reef-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Markus Weimer <mar...@weimo.de>
Subject Re: Regarding checking on cluster on C# side
Date Mon, 07 Dec 2015 20:28:19 GMT
Hi,

broadly +1

Some more detailed notes: there are two usage scenarios, and two levels 
of testing we need. The two usage scenarios are:

   1. Using a *released* version of REEF. I believe we do quite well on
      that, as testing on clusters is part of our release process.
   2. Using the current development state in `master`. That is untested
      on clusters.

The two kinds of tests we need are:

   1. Automated tests on every pull request as we have on the Java side.
      These are local-only tests, but they capture lots of gotchas
      already.
   2. The same, but on an actual cluster. I'm not sure how to find a
      cluster for that. Ideas?

Markus

On 2015-12-07 12:21, Boris Shulman wrote:
> We reached that state more then a year ago (ASA), but I do agree that we better have
e2e tests on real clusters.
>
> -----Original Message-----
> From: "Dhruv Mahajan" <dhruv.mahajan@gmail.com>
> Sent: ‎12/‎7/‎2015 12:04 PM
> To: "dev@reef.incubator.apache.org" <dev@reef.incubator.apache.org>
> Subject: Regarding checking on cluster on C# side
>
> Hi Guys
>
> On c# side I think we have reached at stage where people have started
> running higher level application on REEF constructs on actual cluster. My
> current understanding is that we do not have much serious testing currently
> on C# side on actual test clusters before committers commit the code. I
> strongly feel and think that now we need to check in the code only after
> testing features on actual cluster.
>
> What do you guys think?
>
> Dhruv
>

Mime
View raw message