Hi Ewen,
It seems Leo has fixed the snappy lib for this issue. Here are changes:
https://github.com/xerial/snappy-java/commit/7b86642f75c280debf3c1983053ea7f8635b48a5
Here is Jar with fix:
https://oss.sonatype.org/content/repositories/snapshots/org/xerial/snappy/snappy-java/1.1.1.4-SNAPSHOT/
Hi Ewen,
Thanks for doing the deep analysis on this issue. I have file this issue
with Snappy project and linked this Kafka Issues. Here is details about
the git hub issue: https://github.com/xerial/snappy-java/issues/88
I will follow-up with snappy guys to figure out how to solve this problem
Also, filed https://issues.apache.org/jira/browse/KAFKA-1721 for this
since it either requires an updated version of the upstream library, a
workaround by us, or at a bare minimum clear documentation of the issue.
On Mon, Oct 20, 2014, at 06:23 PM, Ewen Cheslack-Postava wrote:
> I took a quick loo
I took a quick look at this since I noticed the same issue when testing
your code for the issues you filed. I think the issue is that you're
using all your producers across a thread pool and the snappy library
uses ThreadLocal BufferRecyclers. When new Snappy streams are allocated,
they may be allo
