ignite-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Dmitriy Pavlov <dpavlov....@gmail.com>
Subject Re: Bots on dev list
Date Mon, 10 Sep 2018 16:10:57 GMT
Hi Vladimir,

In general, I agree with these ideas, but it will be a lot of work to be
done to make personal notification even possible. We need to match each TC
and GitHub, GitHub account with email, this could become huge manual work.
We need to develop personal notifications code. Would you like to make a
contribution to these changes?

So I found more realistic to
A. group by suite and/or
B. implement some digest-like notification (to avoid mass emails).
C. Fix this strange bug with several notifications about one suite timeout
(I will do my absolute best to find out why it happens).

About point 1 I'm not sure I agree because
1. Nobody will read a separate list,
2. and it not possible to discuss failure with the rest of the community.
Please note that healthy practice is that someone will reply to failure
thread and email one more time when failure is fixed.

So I suggest concentrating on avoiding of failing a test, so the bot will
be silent :).

Sincerely,
Dmitriy Pavlov

пн, 10 сент. 2018 г. в 18:34, Vladimir Ozerov <vozerov@gridgain.com>:

> Folks,
>
> If there will be a single thread for all failures, nobody will read it.
> Simply because it will be always "unread", with tons of messages, where it
> would be impossible to find a failure I am responsible for.
> Several threads - on per suite - makes situation a bit better, but still
> nobody will read it. What is worse, there will be a lot of fat threads atop
> of all other threads, so e.g. the very first page of GMail will always be
> filled with failure messages.
>
> In order to make things comfortable and efficient, we should strive to make
> messages as specific as possible, instead of grouping unrelated failures
> into one thread. IMO we can solve the problem as follows:
> 1) Move bot messages to separate mailing list
> 2) Add personal contributor email addresses to that messages for easy
> filtering
> 3) If needed - add *aggregate* email, sent once a day (or several days) to
> dev list with all the issues.
>
> This way, we first reach contributor directly. Then we send reminder to the
> dev list, without flooding.
>
> Thoughts?
>
>
> On Mon, Sep 10, 2018 at 5:11 PM Alexey Kuznetsov <akuznetsov@apache.org>
> wrote:
>
> >  Dmitriy,
> >
> > I like your idea to add suite name.
> >
> > On Mon, Sep 10, 2018 at 5:42 PM Dmitriy Pavlov <dpavlov.spb@gmail.com>
> > wrote:
> >
> > > Hi Alexey,
> > >
> > > Thank you for that brilliant idea. Build ID was added intentionally to
> > > separate topics. But for timeout failures, it creates odd topics.
> > >
> > > I would like to suggest a modification of the idea: What if we will
> > include
> > > a suite name instead of build ID. It will keep the context of suite
> > > failures and fixes in one topic. E.g. recent problematic suite:
> Zookeeper
> > > will originate 1 or 2 topics, and human reply to it will be kept
> together
> > > with suite failure context. Please share your vision.
> > >
> > > Sincerely.
> > > Dmitriy Pavlov
> > >
> > > пт, 7 сент. 2018 г. в 8:25, Alexey Kuznetsov <akuznetsov@apache.org>:
> > >
> > > > Dmitriy,
> > > >
> > > > How about to change e-mail subject
> > > > from *[MTCGA]: new failures in builds [nnnnn] needs to be handled*
> > > > to *[MTCGA]: new failures in builds needs to be handled* ?
> > > >
> > > > If subject will be the same it will not produce lots of threads on
> dev
> > > list
> > > > and it will be a single thread.
> > > > What do you think?
> > > >
> > > > --
> > > > Alexey Kuznetsov
> > > >
> > >
> >
> >
> > --
> > Alexey Kuznetsov
> >
>

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