reef-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Dhruv Mahajan <dhruv.maha...@gmail.com>
Subject Re: Reef 0.15
Date Fri, 22 Apr 2016 18:07:33 GMT
Hi Boris

I spent a day on this. There is one issue that I found which was related to
the issues at the time of disposing. However, even after I fixed it on my
machine two of the tests related to ThreeNodesCommunicationBothWays and
ThreeNodesCommunicationOneWay get stuck again. I tried logging and
debugging but it seemed to me that everything was fine but messages were
not received even though connections were open. Moreover, adding sleeps
between instantiating each remote manager makes it work which makes me
suspect some sort of situation where probably one of nodes is not ready.
However, logging did not suggest this. So this might take some time since
root issue is not detected yet. The fact that on local machine it never
fails makes it difficult to debug due to overhead associated with each
appveyor run.

Dhruv

On Thu, Apr 21, 2016 at 7:49 PM, Boris Shulman <shulmanb@gmail.com> wrote:

> Hi Dhruv
>
> Any updates on  REEF-1243?
>
> Boris.
>
> On Wed, Apr 20, 2016 at 5:04 PM, Markus Weimer <markus@weimo.de> wrote:
>
> > On 2016-04-20 10:56, Mariia Mykhailova wrote:
> >
> >> We have a history of tests which passed consistently on dev machines
> >> and failed only in Travis which turned out to be actual bugs in our
> >> code, for example, REEF-1040.
> >>
> >
> > +1
> >
> > We should be happy for every deterministically failing test which can can
> > address. We still subject our users to enough heisenbugs, given the
> nature
> > of the beast of distributed systems :)
> >
> > Markus
> >
>

Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message