potiuk commented on issue #17977:
URL: https://github.com/apache/airflow/issues/17977#issuecomment-1397715678
> First, we understand the sorry reality of OSS development in which there's
this huge number of entitled, mostly corporate users that demand updates, plans
and SLA's from a project without giving anything back to the hundreds of people
working on their free time to keep the project running.
No. It's not sorry. It's thriving, and fufliflls requirements and follows
the lead of those who contribute. There is nothing sorry about it.
> We wanted to know if this is an known issue that someone was working on or
if we still needed to provide more information.
Again - this is a complete misunderstanding of how it works here. this is
not a corporate structure where you can ask one person "what's the state". You
have very wrong assumption about how things work in open-source project like
this. There is not a single entity or person, that can answer it. If someone
works on it, then that someone works on it. If somoene submits a fix, then
someone submits a fix. And everyone else sees it. All discussions are public.
And anyone - as any other person here can see it. There is not a single entity
you can get an authoritative answer on whether someone is working on a random
issue. No-one knows. No-one tracks it - other than what happens and is
documented in issues like this one.
> This was specifically asked because we knew this is a pretty old issue
that might be completely unrelated to what we and some other users are
experiencing on 2.3.4. In addition to this, this is the first time we are
trying to debug / collaborate on a core airflow problem, so we are not sure how
the project bug-fixing process works. Sorry if this came across wrong.
It did not come "wrong". I have no "hard" feelings. I was **just** answering
your questions. Simply answering the question "is someone working on it" by
anyone (unless that person is working on fixing it) is impossible. Neither I,
nor anyone else knows, nor track who is working on what and what's the state.
This is a completely distributed way of work. Ther are some "projects" that get
semi-organized, but on random issues like that one, there is no asnwer. It will
get fixed when someone fixes it. this is the best answer anyone can get.
> 1. Would you prefer we open a new issue to track this, since it most
likely is a different issue than the one reported here? I think that would
allow us to provide clearer information about our specific issue. Is GitHub
issues the preferred place to create this or do you track issues somewhere else?
If you think you can provide a reproducible case, then by all means open a
new issue. Or provide mroe details on that one. Especially - if you migrate
first to latest available version. Airlfow has hundreds bugs fixed every
patchlevel release. The original issue has been reported on 2.1.3 whixh is >
1.5 yeear old and likely there are hundreds (if not thousands) bugs fixed
since. And it's a complete loose of everyone's time to try to understand if the
issue has been fixed since. It's way better - if you are really interested in
fixing it - to upgrade to latest and see if you can still see it. So by all
means - please, upgrade to latest airflow that you can and if you can reproduce
it, open an issue describing it with all details you can muster.
> 2. What kind of information would be useful to try and triage this? Again,
this is the first time we try to collaborate on airflow core (we've worked on
custom operators, plugins, etc. before, but never airflow core) and we are a
bit lost on what would be useful.
All. That you can. Explain what investigations you've done so far, what
you've tried and observations you had. Logs, circumstances, related logs from
database around that time (if your investigations indicate they might be
relevant). The thing is that when you report such errors, there is no "paid
support" here. People who create airflow MIGHT help if they dedicate their own,
personal time in trying to understand and diagnose YOUR problem. Which at the
end might happen to be Airlfow problem, but just spending time on trying to
understand it is usually unpaid time that people here might decide to spend
time on. So do whatever you think is necessary to make this time as little as
possible, so those people will eventually spend their time on solving your
problem. But you have to understand that you are asking people for their fee
time - to help your company to solve a commercial problem. So do whatever you
think will make those people to want to spend their time.
One of the best ways to do it, is to show that you spend YOUR time, that you
made effort to analyse the problem and tried to see what is wrong and that you
tested and verified some hypotheses and you can share the result of that that
will help to loose as little time on diagnosing as possible. this is the best
way to make people want to help you - because they see this is not a demand but
cooperation.
There is no easy recipe for that. If we had an easy recipe, we would
automate it. But usually problems like that have no easy recipes because you do
not know what's wrong and it needs a human investing time and effort and
looking at different hypotheses and providing useful insight.
Alternativelly - if you really WANT things to fix it badly - there are
companies and individuals who provide paid support for Airlfow that you can pay
for to get them look at your problem - usually that is connected with paying
for their time and giving them access to your systems (possibly interactive
with your employees that serve as "proxy' if the access cannot be given
directly) to be able to analyse things.
> Thank you, and again, sorry if this came across wrong.
There is nothing to be sorry about. I do not feel anything was "wrong
there". I just want to be plain and straightforward that any support here is
not a "helpdesk" - it is provided on voluntary and "as-is" basis and it's you
who ask for help, and you who should make all the effort that the ask for help
might get successful.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
