Hi,

I thought I would try and do the async version of my mock AVR850 using the vibe.d TCP stuff. This is not HTTP, it is proper networking! ;-)

Problem one is that vibe.d sever processes never seem to terminate.

I am using runApplication or more likely runEventLoop in my main but it never returns. Starting the server from the command line, I Ctrl+C the process and get the message:

[main(----) INF] Received signal 2. Shutting down.

but whilst I get a command line prompt, the process is not terminated.

This is a bit of a blocker so help very much required.

-- 
Russel.
===========================================
Dr Russel Winder      t: +44 20 7585 2200
41 Buckmaster Road    m: +44 7770 465 077
London SW11 1EN, UK   w: www.russel.org.uk

On Saturday, 23 May 2020 at 15:47:59 UTC, Russel Winder wrote:
> Hi,
>
> I thought I would try and do the async version of my mock AVR850 using the vibe.d TCP stuff. This is not HTTP, it is proper networking! ;-)
>
> Problem one is that vibe.d sever processes never seem to terminate.
>
> I am using runApplication or more likely runEventLoop in my main but it never returns. Starting the server from the command line, I Ctrl+C the process and get the message:
>
> [main(----) INF] Received signal 2. Shutting down.
>
> but whilst I get a command line prompt, the process is not terminated.
>
> This is a bit of a blocker so help very much required.

Had similar experiences on non-windows platforms.

I think the solution is just to kill the pid.

On Sat, 2020-05-23 at 16:56 +0000, bauss via Digitalmars-d-learn wrote: […]
> 
> Had similar experiences on non-windows platforms.

I forgot to mention I was on Linux (Debian Sid).

I am (sort of) pleased it isn't just me – definitely unhappy it has happened
to others as well.

> I think the solution is just to kill the pid.

Killing from the command line is not really an option as the process is used in integration tests.

I guess I'll have to send the process SIGKILL rather than SIGTERM, which is a bit annoying.

-- 
Russel.
===========================================
Dr Russel Winder      t: +44 20 7585 2200
41 Buckmaster Road    m: +44 7770 465 077
London SW11 1EN, UK   w: www.russel.org.uk

On Saturday, 23 May 2020 at 17:57:22 UTC, Russel Winder wrote:
> On Sat, 2020-05-23 at 16:56 +0000, bauss via Digitalmars-d-learn wrote: […]
>> 
>> Had similar experiences on non-windows platforms.
>
> I forgot to mention I was on Linux (Debian Sid).
>
> I am (sort of) pleased it isn't just me – definitely unhappy it has happened
> to others as well.
>
>> I think the solution is just to kill the pid.
>
> Killing from the command line is not really an option as the process is used in integration tests.
>
> I guess I'll have to send the process SIGKILL rather than SIGTERM, which is a bit annoying.

Yes, when I saw it happening then I thought it was probably just me.

Since it's not then perhaps an issue should be reported for vibe.d

On Saturday, 23 May 2020 at 15:47:59 UTC, Russel Winder wrote:
> Hi,
>
> I thought I would try and do the async version of my mock AVR850 using the vibe.d TCP stuff. This is not HTTP, it is proper networking! ;-)
>
> Problem one is that vibe.d sever processes never seem to terminate.
>
> I am using runApplication or more likely runEventLoop in my main but it never returns. Starting the server from the command line, I Ctrl+C the process and get the message:
>
> [main(----) INF] Received signal 2. Shutting down.
>
> but whilst I get a command line prompt, the process is not terminated.
>
> This is a bit of a blocker so help very much required.

It’s a recent problem with vibe-core. I’ve had to switch back to version 1.8.1.

https://github.com/vibe-d/vibe-core/issues/205

-Steve

On Sat, 2020-05-23 at 20:52 +0000, Steven Schveighoffer via Digitalmars-d-
learn wrote:
[…]
> 
> It’s a recent problem with vibe-core. I’ve had to switch back to version 1.8.1.
> 
> https://github.com/vibe-d/vibe-core/issues/205

Really good to know this is a genuine problem and not just me doing something wrong! :-)

For my purposes switching to using SIGKILL rather than SIGTERM in my tests seems to work with 1.9.1, so I'll go with that till 1.9.2 or 1.10.0 produces a fix rather than revert to 1.8.1.

-- 
Russel.
===========================================
Dr Russel Winder      t: +44 20 7585 2200
41 Buckmaster Road    m: +44 7770 465 077
London SW11 1EN, UK   w: www.russel.org.uk

On Sun, May 24, 2020 at 10:06 AM Russel Winder via Digitalmars-d-learn <digitalmars-d-learn@puremagic.com> wrote:
>
> For my purposes switching to using SIGKILL rather than SIGTERM in my tests seems to work with 1.9.1, so I'll go with that till 1.9.2 or 1.10.0 produces a fix rather than revert to 1.8.1.
>

You can use VibeHighEventPriority version in your dub as a workaround for now, there is no need to revert to 1.8.1

On 5/25/20 5:49 AM, Daniel Kozak wrote:
> On Sun, May 24, 2020 at 10:06 AM Russel Winder via Digitalmars-d-learn
> <digitalmars-d-learn@puremagic.com> wrote:
>>
>> For my purposes switching to using SIGKILL rather than SIGTERM in my tests
>> seems to work with 1.9.1, so I'll go with that till 1.9.2 or 1.10.0 produces a
>> fix rather than revert to 1.8.1.
>>
> 
> You can use VibeHighEventPriority version in your dub as a workaround
> for now, there is no need to revert to 1.8.1
> 

I'm thinking of actually reverting to something earlier or trying to get 1.9.0 to work. I am getting tons of spurious errors in vibe-core (see [1]). The latest cost me about 2 days of tracking down an error, where occasionally (still not sure of the situation) the connection to the database is spuriously closed before the response can be fully read.

I was not getting these in an earlier version of the build.

-Steve

[1] https://github.com/vibe-d/vibe-core/issues/208