Re: Reasons to think of MTU size and avg value size?

2012-08-20 Thread Patrik Sundberg
Thanks for reply. Just had the thought that if all my values are say 5k and my MTU is 1500 that'd seem possibly unnecessary mismatch. Just preliminary brainstorming from me, no well formed use case or testing yet, just a question that popped in my head. On Mon, Aug 20, 2012 at 6:10 PM, Sean Carey

Re: Reasons to think of MTU size and avg value size?

2012-08-20 Thread Sean Carey
Hey Patrick, You could try jumbo frames and it will buy you some extra CPU. We have done this in a couple of configurations and didn't notice anything crazy. Are you thinking increasing MTU might benefit in some other way, other than CPU. I'd be interested in knowing your thoughts/results if

Re: Riak Search. Ranges. Integer Analyzer Factory.

2012-08-20 Thread Ryan Zezeski
On Mon, Aug 20, 2012 at 12:29 PM, Vladimir Shapovalov wrote: > Hi Ryan, > > Thank you, I didn't know that the data stays unchanged. Similar to 2i. > > While executing curl request I get this record in crash.log: > -- > 2012-08-20 18:1

Re: Riak Search. Ranges. Integer Analyzer Factory.

2012-08-20 Thread Vladimir Shapovalov
Hi Ryan, Thank you, I didn't know that the data stays unchanged. Similar to 2i. While executing curl request I get this record in crash.log: -- 2012-08-20 18:12:22 =CRASH REPORT crasher: initial call: mochiweb_acceptor:init

Re: Riak Search. Indexing special chars. Internal Server Error.

2012-08-20 Thread Ryan Zezeski
On Mon, Aug 20, 2012 at 10:43 AM, Vladimir Shapovalov wrote: > > > It looks like the whitespace_analyzer_factory has > some differences compare to standard_analyzer_factory. > > I am able to index a word containing S-sharp with either analyzer. ___ riak-u

Re: Search is returning 0 results

2012-08-20 Thread Ryan Zezeski
On Fri, Aug 17, 2012 at 7:46 PM, bufer wrote: > > > ### get an specific item data > $ curl -i > http://127.0.0.1:8098/buckets/options/keys/Z8Q1FcBqpGb9fp97CRvia1TMbJq > HTTP/1.1 200 OK > X-Riak-Vclock: a85hYGBgzGDKBVIcypz/fgbo5gVmMCUy5rEyGEouOsmXBQA= > Vary: Accept-Encoding > Server: M

Re: Riak Search. Indexing special chars. Internal Server Error.

2012-08-20 Thread Vladimir Shapovalov
Hi Ryan, I'm using ver. 1.2.0 too. It happens only on bucket with indexing enabled. I think it depends on the Analyzer factory. My index schema was: ... {dynamic_field, [ {name, "*"}, {type, string}, {analyzer_factory, {erlang, text_analyzers, * whites

Re: Riak Search. Ranges. Integer Analyzer Factory.

2012-08-20 Thread Ryan Zezeski
On Mon, Aug 20, 2012 at 4:22 AM, Vladimir Shapovalov wrote: > Hi Alexander, > > thanks for prompt reply. > Didn't work... > And the strings aren't padded in my case. I still have 100 or 200 as > values after insert. > > Thanks > Vladimir > > Vladimir, Alexander is correct. The integers are turne

Re: Riak Search. Indexing special chars. Internal Server Error.

2012-08-20 Thread Ryan Zezeski
On Mon, Aug 20, 2012 at 6:24 AM, Vladimir Shapovalov wrote: > Hi all, > > I'm trying to put some data in a bucket, witch has indexing enabled. > One key value has a special char ß (geman > S-sharp). > It is UTF-8 encoded. > > The errror message i get: > >

Riak Search. Indexing special chars. Internal Server Error.

2012-08-20 Thread Vladimir Shapovalov
Hi all, I'm trying to put some data in a bucket, witch has indexing enabled. One key value has a special char ß (geman S-sharp). It is UTF-8 encoded. The errror message i get:

Reasons to think of MTU size and avg value size?

2012-08-20 Thread Patrik Sundberg
Hi, I was just pondering a bit here - any pointers to think about in terms of MTU settings for a network with a riak cluster and clients? (let's ignore any other traffic on the network) It'd seem a good idea for MTU > median/mean value size (when we're talking about value sizes that aren't huge, k

Re: Riak Search. Ranges. Integer Analyzer Factory.

2012-08-20 Thread Vladimir Shapovalov
Hi Alexander, thanks for prompt reply. Didn't work... And the strings aren't padded in my case. I still have 100 or 200 as values after insert. Thanks Vladimir On Mon, Aug 20, 2012 at 1:03 AM, Alexander Sicular wrote: > Ints are stored as 10 char padded strings. So int:11 is "11". So >