Horton

Re: Review for GetWALAvailability()

Alvaro Herrera <alvherre@2ndquadrant.com>

View thread View on postgr.es
From: Alvaro Herrera <alvherre@2ndquadrant.com>
To: Fujii Masao <masao.fujii@oss.nttdata.com>
Cc: Kyotaro Horiguchi <horikyota.ntt@gmail.com>, pgsql-hackers@lists.postgresql.org
Date: 2020-06-17T18:04:37Z
Lists: pgsql-hackers

Attachments

I think passing the slot name when the slot is also passed is useless
and wasteful; it'd be better to pass NULL for the name and ignore the
strcmp() in that case -- in fact I suggest to forbid passing both name
and slot.  (Any failure there would risk raising an error during
checkpoint, which is undesirable.)

So I propose the following tweaks to your patch, and otherwise +1.

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

Commits

  1. Persist slot invalidation correctly

  2. Adjust max_slot_wal_keep_size behavior per review

  3. Save slot's restart_lsn when invalidated due to size

  4. Fix issues in invalidation of obsolete replication slots.