Re: Memory leak with XLogFileCopy since de768844 (WAL file with .partial)

From: Fujii Masao <masao(dot)fujii(at)gmail(dot)com>
To: Michael Paquier <michael(dot)paquier(at)gmail(dot)com>
Cc: PostgreSQL mailing lists <pgsql-hackers(at)postgresql(dot)org>, Heikki Linnakangas <hlinnaka(at)iki(dot)fi>
Subject: Re: Memory leak with XLogFileCopy since de768844 (WAL file with .partial)
Date: 2015-06-04 13:40:36
Message-ID: CAHGQGwGWACThG=J5vfxzd8UfE87WGjd5Qms3sciTbHNREghLYQ@mail.gmail.com
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-hackers

On Mon, Jun 1, 2015 at 4:24 PM, Michael Paquier
<michael(dot)paquier(at)gmail(dot)com> wrote:
> On Thu, May 28, 2015 at 9:09 PM, Michael Paquier
> <michael(dot)paquier(at)gmail(dot)com> wrote:
>> Since commit de768844, XLogFileCopy of xlog.c returns to caller a
>> pstrdup'd string that can be used afterwards for other things.
>> XLogFileCopy is used in only one place, and it happens that the result
>> string is never freed at all, leaking memory.

That seems to be almost harmless because the startup process will exit
just after calling XLogFileCopy(). No?

> Having a second look at that, XLogFileCopy() is called only in one
> place, and dstfname is never used, hence I think that it would make
> sense to return unconditionally a temporary file name to the caller.

+1

> Also the current error message in case of failure is very C-like:
> elog(ERROR, "InstallXLogFileSegment should not have failed");
> I thought that we to use function names in the error messages.
> Wouldn't something like that be more adapted?
> "could not copy partial log file" or ""could not copy partial log file
> into temporary segment file"

Or we can extend InstallXLogFileSegment so that we can give the log level
to it. When InstallXLogFileSegment is called after XLogFileCopy, we can
give ERROR as the log level and cause an exception to occur if link() or
rename() fails in InstallXLogFileSegment.

Regards,

--
Fujii Masao

In response to

Responses

Browse pgsql-hackers by date

  From Date Subject
Next Message Robert Haas 2015-06-04 13:42:22 Re: [GENERAL] 9.4.1 -> 9.4.2 problem: could not access status of transaction 1
Previous Message Amit Kapila 2015-06-04 13:23:56 Re: Re: [COMMITTERS] pgsql: Map basebackup tablespaces using a tablespace_map file