Commit Graph

5 Commits

Author SHA1 Message Date
Deeman
cf65fa16b6 refactor(infra): consolidate tool installs in setup, strip bootstrap to essentials
- setup_server.sh: add git/curl/ca-certificates apt install, add uv install
  as service user, fix SSH config write (root + chown vs sudo heredoc), remove
  noise log lines after set -e makes them redundant
- bootstrap_supervisor.sh: remove all tool installs (apt, uv, sops, age) —
  setup_server.sh is now the single source of truth; strip to ~45 lines:
  age-key check, clone/fetch, tag checkout, decrypt, uv sync, systemd enable
- readme.md: update step 1 and step 3 descriptions

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
2026-02-26 22:25:31 +01:00
Deeman
0317cb885f feat(infra): use beanflows_service for supervisor
- materia-supervisor.service: User=root → User=beanflows_service,
  add PATH so uv (~/.local/bin) is found without a login shell
- setup_server.sh: full rewrite — creates beanflows_service (nologin),
  generates SSH deploy key + age keypair as service user at XDG path
  (~/.config/sops/age/keys.txt), installs age/sops/rclone as root,
  prints both public keys + numbered next-step instructions
- bootstrap_supervisor.sh: full rewrite — removes GITLAB_READ_TOKEN
  requirement, clones via SSH as service user, installs uv as service
  user, decrypts with SOPS auto-discovery, uv sync as service user,
  systemctl as root
- web/deploy.sh: remove self-contained sops/age install + keypair
  generation; replace with simple sops check (exit if missing) and
  SOPS auto-discovery decrypt (no explicit key file needed)
- infra/readme.md: update architecture diagram for beanflows_service
  paths, update setup steps to match new scripts

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
2026-02-26 21:33:31 +01:00
Deeman
518b50d0f5 docs(claude+infra): expand CLAUDE.md + infra/readme.md for full architecture
CLAUDE.md additions:
- List all 6 extractor packages + extract_core
- Full data flow with all sources + dual-DuckDB
- Foundation-as-ontology: dim_commodity conforms cross-source identifiers
- Two-DuckDB architecture explanation (why not serving.duckdb)
- Extraction pattern: one-package-per-source, state SQLite, adding new source
- Supervisor: croniter scheduling, topological waves, tag-based deploy
- CI/CD: pull-based via git tags, no SSH
- Secrets management: SOPS+age section, file table, server key workflow
- uv workspace management section
- Remove Pulumi ESC references; update env vars table

infra/readme.md:
- Update architecture diagram (add analytics.duckdb, age-key.txt)
- Rewrite setup flow: setup_server.sh → add key to SOPS → bootstrap
- Secrets management section with file table
- Deploy model: pull-based (no SSH/CI credentials)
- Monitoring: add supervisor status + extraction state DB query

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
2026-02-26 12:04:55 +01:00
Deeman
c1d00dcdc4 Refactor to local-first architecture on Hetzner NVMe
Remove distributed R2/Iceberg/SSH pipeline architecture in favor of
local subprocess execution with NVMe storage. Landing data backed up
to R2 via rclone timer.

- Strip Iceberg catalog, httpfs, boto3, paramiko, prefect, pyarrow
- Pipelines run via subprocess.run() with bounded timeouts
- Extract writes to {LANDING_DIR}/psd/{year}/{month}/{etag}.csv.gzip
- SQLMesh reads LANDING_DIR variable, writes to DUCKDB_PATH
- Delete unused provider stubs (ovh, scaleway, oracle)
- Add rclone systemd timer for R2 backup every 6h
- Update supervisor to run pipelines with env vars

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
2026-02-18 19:50:19 +01:00
Deeman
6d4377ccf9 cleanup and prefect service setup 2026-02-04 22:24:55 +01:00