Hi Pavel,
Thanks for trying to run the benchmark. The error arises because an Apache
Ignite node started with the default config does not have the cache with
name '*atomic*'. To run the benchmark you should do the following:
1. Checkout latest '*master*' branch from Apache Ignite repo [1].
2. Compile, install and assembly the Apache Ignite build as described in
DEVNOTES.txt, section "Ignite Fabric Maven Build Instructions", clauses 2
and 4.
3. Unzip the assembly file '*target/bin/apache-ignite-fabric-*-bin.zip*'.
4. Cd to target folder, open '*examples/config/example-cache.xml*' with
an editor, find the '*org.apache.ignite.configuration.CacheConfiguration*'
section and rename '*default*' cache to '*atomic*'.
5. Start local Apache Ignite node with this config: '*./bin/ignite.sh
examples/config/example-cache.xml*'. Either start one more local node
the same way or set one endpoint in *CachePut.js* (via *ENDPOINT*
constant).
6. Start the benchmark: '*node bench-starter.js*'.
Please let me know if you experience any other problems.
Thanks,
Ivan
[1] https://github.com/apache/ignite/
вс, 10 июн. 2018 г. в 23:17, Pavel Petroshenko :
> Hi Ivan,
>
> I've got through some build issues, but it's still crashing on master [1]
> for me:
>
> Thread 1. Client is started
> Thread 1. ERROR: Cache does not exist [cacheId= -1407396309]
> Thread 1. ERROR: Cache does not exist [cacheId= -1407396309]
> Thread 1. ERROR: Cache does not exist [cacheId= -1407396309]
> Thread 1. ERROR: Cache does not exist [cacheId= -1407396309]
>
> It would be nice if you could provide some sufficient instructions on how
> to "install" and run the scripts. And also please specify, which AI
> codebase you used for testing.
>
> Thanks,
> p.
>
> [1]
> https://github.com/nobitlost/ignite/tree/master/modules/platforms/nodejs
>
> On Thu, Jun 7, 2018 at 5:05 PM, Pavel Petroshenko
> wrote:
>
>> Hi Ivan,
>>
>> Thanks for taking care of this. I will give the scripts a try and get
>> back to you if any questions.
>>
>> Could you please update the JIRA ticket [1] so that we keep it up-to-date.
>>
>> Thanks!
>> p.
>>
>> [1] https://issues.apache.org/jira/browse/IGNITE-8733
>>
>>
>> On Thu, Jun 7, 2018 at 4:47 AM, Иван Артюхов wrote:
>>
>>> Hi Igniters!
>>>
>>> I've prepared two scripts to benchmark the throughput of Node.JS thin
>>> client using 'atomic-put' operations [1]. They work in the following way:
>>> - Main script 'bench-starter.js' starts the given number of thin clients
>>> as
>>> sub-processes. AFAIK, Node.JS is one-threaded so we should fork clients
>>> from some parent process to make the benchmark fully utilize all CPU
>>> cores.
>>> - The Node.JS thin client script 'CachePut.js' uses the 'sandra'
>>> benchmark
>>> package [2] which is simple and suites the asynchronous logic of the
>>> Node.JS thin client package itself. Every second 'CachePut.js' calculates
>>> and prints on screen the average throughput for current iteration.
>>>
>>> I tried to make the logic of 'CachePut.js' to be close to Java thin
>>> client
>>> benchmarks for Yardstick framework available in a pull request [3].
>>> Because
>>> I'm not a Node.JS expert, it would be great if someone could review these
>>> two scripts and compare them with Java thin client benchmarks.
>>> Specifically
>>> the 'IgniteThinPutBenchmark.java' benchmark which also does atomic puts.
>>> Any
>>> feedback is greatly appreciated!
>>>
>>> [1] https://gist.github.com/iartiukhov/c02385d265330e2c9192931759616f95
>>> [2] https://www.npmjs.com/package/sandra
>>> [3] https://github.com/apache/ignite/pull/3942
>>>
>>> Thanks,
>>> Ivan
>>>
>>>
>>> вт, 29 мая 2018 г. в 20:05, Denis Magda :
>>>
>>> > Hi Pavel,
>>> >
>>> > Thanks for prompt improvements. I'll check them this week.
>>> >
>>> > --
>>> > Denis
>>> >
>>> > On Sun, May 27, 2018 at 5:04 PM, Pavel Petroshenko <
>>> pa...@petroshenko.com>
>>> > wrote:
>>> >
>>> > > Hi Denis,
>>> > >
>>> > > Thanks for your feedback on the documentation! I addressed all your
>>> > > comments from https://issues.apache.org/jira/browse/IGNITE-8589.
>>> > >
>>> > > Please let me know if you have any questions.
>>> > >
>>> > > Thanks,
>>> > >