[patches] [PATCH 1/1] test: ensure on_disconnect trigger is called on outstanding reqs

[v.shpilevoy at tarantool.org]

In the ticket I have proposed this: https://github.com/tarantool/tarantool/issues/2998#issuecomment-364901700 <https://github.com/tarantool/tarantool/issues/2998#issuecomment-364901700>
@kostja wrote this: https://github.com/tarantool/tarantool/issues/2998#issuecomment-364915567 <https://github.com/tarantool/tarantool/issues/2998#issuecomment-364915567>


> 12 февр. 2018 г., в 17:03, Vladimir Davydov <vdavydov.dev at gmail.com> написал(а):
> 
> On Mon, Feb 12, 2018 at 04:45:34PM +0300, Vladislav Shpilevoy wrote:
>> diff --git a/test/box/net.box.test.lua b/test/box/net.box.test.lua
>> index 378dfd4ab..4d190c22d 100644
>> --- a/test/box/net.box.test.lua
>> +++ b/test/box/net.box.test.lua
>> @@ -850,7 +850,8 @@ x = 0
>> for i = 1, N do x = x + long_call_channel:get() end
>> x
>> 
>> --- Check that a connection does not leak if there is
>> +--
>> +-- gh-2998: check that a connection does not leak if there is
> 
> I don't think this is worthwhile.
> 
> BTW I think original Kostja's concern was about the delay between
> closing the connection and invoking the trigger. Currently, we wait
> until all pending requests have been completed before we invoke the
> on_disconnect trigger. Kostja has suggested to push on_disconnect
> message immediately. Not sure if it's feasible though.
> 
>> -- a long CALL in progress when it is closed.
>> disconnected = false
>> function on_disconnect() disconnected = true end

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.tarantool.org/pipermail/tarantool-patches/attachments/20180212/8d6c5835/attachment.html>