From mboxrd@z Thu Jan  1 00:00:00 1970
Return-Path: <vdavydov.dev@gmail.com>
Date: Thu, 16 Aug 2018 22:42:31 +0300
From: Vladimir Davydov <vdavydov.dev@gmail.com>
Subject: Re: [PATCH] test: fix app-tap/tarantoolctl sporadic failure
Message-ID: <20180816194231.rgvalwtc7laimqai@esperanza>
References: <20180816152400.4237-1-sergepetrenko@tarantool.org>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20180816152400.4237-1-sergepetrenko@tarantool.org>
To: Serge Petrenko <sergepetrenko@tarantool.org>
Cc: tarantool-patches@freelists.org
List-ID: <tarantool-patches.dev.tarantool.org>

On Thu, Aug 16, 2018 at 06:24:00PM +0300, Serge Petrenko wrote:
> In rare cases this test failed on `tarantoolctl status` after
> `tarantoolctl stop` due to `tarantoolctl stop` taking too much time to
> unlink pid file.

This is too vague for an explanation. AFAIU the problem here is that the
pid file is deleted by tarantool (from tarantool_free => pidfile_remove)
while the control socket file is deleted by tarantoolctl so if
`tarantoolctl status` is called immediately after `tarantoolctl stop`,
there's a chance tarantool hasn't exited yet and so the pid file still
exists.

Please update the comment.

> Fix this by waiting for pid file deletion in test.
> 
> Closes #3557
> ---
> https://github.com/tarantool/tarantool/issues/3557
> https://github.com/tarantool/tarantool/tree/sp/gh-3557-tarantoolctl-test-fix
> 
> 
>  test/app-tap/tarantoolctl.test.lua | 9 +++++++++
>  1 file changed, 9 insertions(+)
> 
> diff --git a/test/app-tap/tarantoolctl.test.lua b/test/app-tap/tarantoolctl.test.lua
> index 6946c8312..83a8bfc37 100755
> --- a/test/app-tap/tarantoolctl.test.lua
> +++ b/test/app-tap/tarantoolctl.test.lua
> @@ -34,6 +34,13 @@ local function recursive_rmdir(path)
>      end
>  end
>  
> +local function wait_delete(path)
> +    path = fio.abspath(path)

You don't need to call abspath here.

> +    while fio.path.exists(path) do
> +	fiber.sleep(0.0001)

A tab! Please fix your editor.

Also, please use fiber.sleep(0.01) for consistency with other timeouts
used in this script.

> +    end
> +end
> +

There's tctl_wait() helper that waits for tarantool to start.
Let's rename it to tctl_wait_start() and introduce tctl_wait_stop()
with the same signature that would wait for tarantool to stop.

>  ffi.cdef[[
>  typedef int32_t pid_t;
>  int kill(pid_t pid, int sig);
> @@ -157,6 +164,7 @@ test:plan(6)
>  do
>      local dir = fio.tempdir()
>      local code = [[ box.cfg{memtx_memory = 104857600} ]]
> +    local pid_path = fio.pathjoin(dir, 'script.pid')
>      create_script(dir, 'script.lua', code)
>  
>      local status, err = pcall(function()
> @@ -168,6 +176,7 @@ do
>              check_ok(test_i, dir, 'start',  'script', 1, nil, "is already running")
>              check_ok(test_i, dir, 'status', 'script', 0, nil, "is running")
>              check_ok(test_i, dir, 'stop',   'script', 0, nil, "Stopping")
> +            wait_delete(pid_path)
>              check_ok(test_i, dir, 'status', 'script', 1, nil, "is stopped")
>              check_ok(test_i, dir, 'stop',   'script', 0, nil, "is not running")
>              check_ok(test_i, dir, 'status', 'script', 1, nil, "is stopped" )