Re: [PATCH] Send SIGTERM before SIGKILL on timeout

[Fumiaki Matsushima <mtsm.fm@gmail.com>]

Thank you for your reply!

> Right, it's been rejected several times before:

Sorry, I missed these emails.

> I don't want to encourage lazy app development or Rack apps
written with only unicorn in mind

That's true.
I'll use rack-timeout to solve my timeout problem.

Thanks.

On Sun, Feb 25, 2018 at 12:15 AM, Eric Wong <e@80x24.org> wrote:
> Fumiaki MATSUSHIMA <mtsm.fm@gmail.com> wrote:
>> To output log / send error to error tracking service,
>> we need to receive a signal other than SIGKILL first.
>> ---
>> Hi Unicorn team,
>>
>> I'm not sure this change is accetable though,
>> I can find some articles and patches to prevent SIGKILL
>> on timeout.
>
> Right, it's been rejected several times before:
>
>  https://bogomips.org/unicorn-public/20140416084416.GA9709@dcvr.yhbt.net/t/#u
>  https://bogomips.org/unicorn-public/20180115015740.GA850@dcvr/t/#u
>
>> I think it's great if this feature is supported by unicorn itself.
>>
>> Could you give me your opinion?
>
> Again, I don't want to encourage lazy app development or Rack apps
> written with only unicorn in mind.
>
> Existing `timeout' feature of unicorn is already something I hate
> and don't want to encourage further reliance on:
>
>         https://bogomips.org/unicorn/Application_Timeouts.html
>
> Fwiw, I (with ruby-core) will try to make Timeout in ruby stdlib
> have less overhead for 2.6 (or 2.7) so it can benefit more users
> than just unicorn.  Ditto with various OobGC hacks over the years.
>
> Thanks.