Message-ID: <7579f627-aad4-7a04-1a98-033cc6b3cca5@dunslane.net>
Date: Tue, 4 Oct 2022 01:39:51 -0400
MIME-Version: 1.0
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:91.0) Gecko/20100101
 Thunderbird/91.5.0
Subject: Re: ssl tests aren't concurrency safe due to get_free_port()
Content-Language: en-US
To: Andres Freund <andres@anarazel.de>, pgsql-hackers@postgresql.org,
 Tom Lane <tgl@sss.pgh.pa.us>
References: <20221002164931.d57hlutrcz4d2zi7@awork3.anarazel.de>
From: Andrew Dunstan <andrew@dunslane.net>
In-Reply-To: <20221002164931.d57hlutrcz4d2zi7@awork3.anarazel.de>
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit
Archived-At: 
 <https://www.postgresql.org/message-id/7579f627-aad4-7a04-1a98-033cc6b3cca5%40dunslane.net>
Precedence: bulk


On 2022-10-02 Su 12:49, Andres Freund wrote:
>
> 2) Use a lockfile containing a pid to protect the choice of a port within a
>    build directory. Before accepting a port get_free_port() would check if the
>    a lockfile exists for the port and if so, if the test using it is still
>    alive.  That will protect against racyness between multiple tests inside a
>    build directory, but won't protect against races between multiple builds
>    running concurrently on a machine (e.g. on a buildfarm host)
>
>

I think this is the right solution. To deal with the last issue, the
lockdir should be overrideable, like this:


  my $port_lockdir = $ENV{PG_PORT_LOCKDIR} || $build_dir;


Buildfarm animals could set this, probably to the global lockdir (see
run_branches.pl). Prior to that, buildfarm owners could do that manually.


There are numerous examples of lockfile code in the buildfarm sources.
I'll try to hack something up.


cheers


andrew

--
Andrew Dunstan
EDB: https://www.enterprisedb.com