Replication & HA🔒 Pro lessonSource-grounded

pg_rewind: Reusing a Diverged Old Primary

After a failover, the promoted server becomes the new leader on a new branch of history.

If the old primary comes back, its history has split from the new one, and it may hold changes that never made it across, so you can't just plug it back in as a standby. pg_rewind undoes only the diverged part so the old server can rejoin without a full rebuild.

ProLesson body

Pro members see the rest of this lesson

  • A 4-step state diagram: Diverge → Identify → Copy → Rejoin
  • The full mechanism: what PostgreSQL does internally, in source terms
  • 1 SQL query you can run, each labelled by how it was verified
  • The closing insight, the mistake it prevents, and what it changes in your work

Card required. Cancel before day 7 and you are not charged.

Compare plans

Want to read a full lesson first? Five are free end to end, one from each major track. Start with MVCC internals.
Check it against the source1 citation in postgres/postgres · file, symbol and line verified on REL_17_STABLE
src/bin/pg_rewind/pg_rewind.cVerified · REL_17_STABLE

Primary symbol: main · line 121

Anchored to postgres/postgres on REL_17_STABLE and cross-checked against the manual for PostgreSQL 15–18. Every query here was run and its output captured on a throwaway PostgreSQL 17.10 lab; corrections are noted inline. pg_rewind in the official docs →

Connected

Where this lesson sits

What comes before, after, and alongside it.

Open in the interactive map →
ShareLinkedInX

Finished a free lesson?

Pro opens the rest of the engine course

You felt one mechanism. Pro is the full bodies, interview depth, and the tracks that build on this session.

FollowSubstackLinkedInnew errors · lab notes · hiring loops