in JDBC url
Date Mon, 29 Aug 2016 22:44:22 GMT


ASF GitHub Bot commented on PHOENIX-3189:

Github user joshelser commented on the issue:
    > This solution is not thread safe and will not allow to safely create multiple instances
of a driver on different threads in the JVM. 
    Yes, that's why I directed you over here, bud. That wasn't an initial goal of these changes.
    > With that said I am not sure that you can support multiple users and support renewals
with the way the UGI works.
    Right.. you're catching on to what I was pointing out. This is something that you should
be managing inside of Storm. We cannot do this effectively inside of Phoenix. We can only
put a bandaid on top.
    > Do we want the Phoenix driver to allow multiple instances instantiated with a different
logged in user for each in the same JVM ?
    The only change I think we can do here is to prevent multiple clients from doing what
you're suggesting and hope they don't shoot themselves in the foot.

> HBase/ZooKeeper connection leaks when providing principal/keytab in JDBC url
> ----------------------------------------------------------------------------
>                 Key: PHOENIX-3189
>                 URL:
>             Project: Phoenix
>          Issue Type: Bug
>    Affects Versions: 4.8.0
>            Reporter: Josh Elser
>            Assignee: Josh Elser
>            Priority: Blocker
>             Fix For: 4.9.0, 4.8.1
> We've been doing some more testing after PHOENIX-3126 and, with the help of [~arpitgupta]
and [~harsha_ch], we've found an issue in a test between Storm and Phoenix.
> Storm was configured to create a JDBC Bolt, specifying the principal and keytab in the
JDBC URL, relying on PhoenixDriver to do the Kerberos login for them. After PHOENIX-3126,
a ZK server blacklisted the host running the bolt, and we observed that there were over 140
active ZK threads in the JVM.
> This results in a subtle change where every time the client tries to get a new Connection,
we end up getting a new UGI instance (because the {{ConnectionQueryServicesImpl#openConnection()}}
always does a new login).
> If users are correctly caching Connections, there isn't an issue (best as I can presently
tell). However, if users rely on the getting the same connection every time (the pre-PHOENIX-3126),
they will saturate their local JVM with connections and crash.

This message was sent by Atlassian JIRA

View raw message