On Fri, Sep 9, 2011 at 7:46 PM, Florian Pflug <fgp(at)phlo(dot)org> wrote:
>> I am going to use the whole partition device for the DB (like /dev/sda1)
>> , so no filesystem code will be used. Also I am using asynchronous IO
>> (the aio_read and aio_write) and I don't know if they can be combined
>> with the fdatasync() syscall?
> Someone else (maybe the POSIX spec?) must answer that as I know very little
> about asynchronous IO.
There's an aio_fsync as part of the aio api, But you could use fsync
or fdatasync -- I assume you would have to wait for the aio_write to
have finished before you issue the fsync. But if you're going to
fdatasync all your writes right away you may as well open with O_DSYNC
which is, I gather, exactly how aio is intended to be used.
In response to
pgsql-hackers by date
|Next:||From: Alvaro Herrera||Date: 2011-09-10 01:31:53|
|Subject: nonempty default log_line_prefix (was [COMMITTERS] pgsql: Simplify handling of the timezone GUC)|
|Previous:||From: Greg Stark||Date: 2011-09-09 23:07:28|
|Subject: Re: unite recovery.conf and postgresql.conf|