Michael Paquier - PostgreSQL committer

  • Home
  • Manuals
  • Presentations
  • Projects
  • Resume
  • Tags
  • About

Postgres 10 highlight - Temporary replication slots

2016-12-18 12:25:24 +0000

Postgres 10 has support for an additional feature related to replication slots, a facility holding a way to retain WAL data in pg_xlog depending on the data consumed by clients connected to it. The feature spoken about here has been implemented by the following commit:

Read more...

Postgres 10 highlight - read-write and read-only mode of libpq

2016-12-09 13:44:12 +0000

libpq is getting some improvements in its connection strings to define some properties that are expected from the backend server thanks to the following commit, that will be part of Postgres 10:

Read more...

Postgres 10 highlight - Multiple hosts in connection strings

2016-12-05 06:20:53 +0000

Client applications dependent on libpq can use connection strings to connect to a Postgres server, which is something present for ages. Postgres 10 is introducing a new flavor in that by allowing applications to define multiple connection points, for a feature introduced by the following commit:

Read more...

Postgres 10 highlight - pg_basebackup improvements

2016-11-20 08:40:41 +0000

Lately three improvements and behavior changes have been added to pg_basebackup that will be part of Postgres 10.

Read more...

Postgres 10 highlight - Wait events for latches

2016-10-23 13:48:34 +0000

Postgres 9.6 has added a really cool infrastructure called wait events. This allows a developer to define in the code being run a wait point that gets reported to the system statistics via it PGPROC entry. In short, a custom wait point in the code gets reported and is then visible in system catalogs. In this case this is pg_stat_activity via the columns wait_event_type and wait_event. 9.6 infrastructure shows up information for backend processes holding lightweight locks, heavyweight locks, and buffer pins. While getting a look at this infrastructure, I got surprised by the fact that nothing was reported for latches, more or less code paths calling WaitLatch() to wait for a timeout, a postmaster death, a socket event or just for the latch to be set. As 9.6 was close to shipping when I bumped into the limitation, nothing much could be done for it. So this got delayed to Postgres 10, and has been committed recently with the following change:

Read more...

« Prev 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 Next »

Search

Social

Github

RSS Feeds

Main

Postgres

Sponsor

Unless otherwise specified, the contents of this website are (C)Copyright Michael Paquier 2010-2025 and are licensed for use under CC BY-NC-ND 4.0.