From: | Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> |
---|---|
To: | Gilles Darold <gilles(dot)darold(at)dalibo(dot)com> |
Cc: | "Shulgin, Oleksandr" <oleksandr(dot)shulgin(at)zalando(dot)de>, Robert Haas <robertmhaas(at)gmail(dot)com>, PostgreSQL Hackers <pgsql-hackers(at)postgresql(dot)org> |
Subject: | Re: Patch to implement pg_current_logfile() function |
Date: | 2016-03-11 14:22:03 |
Message-ID: | 676.1457706123@sss.pgh.pa.us |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-hackers |
Gilles Darold <gilles(dot)darold(at)dalibo(dot)com> writes:
> Le 11/03/2016 10:49, Shulgin, Oleksandr a crit :
>> Would it make sense to have it as a symlink instead?
> The only cons I see is that it can be more "difficult" with some
> language to gather the real path, but do we really need it? There is
> also little time where the symlink doesn't exist, this is when it needs
> to be removed before being recreated to point to the new log file.
Yeah, a symlink is impossible to update atomically (on most platforms
anyway). Plain file containing the pathname seems better.
Another point is that we might not necessarily want *only* the pathname in
there. postmaster.pid has accreted more stuff over time, and this file
might too. I can see wanting the syslogger PID in there, for example,
so that onlookers can detect a totally stale file. (Not proposing this
right now, just pointing out that it's a conceivable future feature.)
regards, tom lane
From | Date | Subject | |
---|---|---|---|
Next Message | Petr Jelinek | 2016-03-11 14:24:48 | Re: [PATCH] Logical decoding support for sequence advances |
Previous Message | Julien Rouhaud | 2016-03-11 14:16:19 | Re: auto_explain sample rate |