You are viewing a plain text version of this content. The canonical link for it is here.
Posted to pr@cassandra.apache.org by iksaif <gi...@git.apache.org> on 2017/09/18 13:09:27 UTC

[GitHub] cassandra pull request #151: [CASSANDRA-13651]: Reduce epoll/timerfd CPU usa...

GitHub user iksaif opened a pull request:

    https://github.com/apache/cassandra/pull/151

    [CASSANDRA-13651]: Reduce epoll/timerfd CPU usage

    - Bump Netty to 4.1.15
    - Add a setting to schedule flushes immediatly  

You can merge this pull request into a Git repository by running:

    $ git pull https://github.com/iksaif/cassandra cassandra-13651-trunk

Alternatively you can review and apply these changes as the patch at:

    https://github.com/apache/cassandra/pull/151.patch

To close this pull request, make a commit to your master/trunk branch
with (at least) the following in the commit message:

    This closes #151
    
----
commit 969dcee971242ff08a2ba1baf9e3139935e85ee8
Author: Corentin Chary <c....@criteo.com>
Date:   2017-09-18T12:12:46Z

    Bump netty to 4.1.15
    
    This is to take advantages of the improvements from https://github.com/netty/netty/pull/7042

commit 594ef9eee5ad853eaec4234f9a11bc12e030ae6c
Author: Corentin Chary <c....@criteo.com>
Date:   2017-09-18T12:26:35Z

    CASSANDRA-13651: Reduce CPU used by epoll_wait() / timerfd_create()
    
    By setting -Dcassandra.native_transport_flush_delay_nanoseconds=0 one can
    schedule the flush() immediatly which will be more efficient when
    using epoll() on Linux by reducing the number of calls to epoll_wait().
    
    This is simmilarly more efficient on version of netty that use timerfd
    to get timeouts with microsecond resolution when calling epoll().
    
    On those platforms this can save up to 10% of CPU.

----


---

---------------------------------------------------------------------
To unsubscribe, e-mail: pr-unsubscribe@cassandra.apache.org
For additional commands, e-mail: pr-help@cassandra.apache.org