On Tue, Mar 13, 2012 at 12:03 PM, Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:
> Fujii Masao <masao(dot)fujii(at)gmail(dot)com> writes:
>> On Sat, Mar 10, 2012 at 5:04 AM, Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:
>>> The main actual simplification would be in getting rid of the "hole"
>>> at the end of each 4GB worth of WAL, cf this bit in xlog_internal.h:
>>> If we can't get rid of that and have a continuous 64-bit WAL address
>>> space then it's unlikely we can actually simplify any logic.
>>> Now, doing that doesn't break the naming convention exactly; what it
>>> changes is that there will be WAL files numbered xxxFFFF (for some
>>> number of trailing-1-bits I'm too lazy to work out at the moment) where
>>> before there were not. So the question really is how much external code
>>> there is that is aware of that specific noncontiguous numbering behavior
>>> and would be broken if things stopped being that way.
>> A page header contains WAL location, so getting rid of "hole" seems to
>> break pg_upgrade. No?
> No, why would it do that? The meaning and ordering of WAL addresses is
> the same as before. The only difference is that after the upgrade, the
> system will stop skipping over 16MB of potentially usable WAL addresses
> at the end of each subsequently-used 4GB of space. The holes before
> the switchover point are still holes, but that doesn't matter.
Oh, I see. You're right.
NIPPON TELEGRAPH AND TELEPHONE CORPORATION
NTT Open Source Software Center
In response to
pgsql-hackers by date
|Next:||From: Jeff Davis||Date: 2012-03-13 03:49:40|
|Subject: Re: initdb and fsync|
|Previous:||From: Joachim Wieland||Date: 2012-03-13 03:35:52|
|Subject: Re: patch for parallel pg_dump|