Re: DataStreamer usability

2017-09-05 Thread Denis Mekhanikov
Here is another little example of non-obvious DataStreamer-related API: https://stackoverflow.com/questions/46045987/unknown-pair-exception-when-using-streamtransformer-for-binaryobject/46051447?noredirect=1#comment79063618_46051447 I think, passing new value as the first element of Object array

Re: DataStreamer usability

2017-09-03 Thread Dmitriy Setrakyan
On Sun, Sep 3, 2017 at 11:42 AM, Evgenii Zhuravlev wrote: > >But why would anyone have different configurations for streaming into the > >same cache on the same node? In this case, we should print out a warning > if > >the configuration of one streamer is different

Re: DataStreamer usability

2017-09-03 Thread Evgenii Zhuravlev
>But why would anyone have different configurations for streaming into the >same cache on the same node? In this case, we should print out a warning if >the configuration of one streamer is different from another. Agree? Well, I have not come across the necessity to use different datastreamers on

Re: DataStreamer usability

2017-09-03 Thread Dmitriy Setrakyan
On Sun, Sep 3, 2017 at 10:09 AM, Evgenii Zhuravlev wrote: > Dmitriy, > > >Disagree. This is an Ignite issue, not a user issue. We must detect this > >and either provide a user with a warning message or disallow 2 streamers > >with identical names. > >Makes sense? > >

Re: DataStreamer usability

2017-09-03 Thread Evgenii Zhuravlev
Dmitriy, >Disagree. This is an Ignite issue, not a user issue. We must detect this >and either provide a user with a warning message or disallow 2 streamers >with identical names. >Makes sense? It's not a streamer name we apply as an argument, it's a cache name. So, it's possible to use a

Re: DataStreamer usability

2017-09-03 Thread Dmitriy Setrakyan
On Sun, Sep 3, 2017 at 9:36 AM, Evgenii Zhuravlev wrote: > What about this SO question - I think it could be a good idea to add some > additional validation between nodes configurations at node's start. > Because, if the root of the problem was found correctly, then it

Re: DataStreamer usability

2017-09-03 Thread Evgenii Zhuravlev
What about this SO question - I think it could be a good idea to add some additional validation between nodes configurations at node's start. Because, if the root of the problem was found correctly, then it means that nodes were started with inconsistent configurations. What do you think? Sure,

Re: DataStreamer usability

2017-09-03 Thread Dmitriy Setrakyan
In that case we either provide bad configuration or bad error messages. Could we provide a better error message for this SO issue? Evgenii, can I please ask you to provide links to all the data streamer questions you looked at here? This way we may have a chance to spot some area for improvement.

Re: DataStreamer usability

2017-09-03 Thread Evgenii Zhuravlev
Dmitriy, I've seen several questions on StackOverflow and on the user list, that seems to be connected with Data Streamer at first sight, but after some investigation, it was clear that they were not related to Data Streamer at all. Usually, as it was in this question on SO, it was a wrong

DataStreamer usability

2017-09-02 Thread Dmitriy Setrakyan
Igniters, I am noticing quite a few issues on the user list and Stack Overflow that have to do with Data Streamer usability. Here is just one example: https://stackoverflow.com/questions/46015833/datastreamer-adddata-method-not-work-as-expected-in-cluster I think, as a community, we should try