Horton

Re: Fix lag columns in pg_stat_replication not advancing when replay LSN stalls

Xuneng Zhou <xunengzhou@gmail.com>

View thread View on postgr.es
From: Xuneng Zhou <xunengzhou@gmail.com>
To: Fujii Masao <masao.fujii@gmail.com>
Cc: Shinya Kato <shinya11.kato@gmail.com>, PostgreSQL Hackers <pgsql-hackers@lists.postgresql.org>, Chao Li <li.evan.chao@gmail.com>
Date: 2025-10-21T14:52:39Z
Lists: pgsql-hackers 
Hi Fujii-san,

On Tue, Oct 21, 2025 at 7:45 AM Fujii Masao <masao.fujii@gmail.com> wrote:
>
> On Mon, Oct 20, 2025 at 1:45 PM Shinya Kato <shinya11.kato@gmail.com> wrote:
> >
> > On Mon, Oct 20, 2025 at 10:17 AM Fujii Masao <masao.fujii@gmail.com> wrote:
> > >
> > > On Sun, Oct 19, 2025 at 2:04 AM Shinya Kato <shinya11.kato@gmail.com> wrote:
> > > > Thank you for the patch. I have one comment.
> > > >
> > > > +       if (lag_tracker->overflowed[head].lsn > lsn)
> > > > +           return now - lag_tracker->overflowed[head].time;
> > > >
> > > > Could this return a negative value if the clock somehow went
> > > > backwards? The original code returns -1 in this case, so I'm curious
> > > > about this.
> > >
> > > Thanks for the review!
> > >
> > > Yes, you're right. So I've updated the patch so that -1 is returned
> > > when the current time is earlier than the time in the overflow entry,
> > > treating it as "no new sample found".
> >
> > Thank you for updating the patch. It looks nice to me.
>
> Thanks for the review!
> Unless there are any objections, I'll commit the patch and
> backpatch it to all supported branches.
>

Thanks for working on this.

The patch LGTM.  I am wondering whether it is helpful to add some
comments for this overflowed array
WalTimeSample overflowed[NUM_SYNC_REP_WAIT_MODE];

and replacing literal zeros with the constant InvalidXLogRecPtr for
better readability.

/* InvalidXLogRecPtr means no overflow yet */
if (lag_tracker->overflowed[i].lsn == InvalidXLogRecPtr)

Best,
Xuneng

Commits

  1. Add comments explaining overflow entries in the replication lag tracker.

  2. Fix stalled lag columns in pg_stat_replication when replay LSN stops advancing.