docs: automated documentation update and phase 8 archiving

This commit is contained in:
oib
2026-02-26 19:17:53 +01:00
parent 825f157749
commit 88971a3216
20 changed files with 4327 additions and 1085 deletions

View File

@@ -1303,8 +1303,6 @@ Prepare for the v0.1 public release with comprehensive packaging, deployment, an
|-----------|-------------|----------|--------|
| PyPI packages | Q2 2026 | High | 🔄 In Progress |
| npm packages | Q2 2026 | High | 🔄 In Progress |
| Docker Compose setup | Q2 2026 | High | 🔄 Planned |
| Security audit | Q3 2026 | Critical | 🔄 Planned |
| Prebuilt binaries | Q2 2026 | Medium | 🔄 Planned |
| Documentation | Q2 2026 | High | 🔄 In Progress |

View File

@@ -138,6 +138,48 @@ server {
ssh aitbc-cascade # Direct SSH to container
```
**GPU Access**: No GPU passthrough. All GPU workloads must run on localhost (windsurf host), not inside incus containers.
**Host Proxies (for localhost GPU clients)**
- `127.0.0.1:18000` → container `127.0.0.1:8000` (coordinator/marketplace API)
- Use this to submit offers/bids/contracts/mining requests from localhost GPU miners/dev clients.
## Container: aitbc1 (10.1.223.40) — New Dev Server
### Access
```bash
ssh aitbc1-cascade # Direct SSH to aitbc1 container (incus)
```
### Notes
- Purpose: secondary AITBC dev environment (incus container)
- Host: 10.1.223.40 (Debian trixie), accessible via new SSH alias `aitbc1-cascade`
- Proxy device: incus proxy on host maps 127.0.0.1:18001 → 127.0.0.1:8000 inside container
- AppArmor profile: unconfined (incus raw.lxc)
- Use same deployment patterns as `aitbc` (nginx + services) once provisioned
- **GPU Access**: None. Run GPU-dependent tasks on localhost (windsurf host) only.
**Host Proxies (for localhost GPU clients)**
- `127.0.0.1:18001` → container `127.0.0.1:8000` (coordinator/marketplace API)
- Use this to hit the second marketplace/coordinator from localhost GPU miners/dev clients.
- (Optional) Expose marketplace frontend for aitbc1 via an additional proxy/port if needed for UI tests.
- Health check suggestion: `curl -s http://127.0.0.1:18001/v1/health`
**Localhost dual-miner/dual-client test (shared GPU)**
- Run two miners on localhost (GPU shared), targeting each marketplace:
- Miner A → `http://127.0.0.1:18000`
- Miner B → `http://127.0.0.1:18001`
- Run two clients on localhost for bids/contracts/Ollama answers:
- Client 1 → `http://127.0.0.1:18000`
- Client 2 → `http://127.0.0.1:18001`
- Use a shared dev chain so both marketplaces see the same on-chain events.
- Example commands (adjust to your scripts/flags):
- `miner --id miner-A --gpu 0 --api http://127.0.0.1:18000`
- `miner --id miner-B --gpu 0 --api http://127.0.0.1:18001`
- `client --id client-1 --api http://127.0.0.1:18000 --ollama-model <model>`
- `client --id client-2 --api http://127.0.0.1:18001 --ollama-model <model>`
### Services
| Service | Port | Process | Python Version | Public URL |