diff --git a/CHANGELOG.md b/CHANGELOG.md index 0723f1a..3268741 100644 --- a/CHANGELOG.md +++ b/CHANGELOG.md @@ -20,13 +20,24 @@ is only released once its **exit criterion** is met, not when its code lands. | v0.13.0 | — | The whole system stopped and started from one command; Qwen3Guard judges, on a GPU and CPUs together | released | | v0.14.0 | — | Knowledge bases: documents from people's own machines, kept apart per tenant, answered from | released (v0.14.1: adapters, the tree over the API, the firewall on every machine; v0.14.2: pictures from ComfyUI; v0.14.3: 3D models as jobs; v0.14.4: any ComfyUI workflow as a job, presets, licences; v0.14.5: picture and 3D models that add themselves; v0.14.6: setup that explains every part and lets it be chosen, the gateway on all three HA machines; v0.14.7: the judge sized for its machine; v0.14.8: sign-in with Active Directory, several keys for one outside provider; v0.14.9: sign-in set up from the console, a node's console that shows the machine) | | v0.15.0 | — | The company's MCP tools and VS Code through the gateway; guardrails that ask outside services and put redacted values back; answers to similar questions | released (v0.15.1: Strata, a 125B model of experts on one card; a second mirror kept in step; pickled weights refused; a served model's context changed; v0.15.2: a served model moved to more cards; v0.15.3: a model stopped and kept, or removed with its files) | -| v0.16.0 | — | Weights pinned to a commit and checked against what Hugging Face published; documents embedded by the gateway; Strata offered where models are chosen; cluster upgrades that say how far they are | released (v0.16.1: the mirror's Strata tab, Strata's ready-made packs; v0.16.2: standbys that answer clients, FreeToken, a model fetched once per node; v0.16.3: replicas per profile, a served model's settings on the mirror page, the CPU's speed before serving) | +| v0.16.0 | — | Weights pinned to a commit and checked against what Hugging Face published; documents embedded by the gateway; Strata offered where models are chosen; cluster upgrades that say how far they are | released (v0.16.1: the mirror's Strata tab, Strata's ready-made packs; v0.16.2: standbys that answer clients, FreeToken, a model fetched once per node; v0.16.3: replicas per profile, a served model's settings on the mirror page, the CPU's speed before serving; v0.16.4: Slurm beside llmgw, a model customised from the portal, a GPU node without a driver reported) | | v1.0.0 | 5 | What phase 5 has left: precise KV routing proven on vLLM, the block cost picker, defragmentation across NVLink domains, prefill/decode disaggregation | planned | Until v1.0.0 the minor version is the breaking one: configuration formats and Go APIs change between minors without a deprecation period. -## Unreleased +## v0.16.4 — 2026-11-19 + +Slurm beside llmgw for training and batch work — its users, shared files, containers, accounting priced in llmgw's +books — a model customised from the portal with a file and a button, an organization's admin, and a GPU node whose +driver was never installed told apart from a CPU node. + +**Upgrading.** +- Four migrations (0052–0055: Slurm's users and jobs, held capacity, the portal's training jobs), applied by the + first control plane to start on this version; nothing to do by hand. `llmgw ctl upgrade v0.16.4`: the control + planes carry Slurm, the training page, the admin role and the prices; the agents carry Slurm's node part, the + containers, and the reading of the PCI bus for a card without a driver, without which such a node still joins + as a CPU node. Built with Go 1.27. ### Added @@ -113,6 +124,14 @@ APIs change between minors without a deprecation period. ### Fixed +- **A GPU node with no driver joined quietly as a CPU node.** The agent found + GPUs through `nvidia-smi` alone, so a machine whose driver was never + installed looked exactly like one with no card. The PCI bus is read as + well: a card there with no driver fails the checks before the agent is + installed, with the install and the reboot named (and a driver installed + but not loaded — Secure Boot, nouveau — told apart from a missing one), and + a node already in shows the warning in `ctl nodes` and the console until + the driver is in and its agent restarted. - **The serve forms start at the longest context that fits.** The mirror page's form began every GGUF model at 4096 tokens, and the console's at 4096 or the plan's figure, whatever the model was trained for. Both now diff --git a/README.md b/README.md index a810498..3156068 100644 --- a/README.md +++ b/README.md @@ -42,7 +42,7 @@ Releases track the roadmap phases; see [CHANGELOG.md](CHANGELOG.md). | v0.13.0 | — | The whole system stopped and started from one command; Qwen3Guard judges, on a GPU and CPUs together | released | | v0.14.0 | — | Knowledge bases: documents from people's own machines, kept apart per tenant, answered from | released (v0.14.1: adapters, the tree over the API, the firewall on every machine; v0.14.2: pictures from ComfyUI; v0.14.3: 3D models as jobs; v0.14.4: any ComfyUI workflow as a job, presets, licences; v0.14.5: picture and 3D models that add themselves; v0.14.6: setup that explains every part and lets it be chosen, the gateway on all three HA machines; v0.14.7: the judge sized for its machine; v0.14.8: sign-in with Active Directory, several keys for one outside provider; v0.14.9: sign-in set up from the console, a node's console that shows the machine) | | v0.15.0 | — | The company's MCP tools and VS Code through the gateway; guardrails that ask outside services and put redacted values back; answers to similar questions | released (v0.15.1: Strata, a 125B model of experts on one card; a second mirror kept in step; pickled weights refused; a served model's context changed; v0.15.2: a served model moved to more cards; v0.15.3: a model stopped and kept, or removed with its files) | -| v0.16.0 | — | Weights pinned to a commit and checked against what Hugging Face published; documents embedded by the gateway; Strata offered where models are chosen; cluster upgrades that say how far they are | released (v0.16.1: the mirror's Strata tab, Strata's ready-made packs; v0.16.2: standbys that answer clients, FreeToken, a model fetched once per node; v0.16.3: replicas per profile, a served model's settings on the mirror page, the CPU's speed before serving) | +| v0.16.0 | — | Weights pinned to a commit and checked against what Hugging Face published; documents embedded by the gateway; Strata offered where models are chosen; cluster upgrades that say how far they are | released (v0.16.1: the mirror's Strata tab, Strata's ready-made packs; v0.16.2: standbys that answer clients, FreeToken, a model fetched once per node; v0.16.3: replicas per profile, a served model's settings on the mirror page, the CPU's speed before serving; v0.16.4: Slurm beside llmgw, a model customised from the portal, a GPU node without a driver reported) | | v1.0.0 | 5 | What phase 5 has left: precise KV routing proven on vLLM, the block cost picker, defragmentation across NVLink domains, prefill/decode disaggregation | planned | As of v0.6.0 a cluster runs unattended: agents report their hardware, the