Horton

Re: [COMMITTERS] pgsql: Fix freezing of a dead HOT-updated tuple

Alvaro Herrera <alvherre@alvh.no-ip.org>

View thread View on postgr.es
From: Alvaro Herrera <alvherre@alvh.no-ip.org>
To: Peter Geoghegan <pg@bowt.ie>
Cc: PostgreSQL Hackers <pgsql-hackers@postgresql.org>, "Wood, Dan" <hexpert@amazon.com>
Date: 2017-10-04T12:06:09Z
Lists: pgsql-hackers

Attachments

Alvaro Herrera wrote:
> Peter Geoghegan wrote:
> 
> > I thought that we no longer store FrozenTransactionId (xid 2) as our
> > "raw" xmin while freezing, and yet that's what we see here.
> 
> I'm doing this in 9.3.  I can't tell if the new tuple freezing stuff
> broke things more thoroughly, but it is already broken in earlier
> releases.

In fact, I think in 9.3 we should include this patch, to set the Xmin to
FrozenXid.  9.4 and onwards have commit 37484ad2a "Change the way we
mark tuples as frozen" which uses a combination of infomask bits, but in
9.3 I think leaving the unfrozen value in the xmax field is a bad idea
even if we set the HEAP_XMAX_COMMITTED bit.

-- 
Álvaro Herrera                https://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services

Commits

  1. Backport addition of rs_old_rel to rewriteheap's state.

  2. Perform a lot more sanity checks when freezing tuples.

  3. Revert bogus fixes of HOT-freezing bug

  4. Fix traversal of half-frozen update chains

  5. Fix freezing of a dead HOT-updated tuple

  6. During index build, check and elog (not just Assert) for broken HOT chain.

  7. Fix WAL replay of locking an updated tuple

  8. Change the way we mark tuples as frozen.