chmonitor v0.3.4 is out — Tools, Schema Compare, workspace roles
After four days of agents working day and night — 106 PRs, 267 comments — v0.3.4 is the tag for Tools, workspace roles, Schema Compare, and the shift to self-hosted licenses.
Licenses - We transformed our SaaS Cloud seat model to a self-hosted host-count license. The full argument is We’re selling self-hosted licenses.
https://t.co/YSVMhPk1mr
After a year of ignoring bug reports and complaints from everyone 🫣
I finally decided to burn some more tokens on ClickHouse Monitoring. This is the v0.3 stable release:
- Full rewrite on TanStack Start
- Cleaner design and charts
- Built for speed and run on Cloudflare Workers, Docker, Kubernetes
- Built-in AI agent + AI Insights, or MCP server for your own agent (still bad, working on it)
- Advisors for query tuning, projections, skip indexes, MVs for Claude / Cursor
- Cluster topology, traffic/ingestion, health with drill-down
- Alerting (beta), Postgres (beta) and PeerDB CDC monitoring
If you haven’t heard of https://t.co/rBh1wIZt96 yet: it's a UI to monitor and overview your ClickHouse cluster at any size. Queries, merges, parts, replication, health. An AI advisor on top (of course). Open source.
Thanks to duyetbot working on it day and night. If you find a bug, just open an issue - the fix might land the next morning.
Github: https://t.co/AEr1Gro8bo a star helps a lot 🙏
Homepage: https://t.co/0fNpexLha2
Changelog: https://t.co/zjWNNG11Pv
I just found and purchased a good tool for recording launch videos, but it still lacks features like adding missing videos to the timeline. Do you have any other recommendations?
I am working on the https://t.co/gCWJNGBlOy v0.3 overview video.
You can monitor data ingestion into your @ClickHouseDB cluster: Rows, bytes and insert queries over time, uncompressed ingest vs bytes written to disk, write amplification from merges, and per-table ingestion.
See the whole @ClickHouseDB cluster at a glance
An interactive map of nodes, shards, replicas and Keeper quorum — live replication links, health states and per-node metrics, drawn from system tables.
KIMI K3.
> 2.8T params
> MoE with 896 experts, 16 active.
> 1M context window.
> beats Opus 4.8 MAX on real world knowledge work (GDPval)
> SOTA 91.2 on BrowseComp w zero context tricks.