00:24
<usiegj00>
I'm doing a map reduce on Riak (my first on our production data)
00:24
<usiegj00>
and it's timing out on a 20k entry bucket
00:24
<usiegj00>
it's building a histogram of a small subset of the data
00:24
<usiegj00>
using very specific keyfilters
00:24
<usiegj00>
I'm assuming I'm not doing this right
00:25
<usiegj00>
Is key filter NOT a preferred method for picking a set of inputs?
00:25
<usiegj00>
and if not, what is?
01:13
<drdo>
Is there something in riak-erlang-client that can load balance between a bunch of nodes?
02:34
<usiegj00>
so what're the odds someone has hit a timeout running mapreduce
02:34
<usiegj00>
and worked around it
02:35
<usiegj00>
it's in a key filter range...
09:49
<Robe>
xzilla: you hang out with all the cool kids? :>
10:38
<hukl>
Mirell: had a question but not anymore :) sorry
11:17
<Mirell>
hukl: Heh, just looked at IRC
12:20
pharkmillups joined
12:23
<pharkmillups>
morning
12:56
carlosbrando joined
12:57
<carlosbrando>
Hi. There is a way to turn off the mapreduce cache using the new mapreduce system?
12:57
<carlosbrando>
I'm using the old one and I can disable the cache. But I did'nt find a way to do the same with the new one.
13:00
<carlosbrando>
I need it because I have to log when someone raises a map reduce. I have POS terminais accessing my Riak, and I charging per mapred query.
13:14
<jodaro>
randomly met andy gross on the corner a few minutes ago
13:16
<pharkmillups>
jodaro: haha. oh?
13:16
<pharkmillups>
how did that happen?
13:16
<jodaro>
his ipad (i think?) has a big basho sticker on it
13:17
<jodaro>
i glanced over and saw that and was like "oh, hey, i'm a riak newb. do you work there?"
13:17
<Robe>
we've got quite a few of those around our office here too
13:17
<Vagabond>
a wild argv0 appears
13:17
carlosbrando joined
13:17
<pharkmillups>
jodaro: tell him he's late for a call. :)
13:18
<Robe>
and I'm late for gtfoffice
13:18
<Mirell>
pharkmillups: He's a busy man today
13:19
<jodaro>
he seemed rush but he was cool enough to talk to me for a second
13:19
<jodaro>
so don't give him too much shit
13:29
<jodaro>
seemed rushed
13:29
<jodaro>
he wasn't listening to Rush as far as i could tell
14:17
<j2043>
After deleting 76 million keys from a bit cask, should the memory footprint go down, or does bit cask still have a record of those, but they are marked dead.
14:18
<j2043>
from a riak cluster back ended by bitcask
14:22
<Vagabond>
you have to wait for a compaction/merge to happen
14:22
<j2043>
How often does that happen?
14:23
<j2043>
and is there any way to force it
14:24
<Vagabond>
not sure
14:24
<Vagabond>
I'm sure one of the DAs would know
14:24
<pharkmillups>
j2043: you can control some aspects of the merging
14:24
<pharkmillups>
hang on …
14:27
<pharkmillups>
j2043: http://wiki.basho.com/Bitcask.html#Tuning-Bitcask
14:28
<pharkmillups>
http://wiki.basho.com/Bitcask.html#Configuring-Bitcask should also help
14:30
<dreid>
Do expired keys contribute to the dead keys used to figure out when you should merge yet?
14:30
carlosbrando joined
14:37
<seancribbs>
dreid: in 1.0+ they should, yes
14:55
<j2043>
I seem to have a node half in, half out of a cluster due to a error on my part. Is it possible to force a node to rejoin a ring?
15:04
<seancribbs>
j2043: after it finishes its current transition, yes
15:07
daniel_karlsson joined
15:27
<chip->
Anybody have experience with Redis as a write-through / read-through cache for Riak?
15:27
<pharkmillups>
chip-: ping siculars next time he shows up
15:28
<pharkmillups>
I think there has been some chatter on the ML about it, too.
15:28
<chip->
Cool, thanks. Hoping to find someone who's used Riak hooks / Redis pub/sub to accomplish it without too much custom code.
15:28
<pharkmillups>
might be worth searching risk.markmail.org
15:28
<pharkmillups>
s/risk/riak
15:28
<chip->
I'm so happy that autocorrect happens to you guys too
15:28
<j2043>
who would have thought that would be so annoying
15:29
<j2043>
I find its hard to convince google that I don't want to search for risk
15:29
<j2043>
even doing +riak no longer works
15:29
<chip->
j2043: I search on basho "riak"
15:29
<chip->
er, 'basho "riak"'
15:29
<seancribbs>
worst is when I try to text my wife with "np" and it corrects to "No"
15:30
<* seancribbs>
facepalm
15:30
<seancribbs>
pharkmillups: where can i get this Riak Credit? sounds great
15:32
<Vagabond>
good for 1 free ring join
16:09
<drdo>
http://pastebin.com/8LFuwpxC
16:10
<Vagabond>
why are there random Fs in that paste?
16:16
<PSUdaemon>
i updated a pre-commit hook and i want those changes to take effect. i recall a method where i connect to the riak console and issue a command that reloads modules. does anyone know what i am referring to? posisbly have a link to some info?
16:16
<hoodoos>
hello guys
16:17
<hoodoos>
anyone's hosting production riak in amazon?
16:17
<hoodoos>
pharkmillups: btw, we finally got a piece of english version running ;)
16:18
<pharkmillups>
hoodoos: nice!
16:18
<pharkmillups>
link>
16:18
<drdo>
Vagabond: oops, was half asleep :P
16:19
<drdo>
Vagabond: Any idea why that would happen? Isn't content-type working in riak-erlang-client?
16:23
<hoodoos>
pharkmillups: you got it in private?
16:23
<pharkmillups>
I did. thanks
16:23
<pharkmillups>
looking now
16:24
<pharkmillups>
hoodoos: re - amazon: yes. people run Riak in prod on AWS.
16:25
<hoodoos>
we're planning to move from linode to amazon
16:25
<omarkj>
I'm working on moving to the current version (1.1.0) of riak_core from 0.14.x. After going through the API changes I've found I'm getting an interesting behaviour when joining nodes with different servies together; https://gist.github.com/5da09dbb9582594f87e9, active owners on the node that joins seem to go away. Pretty sure I' just overlooking something trivial. Any ideas?
16:42
randysecrist joined
17:56
<omarkj>
Never mind, I was misunderstanding a few things.
21:50
<Judson>
Is there a way to tell what backend a riak server is using?
21:51
<Judson>
As a client, I mean. For instance, to determine if secondary indexes are available.