Fork me on GitHub
#onyx
<
2015-09-19
>
michaeldrogalis00:09:19

@lowl4tency: Gonna need a little more context there to help you.

michaeldrogalis00:09:37

@arohner: It's not core.async specific, simply omit sending the :done message to a source of input. Onyx uses a streaming engine under the covers, and the :done message is a form of punctuation that it leverages to become a batch hybrid.

michaeldrogalis00:09:15

You can implement everything you described above. Next release will include state and windowing primitives to do all of that for you - but we're at least 6 weeks away.

lowl4tency08:09:51

michaeldrogalis: we have 1 clojure-onyx app and 1 zookeeper process. It's all on 1 host. Sometimes clojure app is stuck. I trying to investigate

lowl4tency08:09:00

It's from zookeeper log file

lowl4tency08:09:14

I don't see any errors or overload

nha09:09:18

Hello, I am trying to add onyx into an app. Just adding the dependency causes an error :

#error {
 :cause Could not locate clojure/tools/reader/impl/ExceptionInfo__init.class or clojure/tools/reader/impl/ExceptionInfo.clj on classpath.
`

nha09:09:15

I tried requiring like that :

[org.onyxplatform/onyx "0.7.3" :exclusions [org.slf4j/slf4j-api org.slf4j/slf4j-log4j12 org.slf4j/slf4j-nop]]

nha09:09:56

In the REPL, I get an error when requiring timbre (same version as in onyx project.clj)

nha09:09:43

But again, removing onyx from my dependencies and the error goes away, so I'm not sure what to do.

nha09:09:42

* I mean in the repl, when moving to the file that fails and recompiling, it gives an error on the require timbre :

CompilerException java.lang.NoClassDefFoundError: Could not initialize class taoensso.timbre__init

tcrayford10:09:57

@nha what version of clojure are you on?

nha10:09:13

@tcrayford: I am using 1.7

tcrayford10:09:19

oh, that's odd 😞

nha10:09:47

well that's odd to me too 😕

lucasbradstreet10:09:17

What version of Java? (probably not it, but it's a weird issue and I don't know what else to suggest)

nha10:09:48

REPL-y 0.3.5, nREPL 0.2.10
Clojure 1.7.0
Java HotSpot(TM) 64-Bit Server VM 1.8.0_45-b14
`

nha10:09:54

(on osx)

nha11:09:36

I use pedestal if it matters

nha11:09:39

I tried onyx 0.7.2 too

lucasbradstreet11:09:51

Yeah, it could be dependency resolution is forcing a lower dependency of something. lein deps :tree could help

nha11:09:35

well there is that :

[org.onyxplatform/onyx "0.7.3" :exclusions [org.slf4j/slf4j-api org.slf4j/slf4j-log4j12 org.slf4j/slf4j-nop]] -> [org.apache.zookeeper/zookeeper "3.4.1" :exclusions [org.slf4j/slf4j-log4j12 io.netty/netty]] -> [jline "0.9.94"] -> [junit "3.8.1"]
 overrides
[siren "0.2.0"] -> [domina "1.0.0"] -> [org.clojure/clojurescript "0.0-1424"] -> [com.google.javascript/closure-compiler "r1918"] -> [junit "4.8.2"]

nha11:09:25

oh wait there is more

nha11:09:10

it's big - sorry about that - but there are messages related to onyx

lucasbradstreet11:09:15

Sorry, getting a flight so I won't be able to help out for a bit

nha11:09:32

Sure no problem. Have a nice flight simple_smile

michaeldrogalis14:09:04

@lowl4tency: I need more context is terms of "what was going on at the time?", what in particular do you mean by "stuck"? Those error messages indicate that your ZooKeeper lost data. It simply tried to read a value from a znode that apparently doesn't exist anymore. Is 4068bfc4-7049-4a9d-a4f3-0dc5dcdb3aa7 the Onyx deployment ID you're using, as indicated by that error message?

michaeldrogalis14:09:47

Specifically, the two values in that error message, the values for job-scheduler and messaging, are values that are required to be known before the Onyx dashboard can start, as they are set by peers which influence the way log entries are applied

tcrayford16:09:18

@michaeldrogalis: those error messages are from creates, not gets - they're trying to create a node that already exists (which is an error in zk)

michaeldrogalis17:09:29

@tcrayford: Ah! Thanks for the correction. "NodeExists" and "NoNodeExists" often play a trick on my eyes. simple_smile