logging clues? [resend]

Jonathan Lee jonjlee at gmail.com
Mon Jan 25 11:34:59 EST 2010


Ken,

I believe that the underlying problem is that Riak uses one connection per
PUT, but HttpClient assumes persistent connections and pipelines requests
(sends requests without waiting for responses).  My guess is that the
problem happens when HttpClient sends a second request before Riak has a
chance to respond to the first one, and then closes the connection.  A
solution is to send the "Connection: close" header with every request.  In
my test, I no longer encounter errors while PUTting with multiple threads.

            RequestMeta meta = new RequestMeta();
            meta.setHeader("Connection", "close");
            riak.store(o, meta);

Jonathan

On Fri, Jan 22, 2010 at 5:30 PM, Ken Sedgwick <ksedgwic at bonsai.com> wrote:

> On 01/22/2010 07:31 AM, Ryan Tilder wrote:
>
>  What level of concurrency of requests are you using?  The default TCP
>> listen queue in riak is 30, which is likely large enough for most
>> situations.
>>
>
> It happens even when I set NTHREADS to 1.
>
> I'm guessing that riak is doing background writes to disk?  Maybe when I
> get ahead of this writing it results in a RST?
>
> iostat shows that my disk is frequently 100% utilized while this is
> happening ...
>
>
> Ken
>
>
> --
> Ken Sedgwick
> Bonsai Software, Inc.
> http://www.bonsai.com/ken/
> (510) 610-4162
> ken+5a4 at bonsai.com <ken%2B5a4 at bonsai.com>
> Public Key: http://www.bonsai.com/ken/ken.asc
> GPG Fingerprint: 851E 3B07 E586 0843 9434  5CC7 4033 3B9B 3F3F 9640
>
> _______________________________________________
> riak-users mailing list
> riak-users at lists.basho.com
> http://lists.basho.com/mailman/listinfo/riak-users_lists.basho.com
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.basho.com/pipermail/riak-users_lists.basho.com/attachments/20100125/21e6a1b6/attachment.html>


More information about the riak-users mailing list