Building and running reliable data platforms that scale and perform. An organizer of Seattle Postgres User Group. My opinions are my own.
Joined May 2011
- Tweets 4,883
- Following 415
- Followers 2,394
- Likes 6,126
351 Photos and videos
Pinned Tweet
8 Jun 2022
Latest PostgreSQL Happiness Hints: collection of things that I think every production PostgreSQL stack should have. Most things here are widely accepted, but a few are biased toward my opinions (eg scaling & dynamic conn pools). Mostly not original ideas; many people helped
9
65
370
Feb 19
We need to re-think OSS contribution attribution in light of AI. More than ever, it's important for committers to give credit on where the ideas are coming from. A committer can copy/paste someone else's ideas into their own prompts, and they need to give appropriate credit.
3
135
Jan 20
Poisoned conn pools in Go can escalate from row-level brownouts to complete db outages. This test suite demonstrates the failure cascade and shows how PgBouncer's peering feature (v1.19 ) prevents the escalation; although fixing leaks at the app level remains the true solution.
Jeremy Schneider (@jer_s): How Blocking-Lock Brownouts Can Escalate from Row-Level to Complete System Outages postgr.es/p/7qE
1
5
324
29 Nov 2025
great read
11 Sep 2025
You asked for longer rants, so here are longer rants! After almost 10 years I'm back to blogging. Thanks for the encouragement.
Link below in the thread 🔻
1
3
1,330
Jeremy Schneider retweeted
23 Oct 2025
21 Oct 2025
Agents are the New Developer
Agents, like Claude Code, feel uncanny. My first time using it, I built a mobile web app that tracked pushups using computer vision. Just for fun, to see what it could do. One hour later (mostly its time, not mine), I had an app that just worked.
That gave me goosebumps. For the first time, it felt like software wasn't something I built, it was something building with me. It felt like something brand new.
I realized: agents had become the new developer.
But software agents don't behave like human developers. Software development tools need to evolve. Agents need a new kind of database made for how they work.
So we built it. And we're launching it today.
Announcing Agentic Postgres: The First Database Built For Agents.
There's a lot of engineering behind this: a new copy-on-write block storage layer, fast zero-copy forks, new Postgres extensions for full text search (BM25) and semantic search, what (we think is) the best MCP server for Postgres ever built, and a new CLI and free tier.
I'm very proud of what this team has built, especially in such a short period of time.
More here: tigerdata.com/blog/postgres-…
Please give it a try. We're just getting started. We’d love your feedback. 🐯🚀
@TimescaleDB @TigerDatabase
1
3
674
17 Oct 2025
Two small tweaks (additions) to this article
1. General Principle: Schema = Source Code
2. When wait events indicate lock contention or increasing IO time, it’s extremely useful to see which tables are being operated on.
ardentperf.com/2025/10/15/sa…
155
16 Oct 2025
After a discussion on the public postgres slack, I POC'd a simple function that returns key parts of a SQL and avoids leaking literals - an alternative to full parsing (eg. PgAnalyze). The aim is to share query intent without exposing sensitive data. ardentperf.com/2025/10/15/sa…
1
2
3
776
6 Oct 2025
CloudNativePG preferred durability allows commits to proceed when sync standbys are unavailable, like Oracle DG Max Availability mode. In Jepsen tests, rapid failures caused [expected] data loss in 48% of 2‑node runs & dropped to 0% with 60s failoverDelay ardentperf.com/2025/10/05/te…
2
2
2,055
5 Oct 2025
A few thoughts about keeping your data safe without letting org silos crush the budget. Fundamentally it's about having an top-to-bottom picture of your stack and making sure teams work together on a holistic design. ardentperf.com/2025/10/04/da…
2
229
29 Sep 2025
Follow-up on Jepsen CloudNativePG: across 1,061 runs with synchronous replication, I saw 0 data-loss events. Async could hit 80% loss under rate/thread extremes and varied by instance type (Azure Dpsv6 was easiest to repro). Details lab scripts inside. ardentperf.com/2025/09/28/lo…
128
Jeremy Schneider retweeted
5 Sep 2025
Software engineering is 90% database and 10% frontend.
229
205
3,944
356,145
1 Sep 2025
Seattle Postgres is starting our fall series with a big meetup!
6pm this Thu Sep 4 special guest @samokhvalov - founder of PostgresAI & co-host of postgres.fm - VERY lucky to have him as a guest!
Please help repost to other social networks meetup.com/seattle-postgres/…
2
8
1,166
Jeremy Schneider retweeted
27 May 2025
Signups for June's #PostgreSQL hacking workshop are closing soon -- signup link is in this blog post rhaas.blogspot.com/2025/05/h… if you are interested in joining us
4
8
859
22 May 2025
1 really great postmortem
21 May 2025
Awesome post:
Root Cause Analysis: PostgreSQL MultiXact member exhaustion incidents (May 2025)
metronome.com/blog/root-caus…
1
1
1
477
22 May 2025
21 May 2025
Last week at @getmetronome, we hit a painful PostgreSQL edge case: MultiXact Member Space exhaustion. It caused multiple outages affecting our customers. I sincerely apologize to everyone impacted. 🧵 (1/4)
194
Jeremy Schneider retweeted
9 May 2025
No disrespect to Linus Torvalds, but this guy is the greatest geek alive 🫡
Created UNIX in 1971 when he was 28 years old.
Created Go in 2009 when he was 66 years old😲
He also developed the B programming language (which led to C), created UTF-8 encoding (making international text possible online), and designed essential tools like grep that developers still rely on daily.
He also helped with the development of Multics (that led to UNIX), Plan 9 from Bell Labs and Inferno operating systems.
That's 4 operating systems in total... Most people don't even use these many OS.
Pretty impressive resume, right? 🔥
And it's a shame that many people, even the ones in the IT and tech industry, don't know him.
Ken Thompson.... Remember the name 🙏
225
1,718
11,166
608,073
Jeremy Schneider retweeted
6 May 2025
Giving a talk on "Scaling Postgres to millions of tenants" tomorrow in Seattle Postgres meetup (and next week at @PGConfDev). I hope to see y'all there!
✅ per-tenant tablespaces
✅ inheritance foreign tables
✅ planning pain w/ 20k partitions
✅ extensions and smgr
Come if you love RLS, relcache, or regrets 😅
cc @jer_s
2
3
18
1,639
Wrote a timestamp to SQLite as "2006-01-02T15:04:05.999" instead of "2006-01-02 15:04:05.999" [1]. Every subsequent query failed. No error. No warning. Just silent failure.
Fellow Postgres users: guys, we've got it pretty good.
—
[1] Strings because SQLite has no time types.
3
2
28
3,686