On Fri, Sep 15, 2023 at 7:47 AM Heikki Linnakangas wrote:
> Thinking about this some more, I think this is still not 100% correct,
> even with the patch I posted earlier:
This is marked as needing review, but that doesn't appear to be
correct, because there's this comment, indicating that the pat
On Wed, Sep 20, 2023 at 11:22:10PM -0700, Noah Misch wrote:
> On Fri, Sep 15, 2023 at 02:47:45PM +0300, Heikki Linnakangas wrote:
> > On 05/09/2023 21:20, Robert Haas wrote:
>
> > Thinking about this some more, I think this is still not 100% correct, even
> > with the patch I posted earlier:
> >
On Fri, Sep 15, 2023 at 02:47:45PM +0300, Heikki Linnakangas wrote:
> On 05/09/2023 21:20, Robert Haas wrote:
> Thinking about this some more, I think this is still not 100% correct, even
> with the patch I posted earlier:
>
> > /*
> > * When we WAL-logged rel pages, we must nonetheless
On 05/09/2023 21:20, Robert Haas wrote:
In other words, somehow it feels like we ought to be trying to defer
the fsync here until a clean shutdown actually occurs, instead of
performing it immediately.
+1
Admittedly, the bookkeeping seems like a problem, so maybe this is
the best we can do, b
On 04/09/2023 16:59, Melanie Plageman wrote:
The patch looks reasonable to me. Is this [1] case in hash index build
that I reported but didn't take the time to reproduce similar?
[1]
https://www.postgresql.org/message-id/CAAKRu_bPc81M121pOEU7W%3D%2BwSWEebiLnrie4NpaFC%2BkWATFtSA%40mail.gmail.com
On Tue, Sep 05, 2023 at 02:20:18PM -0400, Robert Haas wrote:
> The general rule throughout the system is that the init-fork of an
> unlogged relation is treated the same as a permanent relation: it is
> WAL-logged and fsyncd. But the other forks of an unlogged relation are
> neither WAL-logged nor
On Fri, Aug 25, 2023 at 8:47 AM Heikki Linnakangas wrote:
> 1. Create an unlogged table
> 2. ALTER TABLE unlogged_tbl SET TABLESPACE ... -- This calls
> RelationCopyStorage
> 3. a checkpoint happens while the command is running
> 4. After the ALTER TABLE has finished, shut down postgres cleanly.
>
On Fri, Aug 25, 2023 at 8:47 AM Heikki Linnakangas wrote:
>
> I noticed another missing fsync() with unlogged tables, similar to the
> one at [1].
>
> RelationCopyStorage does this:
>
> > /*
> >* When we WAL-logged rel pages, we must nonetheless fsync them. The
> >* reason i
On Fri, Aug 25, 2023 at 03:47:27PM +0300, Heikki Linnakangas wrote:
> That 'copying_initfork' condition is wrong. The first hint of that is that
> 'use_wal' is always true for an init fork. I believe this was meant to check
> for 'relpersistence == RELPERSISTENCE_UNLOGGED'. Otherwise, this bad thin
I noticed another missing fsync() with unlogged tables, similar to the
one at [1].
RelationCopyStorage does this:
/*
* When we WAL-logged rel pages, we must nonetheless fsync them. The
* reason is that since we're copying outside shared buffers, a
CHECKPOINT
10 matches
Mail list logo
