feat(kb): faza 3 krok 5 — cykliczny ingest (systemd timer) + alerting
documents-ingest-cyclic (jobs/documents-ingest/src/documents_ingest/cyclic_ingest.py): orkiestruje paperless_adapter -> chunk_embed -> summarize(--backend anthropic, claude-haiku-4-5) -> summarize(--embed-summaries) bez zmian w samych jobach. Ollama@SOLARIA (availability_target: medium) jest tolerowana offline: probe GET /api/tags przed obu etapami embed, brak -> pominięcie, nie fail (oba embed passy idempotentne, nadrobią się na kolejnym ticku). Czwarty etap (embed-summaries) dopisany ponad plan §7.1 (który wymieniał tylko 3 kroki) — bez niego nowe streszczenia miałyby embedding=NULL i byłyby niewidoczne dla cascade_query (bramka kroku 4, WHERE embedding IS NOT NULL); potwierdzone z Oskarem. Predykaty pass/fail każdego etapu 1:1 z exit-checkiem danego joba (chunks_errors, llm_errors, stats-balance itd.) — etapy izolowane, nie fail-fast (wcześniejszy fail nie blokuje kolejnych, tak jak joby izolują błędy per wiersz). Metryki .prom (atomowy zapis, last_success_timestamp trzymany z poprzedniego pliku przy failu) do /opt/homelab/state/node-exporter/kb-ingest.prom. 36 nowych testów (202/202 pakietu). systemd (jobs/documents-ingest/systemd/): pierwszy systemd-timer w repo — kb-ingest.timer (OnCalendar=*-*-* 03:30, Persistent=true, plan §7.1) + kb-ingest.service (host-level, User oskar, EnvironmentFile /opt/homelab/kb/.env) + kb-ingest-run.sh (log per-run do /opt/homelab/logs/kb-ingest/, konwencja repo). Instalacja i sekrety udokumentowane w README (Faza 3 krok 5) — instalacja na PIHA dopiero po merge. fleet-prometheus (rules/kb-ingest.yml): KbIngestStale (>172800s od last_success, critical) + KbEmbedBacklogGrowing (backlog>0 przez 72h, warning) — dostawa istniejącym torem brain-watchdog->Telegram, bez Alertmanagera (konwencja liveness.yml). node_exporter: owner_node vps -> per-host (service.yaml) + wpis + override (--collector.textfile.directory, bez nowego mountu — czyta przez istniejący /:/host:ro) + topology.yaml dla PIHA. Domyka pozycję z docs/backlog.md "stability-agent / node_exporter owner_node single, biegaja wielomiejscowo -> per-host" (połowę — node_exporter; stability-agent zostaje osobnym follow-upem) w ramach paczki B inwentaryzacji monitoringu dla PIHA. Test end-to-end na żywo na PIHA (2× --apply, po potwierdzeniu z Oskarem): pierwszy run złapał 5 dokumentów faktycznie nowych w Paperless (nieoczekiwane, niezwiązane z tym taskiem) -> 82 nowe chunki (2 ocr_junk), 5 nowych streszczeń, 5 embeddingów streszczeń, 0 błędów, metryki zapisane. Drugi run: pełna idempotencja, wszystko 0. ANTHROPIC_API_KEY dodany przez Oskara ręcznie do /opt/homelab/kb/.env (nigdy nie logowany/generowany). Co dalej: prawdziwa instalacja systemd (services.yaml już przygotowany, po merge), zdecydowanie czy stability-agent też idzie na per-host przy okazji. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
This commit is contained in:
oskar
parent
4083f18673
commit
cb8b60fee6
15
hosts/piha/runtime/node_exporter/docker-compose.override.yml
Normal file
15
hosts/piha/runtime/node_exporter/docker-compose.override.yml
Normal file
|
|
@ -0,0 +1,15 @@
|
||||||
|
# PIHA-specific override for node_exporter.
|
||||||
|
#
|
||||||
|
# WHY: KB module 5 phase 3 step 5 (docs/kb/modules/05-faza3-plan.md §7.2) needs the
|
||||||
|
# textfile collector so kb-ingest's systemd timer can publish
|
||||||
|
# kb_ingest_last_success_timestamp / kb_ingest_last_exit_code / kb_ingest_embed_backlog
|
||||||
|
# etc. for fleet-prometheus to scrape and alert on.
|
||||||
|
#
|
||||||
|
# NO NEW MOUNT: the base compose already binds `/:/host:ro,rslave` — the textfile
|
||||||
|
# directory is read through that existing mount, not a new volume. Command lists are
|
||||||
|
# replaced wholesale by Compose overrides (not merged), so both flags are restated here.
|
||||||
|
services:
|
||||||
|
node_exporter:
|
||||||
|
command:
|
||||||
|
- '--path.rootfs=/host'
|
||||||
|
- '--collector.textfile.directory=/host/opt/homelab/state/node-exporter'
|
||||||
|
|
@ -74,6 +74,20 @@ services:
|
||||||
# .env (TAILSCALE_BIND_IP) lives alongside the compose file; stateless, no data path
|
# .env (TAILSCALE_BIND_IP) lives alongside the compose file; stateless, no data path
|
||||||
config_path: services/llm-gateway
|
config_path: services/llm-gateway
|
||||||
|
|
||||||
|
node_exporter:
|
||||||
|
role: metrics-exporter
|
||||||
|
deployment_model: docker-compose
|
||||||
|
exposure: local-only
|
||||||
|
offline_required: true
|
||||||
|
depends_on:
|
||||||
|
local: []
|
||||||
|
external: []
|
||||||
|
runtime:
|
||||||
|
# textfile collector reads /opt/homelab/state/node-exporter (module 5 phase 3 step 5,
|
||||||
|
# docs/kb/modules/05-faza3-plan.md §7.2 — kb-ingest.prom) via the existing /:/host:ro
|
||||||
|
# mount, see hosts/piha/runtime/node_exporter/docker-compose.override.yml.
|
||||||
|
data_path: /opt/homelab/state/node-exporter
|
||||||
|
|
||||||
kb-postgres:
|
kb-postgres:
|
||||||
role: kb-database # KB spine: Postgres 16 + pgvector (always-on)
|
role: kb-database # KB spine: Postgres 16 + pgvector (always-on)
|
||||||
deployment_model: docker-compose
|
deployment_model: docker-compose
|
||||||
|
|
|
||||||
|
|
@ -49,6 +49,7 @@ nodes:
|
||||||
- vikunja # Task management (vikunja + postgres), public via npm
|
- vikunja # Task management (vikunja + postgres), public via npm
|
||||||
- kb-postgres # KB spine: Postgres 16 + pgvector, port 5433 (always-on)
|
- kb-postgres # KB spine: Postgres 16 + pgvector, port 5433 (always-on)
|
||||||
- llm-gateway # FastAPI router -> Ollama @ SOLARIA (Tailscale-internal :8080)
|
- llm-gateway # FastAPI router -> Ollama @ SOLARIA (Tailscale-internal :8080)
|
||||||
|
- node_exporter # per-host (also vps) — textfile collector for kb-ingest metrics
|
||||||
|
|
||||||
solaria:
|
solaria:
|
||||||
roles:
|
roles:
|
||||||
|
|
|
||||||
|
|
@ -535,3 +535,111 @@ cd jobs/documents-ingest && pytest
|
||||||
all envelopes, cascade stage-1-narrows-stage-2, an envelope whose summary exists but has no
|
all envelopes, cascade stage-1-narrows-stage-2, an envelope whose summary exists but has no
|
||||||
active chunks, N larger than the number of summarized envelopes, the no-summaries
|
active chunks, N larger than the number of summarized envelopes, the no-summaries
|
||||||
short-circuit (stage 2 never queried), and both query entry points embedding exactly once.
|
short-circuit (stage 2 never queried), and both query entry points embedding exactly once.
|
||||||
|
|
||||||
|
---
|
||||||
|
|
||||||
|
## Phase 3 step 5 — cyclic ingest (`documents-ingest-cyclic`) + systemd timer
|
||||||
|
|
||||||
|
Module 5, phase 3, plan step 5 (`docs/kb/modules/05-faza3-plan.md`, §7). Orchestrates one
|
||||||
|
run of the recurring ingest pipeline: `paperless_adapter.run()` (new `source='paperless'`
|
||||||
|
envelopes) → `chunk_embed.run()` (new `document_chunk` rows) →
|
||||||
|
`summarize.run_summarize(backend='anthropic')` (new `document_summary` rows,
|
||||||
|
`model='claude-haiku-4-5'` — plan §2 decision 3) → `summarize.run_embed_summaries()`
|
||||||
|
(embeds those summaries). All four are the same job functions used elsewhere in this
|
||||||
|
package, called directly — no changes to `paperless_adapter.py` / `chunk_embed.py` /
|
||||||
|
`summarize.py`, no new CLI flags on them.
|
||||||
|
|
||||||
|
### Ollama-offline tolerance
|
||||||
|
|
||||||
|
SOLARIA has `availability_target: medium` (planned power-off, plan §1.3). The wrapper
|
||||||
|
probes `GET {OLLAMA_URL}/api/tags` before the two embed stages (chunk embedding, summary
|
||||||
|
embedding); unreachable means **skip, not fail** — both embed passes are idempotent, so
|
||||||
|
new chunks/summaries left unembedded this tick are picked up whole on the next one. A
|
||||||
|
growing backlog is what `kb_ingest_embed_backlog` + the `KbEmbedBacklogGrowing` alert are
|
||||||
|
for, not this wrapper's exit code.
|
||||||
|
|
||||||
|
Anything else failing **is** a hard failure: Paperless unreachable, a DB error, a non-zero
|
||||||
|
job error counter, a broken stats-balance invariant, the Anthropic API failing. Each
|
||||||
|
stage's pass/fail predicate mirrors that job's own `main()` exit check 1:1 (see
|
||||||
|
`cyclic_ingest.py`'s module docstring). Stages are isolated, not fail-fast — an earlier
|
||||||
|
stage failing never skips a later one, mirroring the per-row isolation the underlying jobs
|
||||||
|
already use.
|
||||||
|
|
||||||
|
### Usage
|
||||||
|
|
||||||
|
```bash
|
||||||
|
# Dry run (default) — same idempotent counting as every other job in this family, no writes:
|
||||||
|
documents-ingest-cyclic --dsn postgresql://kb:<pw>@localhost:5433/kb \
|
||||||
|
--paperless-token <token> --anthropic-api-key <key>
|
||||||
|
|
||||||
|
# Real run (what the timer invokes):
|
||||||
|
documents-ingest-cyclic --dsn ... --paperless-token ... --anthropic-api-key ... --apply
|
||||||
|
```
|
||||||
|
|
||||||
|
`--dsn`/`--paperless-token`/`--anthropic-api-key` also read from `KB_DSN` /
|
||||||
|
`PAPERLESS_API_TOKEN` / `ANTHROPIC_API_KEY` env vars — never logged. `--ollama-url`
|
||||||
|
defaults to `http://solaria:11434` (this wrapper always runs on PIHA, unlike
|
||||||
|
`chunk_embed`/`summarize`'s own CLI defaults which assume co-location with Ollama).
|
||||||
|
|
||||||
|
### Metrics (Prometheus textfile collector)
|
||||||
|
|
||||||
|
Every run — success or failure — writes `--prom-path`
|
||||||
|
(default `/opt/homelab/state/node-exporter/kb-ingest.prom`) atomically (tmp + rename):
|
||||||
|
|
||||||
|
| Metric | Meaning |
|
||||||
|
|---|---|
|
||||||
|
| `kb_ingest_last_run_timestamp` | Unix ts of the last run, success or failure |
|
||||||
|
| `kb_ingest_last_success_timestamp` | Unix ts of the last run with no hard failure — carried forward from the previous file on a failing run, never reset to 0/now |
|
||||||
|
| `kb_ingest_last_exit_code` | 0 or 1 |
|
||||||
|
| `kb_ingest_documents_inserted` | New envelope rows this run |
|
||||||
|
| `kb_ingest_chunks_inserted` | New document_chunk rows this run (0 if the embed stage was skipped) |
|
||||||
|
| `kb_ingest_summaries_inserted` | New document_summary rows this run |
|
||||||
|
| `kb_ingest_embed_skipped` | 1 if Ollama was unreachable this run (both embed stages skipped), 0 otherwise |
|
||||||
|
| `kb_ingest_embed_backlog` | Active chunks (`excluded_reason IS NULL`) still missing an embedding |
|
||||||
|
|
||||||
|
Scraped by fleet-prometheus via node_exporter's textfile collector on PIHA
|
||||||
|
(`hosts/piha/runtime/node_exporter/docker-compose.override.yml`); alert rules in
|
||||||
|
`services/fleet-prometheus/rules/kb-ingest.yml`.
|
||||||
|
|
||||||
|
### Install (PIHA)
|
||||||
|
|
||||||
|
1. Dedicated venv (per plan §7.1 — not the ad hoc rsync-to-`/tmp` pattern used for the
|
||||||
|
one-shot jobs elsewhere in this README; this is a permanent, recurring installation):
|
||||||
|
```bash
|
||||||
|
python3 -m venv /opt/homelab/kb/venv
|
||||||
|
/opt/homelab/kb/venv/bin/pip install -e packages/kb-mail -e jobs/documents-ingest
|
||||||
|
```
|
||||||
|
(run from a checkout of this repo on PIHA — the checkout is used as an install source
|
||||||
|
only, per CLAUDE.md's "deploy-only" rule; no development happens there).
|
||||||
|
2. Secrets in `/opt/homelab/kb/.env` (already holds `PAPERLESS_API_TOKEN`; add
|
||||||
|
`KB_DSN=postgresql://kb:<pw>@localhost:5433/kb` and `ANTHROPIC_API_KEY=<key>`),
|
||||||
|
`chmod 600`, never in Git.
|
||||||
|
3. Copy `jobs/documents-ingest/systemd/kb-ingest-run.sh` to `/opt/homelab/kb/` and
|
||||||
|
`chmod +x` it.
|
||||||
|
4. Copy (or symlink) `kb-ingest.service` and `kb-ingest.timer` to `/etc/systemd/system/`,
|
||||||
|
then:
|
||||||
|
```bash
|
||||||
|
systemctl daemon-reload
|
||||||
|
systemctl enable --now kb-ingest.timer
|
||||||
|
```
|
||||||
|
5. Verify: `systemctl list-timers kb-ingest.timer`, `journalctl -u kb-ingest.service`,
|
||||||
|
`/opt/homelab/logs/kb-ingest/run-YYYYMMDD.log`, and
|
||||||
|
`/opt/homelab/state/node-exporter/kb-ingest.prom` after the first run (manual
|
||||||
|
`systemctl start kb-ingest.service` to trigger one immediately without waiting for
|
||||||
|
03:30).
|
||||||
|
|
||||||
|
### Tests
|
||||||
|
|
||||||
|
```bash
|
||||||
|
pip install -e packages/kb-mail/
|
||||||
|
pip install -e jobs/documents-ingest/
|
||||||
|
cd jobs/documents-ingest && pytest
|
||||||
|
```
|
||||||
|
|
||||||
|
`tests/test_cyclic_ingest.py` — pure unit tests, no DB or real HTTP/Ollama/Anthropic; every
|
||||||
|
stage function and the Ollama probe are monkeypatched. Covers: each stage's failure
|
||||||
|
predicate (pinned against its source job's own exit check), the Ollama-down skip path
|
||||||
|
(chunk_embed/embed_summaries never even called), stage isolation (an earlier stage failing
|
||||||
|
never skips a later one, whether via a failed predicate or a raised exception), `.prom`
|
||||||
|
rendering, atomic write, `last_success_timestamp` carry-forward across a failing run, and
|
||||||
|
`main()`'s CLI guardrails + exit-code propagation.
|
||||||
|
|
|
||||||
|
|
@ -20,6 +20,7 @@ documents-ingest = "documents_ingest.extractor:main"
|
||||||
documents-ingest-paperless = "documents_ingest.paperless_adapter:main"
|
documents-ingest-paperless = "documents_ingest.paperless_adapter:main"
|
||||||
documents-ingest-embed = "documents_ingest.chunk_embed:main"
|
documents-ingest-embed = "documents_ingest.chunk_embed:main"
|
||||||
documents-ingest-summarize = "documents_ingest.summarize:main"
|
documents-ingest-summarize = "documents_ingest.summarize:main"
|
||||||
|
documents-ingest-cyclic = "documents_ingest.cyclic_ingest:main"
|
||||||
|
|
||||||
[tool.setuptools.packages.find]
|
[tool.setuptools.packages.find]
|
||||||
where = ["src"]
|
where = ["src"]
|
||||||
|
|
|
||||||
396
jobs/documents-ingest/src/documents_ingest/cyclic_ingest.py
Normal file
396
jobs/documents-ingest/src/documents_ingest/cyclic_ingest.py
Normal file
|
|
@ -0,0 +1,396 @@
|
||||||
|
"""Cyclic ingest wrapper -- module 5 phase 3, plan step 5 (docs/kb/modules/05-faza3-plan.md,
|
||||||
|
§7). Orchestrates one run of the recurring ingest pipeline for `kb-ingest.timer` on PIHA:
|
||||||
|
|
||||||
|
paperless_adapter.run() -- new source='paperless' envelopes
|
||||||
|
-> chunk_embed.run() -- new document_chunk rows (bge-m3 via Ollama@SOLARIA)
|
||||||
|
-> summarize.run_summarize(backend='anthropic') -- new document_summary rows
|
||||||
|
-> summarize.run_embed_summaries() -- embeds those summaries (bge-m3)
|
||||||
|
|
||||||
|
The two embed stages are gated on a pre-flight Ollama reachability probe
|
||||||
|
(`GET /api/tags`) because SOLARIA has `availability_target: medium` (planned power-off,
|
||||||
|
plan §1.3): "Ollama unreachable" is skipped, not failed -- new chunks/summaries are picked
|
||||||
|
up by the next tick since both embed passes are idempotent. Without the fourth stage,
|
||||||
|
new summaries would sit with `embedding IS NULL` and stay invisible to `cascade_query`
|
||||||
|
(module 5 phase 3 step 4, `retrieval.py`), which filters on `embedding IS NOT NULL` --
|
||||||
|
so the write pass and the embed pass both run every tick, gated by the same probe.
|
||||||
|
|
||||||
|
A real failure anywhere else -- Paperless unreachable, a DB error, a non-zero job error
|
||||||
|
counter, a broken stats-balance invariant, the Anthropic API failing -- is NOT tolerated:
|
||||||
|
each stage's pass/fail predicate mirrors that job's own `main()` exit-code check 1:1 (see
|
||||||
|
`_adapter_failed` / `_chunk_embed_failed` / `_summarize_failed` / `_embed_summaries_failed`
|
||||||
|
below -- duplicated deliberately, since each job's real predicate lives inline in argparse
|
||||||
|
wiring, not as an importable function; a test pins each one against its source job).
|
||||||
|
|
||||||
|
Stages are isolated, not fail-fast: an earlier stage failing doesn't skip later ones,
|
||||||
|
mirroring the per-row isolation the underlying jobs already use (a stale Paperless token
|
||||||
|
this run shouldn't also block chunk-embedding of documents already adapted in prior runs).
|
||||||
|
The wrapper's own exit code is 1 if ANY stage failed, 0 otherwise (Ollama-skips don't count).
|
||||||
|
|
||||||
|
Writes a Prometheus textfile-collector `.prom` file (plan §7.2) alongside a structured log
|
||||||
|
line; `kb_ingest_last_success_timestamp` is carried forward from any previous file when this
|
||||||
|
run fails, so a transient failure doesn't reset the `KbIngestStale` alert's clock to zero.
|
||||||
|
|
||||||
|
This module only ever calls the existing job functions -- no changes to
|
||||||
|
paperless_adapter.py / chunk_embed.py / summarize.py, and no new CLI flags on them.
|
||||||
|
"""
|
||||||
|
from __future__ import annotations
|
||||||
|
|
||||||
|
import argparse
|
||||||
|
import asyncio
|
||||||
|
import os
|
||||||
|
import re
|
||||||
|
import sys
|
||||||
|
import time
|
||||||
|
from pathlib import Path
|
||||||
|
from typing import Optional
|
||||||
|
|
||||||
|
import aiohttp
|
||||||
|
import asyncpg
|
||||||
|
import structlog
|
||||||
|
|
||||||
|
from documents_ingest import chunk_embed, paperless_adapter, summarize
|
||||||
|
|
||||||
|
_log = structlog.get_logger(__name__)
|
||||||
|
|
||||||
|
# Wrapper always runs on PIHA -- Ollama lives on SOLARIA, reached over Tailscale (plan §7.1).
|
||||||
|
DEFAULT_OLLAMA_URL = "http://solaria:11434"
|
||||||
|
# plan §2 decision 3, resolved 2026-07-17: claude-haiku-4-5 is the compilation-track model;
|
||||||
|
# cascade_query's default summary_model matches this.
|
||||||
|
DEFAULT_SUMMARY_MODEL = "claude-haiku-4-5"
|
||||||
|
DEFAULT_EMBED_MODEL = "bge-m3"
|
||||||
|
DEFAULT_OLLAMA_PROBE_TIMEOUT = 5.0
|
||||||
|
DEFAULT_PROM_PATH = Path("/opt/homelab/state/node-exporter/kb-ingest.prom")
|
||||||
|
|
||||||
|
_BACKLOG_QUERY = (
|
||||||
|
"SELECT count(*) AS n FROM document_chunk WHERE excluded_reason IS NULL AND embedding IS NULL"
|
||||||
|
)
|
||||||
|
|
||||||
|
PROM_METRIC_HELP: dict[str, tuple[str, str]] = {
|
||||||
|
"kb_ingest_last_run_timestamp": (
|
||||||
|
"gauge", "Unix timestamp of the last kb-ingest wrapper run (success or failure).",
|
||||||
|
),
|
||||||
|
"kb_ingest_last_success_timestamp": (
|
||||||
|
"gauge",
|
||||||
|
"Unix timestamp of the last kb-ingest run with no hard failure "
|
||||||
|
"(Ollama-offline skips do not count as failure).",
|
||||||
|
),
|
||||||
|
"kb_ingest_last_exit_code": ("gauge", "Exit code of the last kb-ingest wrapper run."),
|
||||||
|
"kb_ingest_documents_inserted": (
|
||||||
|
"gauge", "New envelope rows inserted by the Paperless adapter stage in the last run.",
|
||||||
|
),
|
||||||
|
"kb_ingest_chunks_inserted": (
|
||||||
|
"gauge",
|
||||||
|
"New document_chunk rows inserted by the chunk/embed stage in the last run "
|
||||||
|
"(0 if skipped -- Ollama unreachable).",
|
||||||
|
),
|
||||||
|
"kb_ingest_summaries_inserted": (
|
||||||
|
"gauge", "New document_summary rows inserted by the summarize stage in the last run.",
|
||||||
|
),
|
||||||
|
"kb_ingest_embed_skipped": (
|
||||||
|
"gauge",
|
||||||
|
"1 if the embed stages (chunk + summary) were skipped this run because "
|
||||||
|
"Ollama@SOLARIA was unreachable, 0 otherwise.",
|
||||||
|
),
|
||||||
|
"kb_ingest_embed_backlog": (
|
||||||
|
"gauge",
|
||||||
|
"Active document_chunk rows (excluded_reason IS NULL) still missing an embedding.",
|
||||||
|
),
|
||||||
|
}
|
||||||
|
|
||||||
|
_PROM_LINE_RE = re.compile(r"^(kb_ingest_\w+)\s+([0-9.eE+-]+)\s*$")
|
||||||
|
|
||||||
|
|
||||||
|
def _stage_result(stats: Optional[dict] = None, failed: bool = False, skipped: bool = False,
|
||||||
|
error: Optional[str] = None) -> dict:
|
||||||
|
return {"stats": stats, "failed": failed, "skipped": skipped, "error": error}
|
||||||
|
|
||||||
|
|
||||||
|
def _adapter_failed(stats: dict) -> bool:
|
||||||
|
"""Mirrors paperless_adapter.main()'s exit predicate."""
|
||||||
|
balanced = stats["fetched"] == stats["already_in_db"] + stats["inserted"] + stats["errors"]
|
||||||
|
return stats["errors"] > 0 or not balanced
|
||||||
|
|
||||||
|
|
||||||
|
def _chunk_embed_failed(stats: dict) -> bool:
|
||||||
|
"""Mirrors chunk_embed.main()'s exit predicate."""
|
||||||
|
balanced = (
|
||||||
|
stats["documents_fetched"] == stats["empty_content"] + stats["documents_chunked"]
|
||||||
|
and stats["chunks_total"] == (
|
||||||
|
stats["chunks_already_embedded"] + stats["chunks_inserted"]
|
||||||
|
+ stats["chunks_junk_flagged"] + stats["chunks_conflict_skipped"]
|
||||||
|
+ stats["chunks_errors"]
|
||||||
|
)
|
||||||
|
)
|
||||||
|
return stats["chunks_errors"] > 0 or stats["chunks_conflict_skipped"] > 0 or not balanced
|
||||||
|
|
||||||
|
|
||||||
|
def _summarize_failed(stats: dict) -> bool:
|
||||||
|
"""Mirrors summarize.main()'s write-mode exit predicate."""
|
||||||
|
balanced = stats["documents_fetched"] == (
|
||||||
|
stats["duplicates_skipped"] + stats["no_active_chunks"] + stats["already_summarized"]
|
||||||
|
+ stats["summarized"] + stats["llm_errors"]
|
||||||
|
)
|
||||||
|
return stats["llm_errors"] > 0 or not balanced
|
||||||
|
|
||||||
|
|
||||||
|
def _embed_summaries_failed(stats: dict) -> bool:
|
||||||
|
"""Mirrors summarize.main()'s --embed-summaries exit predicate."""
|
||||||
|
balanced = stats["embedded"] + stats["errors"] == stats["summaries_fetched"]
|
||||||
|
return stats["errors"] > 0 or not balanced
|
||||||
|
|
||||||
|
|
||||||
|
async def probe_ollama(ollama_url: str, timeout: float = DEFAULT_OLLAMA_PROBE_TIMEOUT) -> bool:
|
||||||
|
"""`GET /api/tags` reachability check (plan §7.1). Any exception or non-2xx means
|
||||||
|
"treat as offline, skip the embed stages this tick" -- never raises."""
|
||||||
|
try:
|
||||||
|
async with aiohttp.ClientSession(timeout=aiohttp.ClientTimeout(total=timeout)) as session:
|
||||||
|
async with session.get(f"{ollama_url}/api/tags") as resp:
|
||||||
|
return resp.status == 200
|
||||||
|
except Exception:
|
||||||
|
return False
|
||||||
|
|
||||||
|
|
||||||
|
async def fetch_embed_backlog(dsn: str) -> Optional[int]:
|
||||||
|
"""Read-only count for `kb_ingest_embed_backlog` (plan §7.2). Returns None (metric
|
||||||
|
omitted, not zeroed) on a DB error -- a real DB outage is already reflected by every
|
||||||
|
other stage's failure, this is purely observability."""
|
||||||
|
try:
|
||||||
|
conn = await asyncpg.connect(dsn)
|
||||||
|
except Exception:
|
||||||
|
return None
|
||||||
|
try:
|
||||||
|
row = await conn.fetchrow(_BACKLOG_QUERY)
|
||||||
|
return int(row["n"])
|
||||||
|
except Exception:
|
||||||
|
return None
|
||||||
|
finally:
|
||||||
|
await conn.close()
|
||||||
|
|
||||||
|
|
||||||
|
async def run_cyclic(
|
||||||
|
dsn: str,
|
||||||
|
paperless_url: str,
|
||||||
|
paperless_token: str,
|
||||||
|
ollama_url: str = DEFAULT_OLLAMA_URL,
|
||||||
|
anthropic_api_key: Optional[str] = None,
|
||||||
|
summary_model: str = DEFAULT_SUMMARY_MODEL,
|
||||||
|
embed_model: str = DEFAULT_EMBED_MODEL,
|
||||||
|
tags_vocab_path: Path = summarize.DEFAULT_TAGS_VOCAB_PATH,
|
||||||
|
apply: bool = False,
|
||||||
|
ollama_probe_timeout: float = DEFAULT_OLLAMA_PROBE_TIMEOUT,
|
||||||
|
) -> dict:
|
||||||
|
"""Runs all four stages, isolating failures per stage (an earlier failure never skips
|
||||||
|
a later stage). Returns a dict with one entry per stage (`adapter`, `chunk_embed`,
|
||||||
|
`summarize`, `embed_summaries`, each `_stage_result()`-shaped), plus `ollama_up`,
|
||||||
|
`embed_backlog`, and the aggregate `failed` bool the caller should exit non-zero on."""
|
||||||
|
ollama_up = await probe_ollama(ollama_url, timeout=ollama_probe_timeout)
|
||||||
|
|
||||||
|
adapter_stage = _stage_result()
|
||||||
|
try:
|
||||||
|
stats = await paperless_adapter.run(
|
||||||
|
dsn=dsn, paperless_url=paperless_url, paperless_token=paperless_token, apply=apply,
|
||||||
|
)
|
||||||
|
adapter_stage = _stage_result(stats=stats, failed=_adapter_failed(stats))
|
||||||
|
except Exception as exc:
|
||||||
|
_log.error("adapter_exception", error=str(exc))
|
||||||
|
adapter_stage = _stage_result(failed=True, error=str(exc))
|
||||||
|
|
||||||
|
chunk_stage = _stage_result(skipped=not ollama_up)
|
||||||
|
if ollama_up:
|
||||||
|
try:
|
||||||
|
stats = await chunk_embed.run(
|
||||||
|
dsn=dsn, ollama_url=ollama_url, model=embed_model, apply=apply,
|
||||||
|
)
|
||||||
|
chunk_stage = _stage_result(stats=stats, failed=_chunk_embed_failed(stats))
|
||||||
|
except Exception as exc:
|
||||||
|
_log.error("chunk_embed_exception", error=str(exc))
|
||||||
|
chunk_stage = _stage_result(failed=True, error=str(exc))
|
||||||
|
else:
|
||||||
|
_log.warning("skip.chunk_embed", reason="ollama_unreachable")
|
||||||
|
|
||||||
|
summarize_stage = _stage_result()
|
||||||
|
try:
|
||||||
|
stats = await summarize.run_summarize(
|
||||||
|
dsn=dsn,
|
||||||
|
backend_name="anthropic",
|
||||||
|
model=summary_model,
|
||||||
|
anthropic_api_key=anthropic_api_key,
|
||||||
|
tags_vocab_path=tags_vocab_path,
|
||||||
|
apply=apply,
|
||||||
|
)
|
||||||
|
summarize_stage = _stage_result(stats=stats, failed=_summarize_failed(stats))
|
||||||
|
except Exception as exc:
|
||||||
|
_log.error("summarize_exception", error=str(exc))
|
||||||
|
summarize_stage = _stage_result(failed=True, error=str(exc))
|
||||||
|
|
||||||
|
embed_summaries_stage = _stage_result(skipped=not ollama_up)
|
||||||
|
if ollama_up:
|
||||||
|
try:
|
||||||
|
stats = await summarize.run_embed_summaries(
|
||||||
|
dsn=dsn, ollama_url=ollama_url, embed_model=embed_model,
|
||||||
|
model_filter=summary_model, apply=apply,
|
||||||
|
)
|
||||||
|
embed_summaries_stage = _stage_result(stats=stats, failed=_embed_summaries_failed(stats))
|
||||||
|
except Exception as exc:
|
||||||
|
_log.error("embed_summaries_exception", error=str(exc))
|
||||||
|
embed_summaries_stage = _stage_result(failed=True, error=str(exc))
|
||||||
|
else:
|
||||||
|
_log.warning("skip.embed_summaries", reason="ollama_unreachable")
|
||||||
|
|
||||||
|
embed_backlog = await fetch_embed_backlog(dsn)
|
||||||
|
|
||||||
|
failed = any(
|
||||||
|
stage["failed"]
|
||||||
|
for stage in (adapter_stage, chunk_stage, summarize_stage, embed_summaries_stage)
|
||||||
|
)
|
||||||
|
|
||||||
|
return {
|
||||||
|
"ollama_up": ollama_up,
|
||||||
|
"adapter": adapter_stage,
|
||||||
|
"chunk_embed": chunk_stage,
|
||||||
|
"summarize": summarize_stage,
|
||||||
|
"embed_summaries": embed_summaries_stage,
|
||||||
|
"embed_backlog": embed_backlog,
|
||||||
|
"failed": failed,
|
||||||
|
}
|
||||||
|
|
||||||
|
|
||||||
|
def _read_prev_metric(path: Path, name: str) -> Optional[float]:
|
||||||
|
if not path.exists():
|
||||||
|
return None
|
||||||
|
try:
|
||||||
|
text = path.read_text()
|
||||||
|
except OSError:
|
||||||
|
return None
|
||||||
|
for line in text.splitlines():
|
||||||
|
m = _PROM_LINE_RE.match(line)
|
||||||
|
if m and m.group(1) == name:
|
||||||
|
return float(m.group(2))
|
||||||
|
return None
|
||||||
|
|
||||||
|
|
||||||
|
def build_metrics(result: dict, now_ts: float, prom_path: Path) -> dict:
|
||||||
|
"""Turns a `run_cyclic()` result into the flat name->value map `render_prom()` expects.
|
||||||
|
`kb_ingest_last_success_timestamp` carries forward the previous file's value on failure
|
||||||
|
-- never resets to 0/now, or `KbIngestStale` would flap on every transient error."""
|
||||||
|
failed = result["failed"]
|
||||||
|
prev_success = _read_prev_metric(prom_path, "kb_ingest_last_success_timestamp")
|
||||||
|
last_success = now_ts if not failed else (prev_success if prev_success is not None else 0.0)
|
||||||
|
|
||||||
|
metrics = {
|
||||||
|
"kb_ingest_last_run_timestamp": now_ts,
|
||||||
|
"kb_ingest_last_success_timestamp": last_success,
|
||||||
|
"kb_ingest_last_exit_code": 1 if failed else 0,
|
||||||
|
"kb_ingest_documents_inserted": (result["adapter"]["stats"] or {}).get("inserted", 0),
|
||||||
|
"kb_ingest_chunks_inserted": (result["chunk_embed"]["stats"] or {}).get("chunks_inserted", 0),
|
||||||
|
"kb_ingest_summaries_inserted": (result["summarize"]["stats"] or {}).get("summarized", 0),
|
||||||
|
"kb_ingest_embed_skipped": 0 if result["ollama_up"] else 1,
|
||||||
|
}
|
||||||
|
if result["embed_backlog"] is not None:
|
||||||
|
metrics["kb_ingest_embed_backlog"] = result["embed_backlog"]
|
||||||
|
return metrics
|
||||||
|
|
||||||
|
|
||||||
|
def render_prom(metrics: dict) -> str:
|
||||||
|
lines = []
|
||||||
|
for name, (mtype, help_text) in PROM_METRIC_HELP.items():
|
||||||
|
if name not in metrics:
|
||||||
|
continue
|
||||||
|
lines.append(f"# HELP {name} {help_text}")
|
||||||
|
lines.append(f"# TYPE {name} {mtype}")
|
||||||
|
lines.append(f"{name} {metrics[name]}")
|
||||||
|
return "\n".join(lines) + "\n"
|
||||||
|
|
||||||
|
|
||||||
|
def write_prom_atomic(path: Path, content: str) -> None:
|
||||||
|
"""tmp-write + rename (plan §7.2) -- node_exporter's textfile collector never observes
|
||||||
|
a half-written file."""
|
||||||
|
path.parent.mkdir(parents=True, exist_ok=True)
|
||||||
|
tmp_path = path.with_name(f"{path.name}.tmp{os.getpid()}")
|
||||||
|
tmp_path.write_text(content)
|
||||||
|
tmp_path.replace(path)
|
||||||
|
|
||||||
|
|
||||||
|
def main() -> None:
|
||||||
|
parser = argparse.ArgumentParser(
|
||||||
|
description="Cyclic KB ingest wrapper: paperless_adapter -> chunk_embed -> "
|
||||||
|
"summarize -> embed-summaries (module 5 phase 3, plan §7). Tolerates "
|
||||||
|
"Ollama@SOLARIA being offline; does not tolerate anything else failing."
|
||||||
|
)
|
||||||
|
parser.add_argument("--dsn", default=os.environ.get("KB_DSN"),
|
||||||
|
help="asyncpg DSN for kb-postgres (or set KB_DSN env var)")
|
||||||
|
parser.add_argument("--paperless-url",
|
||||||
|
default=os.environ.get("PAPERLESS_URL", paperless_adapter.DEFAULT_PAPERLESS_URL),
|
||||||
|
help="Paperless base URL (or set PAPERLESS_URL env var)")
|
||||||
|
parser.add_argument("--paperless-token", default=os.environ.get("PAPERLESS_API_TOKEN"),
|
||||||
|
help="Paperless API token (or set PAPERLESS_API_TOKEN env var) -- never logged")
|
||||||
|
parser.add_argument("--ollama-url", default=os.environ.get("OLLAMA_URL", DEFAULT_OLLAMA_URL),
|
||||||
|
help=f"Ollama base URL (default: {DEFAULT_OLLAMA_URL}, or set OLLAMA_URL)")
|
||||||
|
parser.add_argument("--anthropic-api-key", default=os.environ.get("ANTHROPIC_API_KEY"),
|
||||||
|
help="Anthropic API key (or set ANTHROPIC_API_KEY env var) -- never logged")
|
||||||
|
parser.add_argument("--summary-model",
|
||||||
|
default=os.environ.get("KB_SUMMARY_MODEL", DEFAULT_SUMMARY_MODEL),
|
||||||
|
help=f"Summarizing model, both write+embed filter (default: {DEFAULT_SUMMARY_MODEL})")
|
||||||
|
parser.add_argument("--embed-model", default=os.environ.get("KB_EMBED_MODEL", DEFAULT_EMBED_MODEL),
|
||||||
|
help=f"Ollama embedding model, chunks+summaries (default: {DEFAULT_EMBED_MODEL})")
|
||||||
|
parser.add_argument("--tags-vocab", type=Path, default=summarize.DEFAULT_TAGS_VOCAB_PATH,
|
||||||
|
help="Path to tags-vocab.yaml")
|
||||||
|
parser.add_argument("--prom-path", type=Path,
|
||||||
|
default=Path(os.environ.get("KB_INGEST_PROM_PATH", str(DEFAULT_PROM_PATH))),
|
||||||
|
help=f"Textfile-collector output path (default: {DEFAULT_PROM_PATH})")
|
||||||
|
parser.add_argument("--ollama-probe-timeout", type=float, default=DEFAULT_OLLAMA_PROBE_TIMEOUT,
|
||||||
|
help=f"Seconds before the Ollama reachability probe gives up (default: {DEFAULT_OLLAMA_PROBE_TIMEOUT})")
|
||||||
|
parser.add_argument("--apply", action="store_true",
|
||||||
|
help="Actually write to Paperless-read/DB/call the LLM+Ollama. Default is dry-run.")
|
||||||
|
args = parser.parse_args()
|
||||||
|
|
||||||
|
if not args.dsn:
|
||||||
|
_log.error("missing_dsn", hint="pass --dsn or set KB_DSN")
|
||||||
|
sys.exit(1)
|
||||||
|
if not args.paperless_token:
|
||||||
|
_log.error("missing_paperless_token", hint="pass --paperless-token or set PAPERLESS_API_TOKEN")
|
||||||
|
sys.exit(1)
|
||||||
|
if args.apply and not args.anthropic_api_key:
|
||||||
|
_log.error("missing_anthropic_api_key", hint="pass --anthropic-api-key or set ANTHROPIC_API_KEY")
|
||||||
|
sys.exit(1)
|
||||||
|
|
||||||
|
result = asyncio.run(run_cyclic(
|
||||||
|
dsn=args.dsn,
|
||||||
|
paperless_url=args.paperless_url,
|
||||||
|
paperless_token=args.paperless_token,
|
||||||
|
ollama_url=args.ollama_url,
|
||||||
|
anthropic_api_key=args.anthropic_api_key,
|
||||||
|
summary_model=args.summary_model,
|
||||||
|
embed_model=args.embed_model,
|
||||||
|
tags_vocab_path=args.tags_vocab,
|
||||||
|
apply=args.apply,
|
||||||
|
ollama_probe_timeout=args.ollama_probe_timeout,
|
||||||
|
))
|
||||||
|
|
||||||
|
now_ts = time.time()
|
||||||
|
metrics = build_metrics(result, now_ts, args.prom_path)
|
||||||
|
write_prom_atomic(args.prom_path, render_prom(metrics))
|
||||||
|
|
||||||
|
mode = "APPLY" if args.apply else "DRY-RUN"
|
||||||
|
_log.info(
|
||||||
|
"cyclic_summary",
|
||||||
|
mode=mode,
|
||||||
|
ollama_up=result["ollama_up"],
|
||||||
|
failed=result["failed"],
|
||||||
|
adapter=result["adapter"]["stats"],
|
||||||
|
adapter_error=result["adapter"]["error"],
|
||||||
|
chunk_embed=result["chunk_embed"]["stats"],
|
||||||
|
chunk_embed_skipped=result["chunk_embed"]["skipped"],
|
||||||
|
chunk_embed_error=result["chunk_embed"]["error"],
|
||||||
|
summarize=result["summarize"]["stats"],
|
||||||
|
summarize_error=result["summarize"]["error"],
|
||||||
|
embed_summaries=result["embed_summaries"]["stats"],
|
||||||
|
embed_summaries_skipped=result["embed_summaries"]["skipped"],
|
||||||
|
embed_summaries_error=result["embed_summaries"]["error"],
|
||||||
|
embed_backlog=result["embed_backlog"],
|
||||||
|
)
|
||||||
|
sys.exit(1 if result["failed"] else 0)
|
||||||
|
|
||||||
|
|
||||||
|
if __name__ == "__main__":
|
||||||
|
main()
|
||||||
20
jobs/documents-ingest/systemd/kb-ingest-run.sh
Executable file
20
jobs/documents-ingest/systemd/kb-ingest-run.sh
Executable file
|
|
@ -0,0 +1,20 @@
|
||||||
|
#!/usr/bin/env bash
|
||||||
|
# kb-ingest-run.sh — thin launcher for kb-ingest.service (module 5 phase 3 step 5,
|
||||||
|
# docs/kb/modules/05-faza3-plan.md §7.1).
|
||||||
|
#
|
||||||
|
# All sequencing/tolerance/exit-code logic lives in documents-ingest-cyclic (Python,
|
||||||
|
# jobs/documents-ingest/src/documents_ingest/cyclic_ingest.py) — this script only computes
|
||||||
|
# a date-stamped log path and redirects, per the repo convention that a long run must never
|
||||||
|
# log only to a terminal/journal (lesson from the 2026-07 backfill that lost 4999 rows to a
|
||||||
|
# closed tmux — see the kb-piha-job-execution memory note).
|
||||||
|
#
|
||||||
|
# Installed at /opt/homelab/kb/kb-ingest-run.sh (copied from repo at deploy time), invoked
|
||||||
|
# by kb-ingest.service. `exec` at the end preserves documents-ingest-cyclic's exit code as
|
||||||
|
# this script's own, so systemd sees the real success/failure.
|
||||||
|
set -euo pipefail
|
||||||
|
|
||||||
|
LOG_DIR="/opt/homelab/logs/kb-ingest"
|
||||||
|
mkdir -p "$LOG_DIR"
|
||||||
|
LOG_FILE="${LOG_DIR}/run-$(date +%Y%m%d).log"
|
||||||
|
|
||||||
|
exec /opt/homelab/kb/venv/bin/documents-ingest-cyclic --apply >> "$LOG_FILE" 2>&1
|
||||||
19
jobs/documents-ingest/systemd/kb-ingest.service
Normal file
19
jobs/documents-ingest/systemd/kb-ingest.service
Normal file
|
|
@ -0,0 +1,19 @@
|
||||||
|
# kb-ingest.service — module 5 phase 3 step 5 (docs/kb/modules/05-faza3-plan.md §7.1).
|
||||||
|
# First systemd unit in this repo: deliberately host-level, not a container — the job
|
||||||
|
# needs simultaneous LAN (Paperless), local DB (kb-postgres@PIHA), and Tailscale (Ollama@
|
||||||
|
# SOLARIA) access; containerizing it buys nothing here.
|
||||||
|
#
|
||||||
|
# Install: copy or symlink this file (and kb-ingest.timer) to /etc/systemd/system/, then
|
||||||
|
# `systemctl daemon-reload`. See ../README.md "Cyclic ingest (systemd timer)" for the full
|
||||||
|
# install sequence (venv, .env, kb-ingest-run.sh).
|
||||||
|
[Unit]
|
||||||
|
Description=KB cyclic ingest (Paperless adapter + chunk embed + summarize)
|
||||||
|
After=network-online.target
|
||||||
|
Wants=network-online.target
|
||||||
|
|
||||||
|
[Service]
|
||||||
|
Type=oneshot
|
||||||
|
User=oskar
|
||||||
|
EnvironmentFile=/opt/homelab/kb/.env
|
||||||
|
WorkingDirectory=/opt/homelab/kb
|
||||||
|
ExecStart=/opt/homelab/kb/kb-ingest-run.sh
|
||||||
14
jobs/documents-ingest/systemd/kb-ingest.timer
Normal file
14
jobs/documents-ingest/systemd/kb-ingest.timer
Normal file
|
|
@ -0,0 +1,14 @@
|
||||||
|
# kb-ingest.timer — module 5 phase 3 step 5 (docs/kb/modules/05-faza3-plan.md §7.1).
|
||||||
|
# Daily at 03:30, not hourly: the plan fixes this schedule explicitly (low-traffic window,
|
||||||
|
# and the underlying jobs are full-corpus re-scans each tick — cheap at 186 documents
|
||||||
|
# today, but daily keeps headroom as the corpus grows). Persistent=true catches up after a
|
||||||
|
# PIHA reboot instead of silently skipping the day's run.
|
||||||
|
[Unit]
|
||||||
|
Description=Run kb-ingest.service daily (module 5 phase 3 step 5)
|
||||||
|
|
||||||
|
[Timer]
|
||||||
|
OnCalendar=*-*-* 03:30:00
|
||||||
|
Persistent=true
|
||||||
|
|
||||||
|
[Install]
|
||||||
|
WantedBy=timers.target
|
||||||
428
jobs/documents-ingest/tests/test_cyclic_ingest.py
Normal file
428
jobs/documents-ingest/tests/test_cyclic_ingest.py
Normal file
|
|
@ -0,0 +1,428 @@
|
||||||
|
"""Unit tests for the cyclic ingest wrapper — no DB, no real HTTP, no real
|
||||||
|
Ollama/Anthropic/Paperless. Every stage function (paperless_adapter.run, chunk_embed.run,
|
||||||
|
summarize.run_summarize, summarize.run_embed_summaries) and probe_ollama/fetch_embed_backlog
|
||||||
|
are monkeypatched with fakes; only cyclic_ingest's own orchestration, predicates, and
|
||||||
|
.prom rendering are under test."""
|
||||||
|
from __future__ import annotations
|
||||||
|
|
||||||
|
import sys
|
||||||
|
|
||||||
|
import pytest
|
||||||
|
|
||||||
|
from documents_ingest import cyclic_ingest
|
||||||
|
from documents_ingest.cyclic_ingest import (
|
||||||
|
PROM_METRIC_HELP,
|
||||||
|
_adapter_failed,
|
||||||
|
_chunk_embed_failed,
|
||||||
|
_embed_summaries_failed,
|
||||||
|
_read_prev_metric,
|
||||||
|
_summarize_failed,
|
||||||
|
build_metrics,
|
||||||
|
main,
|
||||||
|
probe_ollama,
|
||||||
|
render_prom,
|
||||||
|
run_cyclic,
|
||||||
|
write_prom_atomic,
|
||||||
|
)
|
||||||
|
|
||||||
|
|
||||||
|
# ---------------------------------------------------------------------------
|
||||||
|
# Per-stage failure predicates — pinned against each job's own main() exit check
|
||||||
|
# ---------------------------------------------------------------------------
|
||||||
|
|
||||||
|
class TestAdapterFailed:
|
||||||
|
def test_balanced_no_errors_passes(self):
|
||||||
|
stats = {"fetched": 3, "already_in_db": 1, "inserted": 2, "errors": 0}
|
||||||
|
assert _adapter_failed(stats) is False
|
||||||
|
|
||||||
|
def test_any_error_fails(self):
|
||||||
|
stats = {"fetched": 3, "already_in_db": 1, "inserted": 1, "errors": 1}
|
||||||
|
assert _adapter_failed(stats) is True
|
||||||
|
|
||||||
|
def test_imbalance_fails_even_with_zero_errors(self):
|
||||||
|
stats = {"fetched": 3, "already_in_db": 1, "inserted": 1, "errors": 0}
|
||||||
|
assert _adapter_failed(stats) is True
|
||||||
|
|
||||||
|
|
||||||
|
class TestChunkEmbedFailed:
|
||||||
|
def _stats(self, **overrides):
|
||||||
|
stats = {
|
||||||
|
"documents_fetched": 2, "empty_content": 0, "documents_chunked": 2,
|
||||||
|
"chunks_total": 4, "chunks_already_embedded": 0, "chunks_inserted": 4,
|
||||||
|
"chunks_junk_flagged": 0, "chunks_conflict_skipped": 0, "chunks_errors": 0,
|
||||||
|
}
|
||||||
|
stats.update(overrides)
|
||||||
|
return stats
|
||||||
|
|
||||||
|
def test_balanced_passes(self):
|
||||||
|
assert _chunk_embed_failed(self._stats()) is False
|
||||||
|
|
||||||
|
def test_chunks_errors_fails(self):
|
||||||
|
assert _chunk_embed_failed(self._stats(chunks_errors=1, chunks_inserted=3)) is True
|
||||||
|
|
||||||
|
def test_conflict_skipped_fails(self):
|
||||||
|
assert _chunk_embed_failed(self._stats(chunks_conflict_skipped=1, chunks_inserted=3)) is True
|
||||||
|
|
||||||
|
def test_doc_balance_break_fails(self):
|
||||||
|
assert _chunk_embed_failed(self._stats(documents_fetched=99)) is True
|
||||||
|
|
||||||
|
|
||||||
|
class TestSummarizeFailed:
|
||||||
|
def _stats(self, **overrides):
|
||||||
|
stats = {
|
||||||
|
"documents_fetched": 2, "duplicates_skipped": 0, "no_active_chunks": 0,
|
||||||
|
"already_summarized": 0, "summarized": 2, "llm_errors": 0,
|
||||||
|
}
|
||||||
|
stats.update(overrides)
|
||||||
|
return stats
|
||||||
|
|
||||||
|
def test_balanced_passes(self):
|
||||||
|
assert _summarize_failed(self._stats()) is False
|
||||||
|
|
||||||
|
def test_llm_errors_fails(self):
|
||||||
|
assert _summarize_failed(self._stats(llm_errors=1, summarized=1)) is True
|
||||||
|
|
||||||
|
def test_imbalance_fails(self):
|
||||||
|
assert _summarize_failed(self._stats(documents_fetched=5)) is True
|
||||||
|
|
||||||
|
|
||||||
|
class TestEmbedSummariesFailed:
|
||||||
|
def test_balanced_passes(self):
|
||||||
|
assert _embed_summaries_failed({"summaries_fetched": 2, "embedded": 2, "errors": 0}) is False
|
||||||
|
|
||||||
|
def test_errors_fails(self):
|
||||||
|
assert _embed_summaries_failed({"summaries_fetched": 2, "embedded": 1, "errors": 1}) is True
|
||||||
|
|
||||||
|
def test_imbalance_fails(self):
|
||||||
|
assert _embed_summaries_failed({"summaries_fetched": 5, "embedded": 2, "errors": 0}) is True
|
||||||
|
|
||||||
|
|
||||||
|
# ---------------------------------------------------------------------------
|
||||||
|
# probe_ollama — reachability probe, must never raise
|
||||||
|
# ---------------------------------------------------------------------------
|
||||||
|
|
||||||
|
class _FakeProbeResponse:
|
||||||
|
def __init__(self, status):
|
||||||
|
self.status = status
|
||||||
|
|
||||||
|
async def __aenter__(self):
|
||||||
|
return self
|
||||||
|
|
||||||
|
async def __aexit__(self, *exc):
|
||||||
|
return False
|
||||||
|
|
||||||
|
|
||||||
|
class _FakeProbeSession:
|
||||||
|
def __init__(self, status=200, raise_on_get=None):
|
||||||
|
self._status = status
|
||||||
|
self._raise_on_get = raise_on_get
|
||||||
|
|
||||||
|
async def __aenter__(self):
|
||||||
|
return self
|
||||||
|
|
||||||
|
async def __aexit__(self, *exc):
|
||||||
|
return False
|
||||||
|
|
||||||
|
def get(self, url):
|
||||||
|
if self._raise_on_get is not None:
|
||||||
|
raise self._raise_on_get
|
||||||
|
return _FakeProbeResponse(self._status)
|
||||||
|
|
||||||
|
|
||||||
|
class TestProbeOllama:
|
||||||
|
async def test_200_is_up(self, monkeypatch):
|
||||||
|
monkeypatch.setattr(
|
||||||
|
"documents_ingest.cyclic_ingest.aiohttp.ClientSession",
|
||||||
|
lambda *a, **kw: _FakeProbeSession(status=200),
|
||||||
|
)
|
||||||
|
assert await probe_ollama("http://solaria:11434") is True
|
||||||
|
|
||||||
|
async def test_non_200_is_down(self, monkeypatch):
|
||||||
|
monkeypatch.setattr(
|
||||||
|
"documents_ingest.cyclic_ingest.aiohttp.ClientSession",
|
||||||
|
lambda *a, **kw: _FakeProbeSession(status=500),
|
||||||
|
)
|
||||||
|
assert await probe_ollama("http://solaria:11434") is False
|
||||||
|
|
||||||
|
async def test_connection_error_is_down_not_raised(self, monkeypatch):
|
||||||
|
monkeypatch.setattr(
|
||||||
|
"documents_ingest.cyclic_ingest.aiohttp.ClientSession",
|
||||||
|
lambda *a, **kw: _FakeProbeSession(raise_on_get=ConnectionError("refused")),
|
||||||
|
)
|
||||||
|
assert await probe_ollama("http://solaria:11434") is False
|
||||||
|
|
||||||
|
|
||||||
|
# ---------------------------------------------------------------------------
|
||||||
|
# run_cyclic — sequencing, Ollama-offline tolerance, stage isolation
|
||||||
|
# ---------------------------------------------------------------------------
|
||||||
|
|
||||||
|
ADAPTER_OK = {"fetched": 1, "already_in_db": 0, "inserted": 1, "source_mail_linked": 0,
|
||||||
|
"empty_content": 0, "errors": 0}
|
||||||
|
CHUNK_OK = {"documents_fetched": 1, "empty_content": 0, "documents_chunked": 1,
|
||||||
|
"chunks_total": 1, "chunks_already_embedded": 0, "chunks_inserted": 1,
|
||||||
|
"chunks_junk_flagged": 0, "chunks_conflict_skipped": 0, "chunks_errors": 0,
|
||||||
|
"embed_calls": 1, "embed_seconds_total": 0.1}
|
||||||
|
SUMMARIZE_OK = {"documents_fetched": 1, "duplicates_skipped": 0, "no_active_chunks": 0,
|
||||||
|
"already_summarized": 0, "summarized": 1, "llm_errors": 0,
|
||||||
|
"documents_mapreduce": 0, "tags_truncated": 0}
|
||||||
|
EMBED_SUMMARIES_OK = {"summaries_fetched": 1, "embedded": 1, "errors": 0}
|
||||||
|
|
||||||
|
|
||||||
|
def _patch_stages(monkeypatch, *, ollama_up=True, adapter=None, chunk=None, summarize=None,
|
||||||
|
embed_summaries=None, backlog=0):
|
||||||
|
calls = {"adapter": 0, "chunk_embed": 0, "summarize": 0, "embed_summaries": 0}
|
||||||
|
|
||||||
|
async def _fake_probe(url, timeout=5.0):
|
||||||
|
return ollama_up
|
||||||
|
|
||||||
|
async def _fake_adapter_run(**kwargs):
|
||||||
|
calls["adapter"] += 1
|
||||||
|
return adapter if adapter is not None else dict(ADAPTER_OK)
|
||||||
|
|
||||||
|
async def _fake_chunk_run(**kwargs):
|
||||||
|
calls["chunk_embed"] += 1
|
||||||
|
if chunk is not None and isinstance(chunk, Exception):
|
||||||
|
raise chunk
|
||||||
|
return chunk if chunk is not None else dict(CHUNK_OK)
|
||||||
|
|
||||||
|
async def _fake_run_summarize(**kwargs):
|
||||||
|
calls["summarize"] += 1
|
||||||
|
if summarize is not None and isinstance(summarize, Exception):
|
||||||
|
raise summarize
|
||||||
|
return summarize if summarize is not None else dict(SUMMARIZE_OK)
|
||||||
|
|
||||||
|
async def _fake_run_embed_summaries(**kwargs):
|
||||||
|
calls["embed_summaries"] += 1
|
||||||
|
return embed_summaries if embed_summaries is not None else dict(EMBED_SUMMARIES_OK)
|
||||||
|
|
||||||
|
async def _fake_backlog(dsn):
|
||||||
|
return backlog
|
||||||
|
|
||||||
|
monkeypatch.setattr(cyclic_ingest, "probe_ollama", _fake_probe)
|
||||||
|
monkeypatch.setattr(cyclic_ingest.paperless_adapter, "run", _fake_adapter_run)
|
||||||
|
monkeypatch.setattr(cyclic_ingest.chunk_embed, "run", _fake_chunk_run)
|
||||||
|
monkeypatch.setattr(cyclic_ingest.summarize, "run_summarize", _fake_run_summarize)
|
||||||
|
monkeypatch.setattr(cyclic_ingest.summarize, "run_embed_summaries", _fake_run_embed_summaries)
|
||||||
|
monkeypatch.setattr(cyclic_ingest, "fetch_embed_backlog", _fake_backlog)
|
||||||
|
return calls
|
||||||
|
|
||||||
|
|
||||||
|
class TestRunCyclic:
|
||||||
|
async def test_all_stages_succeed_ollama_up(self, monkeypatch):
|
||||||
|
calls = _patch_stages(monkeypatch, ollama_up=True)
|
||||||
|
result = await run_cyclic(
|
||||||
|
dsn="dsn", paperless_url="http://paperless", paperless_token="tok",
|
||||||
|
anthropic_api_key="key", apply=True,
|
||||||
|
)
|
||||||
|
assert result["failed"] is False
|
||||||
|
assert result["ollama_up"] is True
|
||||||
|
assert calls == {"adapter": 1, "chunk_embed": 1, "summarize": 1, "embed_summaries": 1}
|
||||||
|
assert result["chunk_embed"]["skipped"] is False
|
||||||
|
assert result["embed_summaries"]["skipped"] is False
|
||||||
|
assert result["embed_backlog"] == 0
|
||||||
|
|
||||||
|
async def test_ollama_down_skips_embed_stages_not_a_failure(self, monkeypatch):
|
||||||
|
calls = _patch_stages(monkeypatch, ollama_up=False)
|
||||||
|
result = await run_cyclic(
|
||||||
|
dsn="dsn", paperless_url="http://paperless", paperless_token="tok",
|
||||||
|
anthropic_api_key="key", apply=True,
|
||||||
|
)
|
||||||
|
assert result["failed"] is False
|
||||||
|
assert result["ollama_up"] is False
|
||||||
|
# chunk_embed and embed_summaries must never even be invoked when Ollama is down.
|
||||||
|
assert calls["chunk_embed"] == 0
|
||||||
|
assert calls["embed_summaries"] == 0
|
||||||
|
assert calls["adapter"] == 1
|
||||||
|
assert calls["summarize"] == 1
|
||||||
|
assert result["chunk_embed"]["skipped"] is True
|
||||||
|
assert result["chunk_embed"]["stats"] is None
|
||||||
|
assert result["embed_summaries"]["skipped"] is True
|
||||||
|
|
||||||
|
async def test_adapter_failure_does_not_skip_later_stages(self, monkeypatch):
|
||||||
|
broken_adapter = dict(ADAPTER_OK, errors=1)
|
||||||
|
calls = _patch_stages(monkeypatch, ollama_up=True, adapter=broken_adapter)
|
||||||
|
result = await run_cyclic(
|
||||||
|
dsn="dsn", paperless_url="http://paperless", paperless_token="tok",
|
||||||
|
anthropic_api_key="key", apply=True,
|
||||||
|
)
|
||||||
|
assert result["failed"] is True
|
||||||
|
assert result["adapter"]["failed"] is True
|
||||||
|
# every other stage still ran despite the adapter failing this tick.
|
||||||
|
assert calls == {"adapter": 1, "chunk_embed": 1, "summarize": 1, "embed_summaries": 1}
|
||||||
|
assert result["chunk_embed"]["failed"] is False
|
||||||
|
assert result["summarize"]["failed"] is False
|
||||||
|
assert result["embed_summaries"]["failed"] is False
|
||||||
|
|
||||||
|
async def test_chunk_embed_exception_is_isolated(self, monkeypatch):
|
||||||
|
calls = _patch_stages(monkeypatch, ollama_up=True, chunk=RuntimeError("ollama 500"))
|
||||||
|
result = await run_cyclic(
|
||||||
|
dsn="dsn", paperless_url="http://paperless", paperless_token="tok",
|
||||||
|
anthropic_api_key="key", apply=True,
|
||||||
|
)
|
||||||
|
assert result["failed"] is True
|
||||||
|
assert result["chunk_embed"]["failed"] is True
|
||||||
|
assert "ollama 500" in result["chunk_embed"]["error"]
|
||||||
|
# summarize + embed_summaries still ran despite chunk_embed raising.
|
||||||
|
assert calls["summarize"] == 1
|
||||||
|
assert calls["embed_summaries"] == 1
|
||||||
|
assert result["summarize"]["failed"] is False
|
||||||
|
|
||||||
|
async def test_stats_mismatch_marks_stage_failed_without_exception(self, monkeypatch):
|
||||||
|
imbalanced_summarize = dict(SUMMARIZE_OK, documents_fetched=99)
|
||||||
|
_patch_stages(monkeypatch, ollama_up=True, summarize=imbalanced_summarize)
|
||||||
|
result = await run_cyclic(
|
||||||
|
dsn="dsn", paperless_url="http://paperless", paperless_token="tok",
|
||||||
|
anthropic_api_key="key", apply=True,
|
||||||
|
)
|
||||||
|
assert result["failed"] is True
|
||||||
|
assert result["summarize"]["failed"] is True
|
||||||
|
assert result["summarize"]["error"] is None # no exception, just a failed predicate
|
||||||
|
|
||||||
|
|
||||||
|
# ---------------------------------------------------------------------------
|
||||||
|
# .prom rendering + carry-forward
|
||||||
|
# ---------------------------------------------------------------------------
|
||||||
|
|
||||||
|
class TestRenderProm:
|
||||||
|
def test_includes_help_and_type_for_every_metric(self):
|
||||||
|
metrics = {name: 1 for name in PROM_METRIC_HELP}
|
||||||
|
content = render_prom(metrics)
|
||||||
|
for name in PROM_METRIC_HELP:
|
||||||
|
assert f"# HELP {name} " in content
|
||||||
|
assert f"# TYPE {name} gauge" in content
|
||||||
|
assert f"\n{name} 1\n" in content
|
||||||
|
|
||||||
|
def test_omits_metrics_not_present(self):
|
||||||
|
content = render_prom({"kb_ingest_last_run_timestamp": 123.0})
|
||||||
|
assert "kb_ingest_last_run_timestamp 123.0" in content
|
||||||
|
assert "kb_ingest_embed_backlog" not in content
|
||||||
|
|
||||||
|
|
||||||
|
class TestWriteAndReadPromFile:
|
||||||
|
def test_write_then_read_prev_metric(self, tmp_path):
|
||||||
|
path = tmp_path / "kb-ingest.prom"
|
||||||
|
content = render_prom({"kb_ingest_last_success_timestamp": 1700000000.0})
|
||||||
|
write_prom_atomic(path, content)
|
||||||
|
|
||||||
|
assert path.read_text() == content
|
||||||
|
assert list(tmp_path.iterdir()) == [path] # no leftover .tmp file
|
||||||
|
|
||||||
|
assert _read_prev_metric(path, "kb_ingest_last_success_timestamp") == 1700000000.0
|
||||||
|
assert _read_prev_metric(path, "kb_ingest_missing_metric") is None
|
||||||
|
|
||||||
|
def test_read_prev_metric_missing_file_returns_none(self, tmp_path):
|
||||||
|
assert _read_prev_metric(tmp_path / "nope.prom", "kb_ingest_last_success_timestamp") is None
|
||||||
|
|
||||||
|
|
||||||
|
class TestBuildMetrics:
|
||||||
|
def _result(self, failed, ollama_up=True, embed_backlog=0):
|
||||||
|
return {
|
||||||
|
"ollama_up": ollama_up,
|
||||||
|
"adapter": {"stats": dict(ADAPTER_OK), "failed": False, "skipped": False, "error": None},
|
||||||
|
"chunk_embed": {"stats": dict(CHUNK_OK), "failed": False, "skipped": False, "error": None},
|
||||||
|
"summarize": {"stats": dict(SUMMARIZE_OK), "failed": False, "skipped": False, "error": None},
|
||||||
|
"embed_summaries": {"stats": dict(EMBED_SUMMARIES_OK), "failed": False, "skipped": False, "error": None},
|
||||||
|
"embed_backlog": embed_backlog,
|
||||||
|
"failed": failed,
|
||||||
|
}
|
||||||
|
|
||||||
|
def test_success_sets_last_success_to_now(self, tmp_path):
|
||||||
|
prom_path = tmp_path / "kb-ingest.prom"
|
||||||
|
metrics = build_metrics(self._result(failed=False), now_ts=1000.0, prom_path=prom_path)
|
||||||
|
assert metrics["kb_ingest_last_success_timestamp"] == 1000.0
|
||||||
|
assert metrics["kb_ingest_last_exit_code"] == 0
|
||||||
|
assert metrics["kb_ingest_documents_inserted"] == 1
|
||||||
|
assert metrics["kb_ingest_embed_skipped"] == 0
|
||||||
|
|
||||||
|
def test_failure_carries_forward_previous_success(self, tmp_path):
|
||||||
|
prom_path = tmp_path / "kb-ingest.prom"
|
||||||
|
write_prom_atomic(prom_path, render_prom({"kb_ingest_last_success_timestamp": 500.0}))
|
||||||
|
|
||||||
|
metrics = build_metrics(self._result(failed=True), now_ts=1000.0, prom_path=prom_path)
|
||||||
|
assert metrics["kb_ingest_last_success_timestamp"] == 500.0
|
||||||
|
assert metrics["kb_ingest_last_run_timestamp"] == 1000.0
|
||||||
|
assert metrics["kb_ingest_last_exit_code"] == 1
|
||||||
|
|
||||||
|
def test_failure_with_no_previous_file_defaults_to_zero(self, tmp_path):
|
||||||
|
prom_path = tmp_path / "kb-ingest.prom"
|
||||||
|
metrics = build_metrics(self._result(failed=True), now_ts=1000.0, prom_path=prom_path)
|
||||||
|
assert metrics["kb_ingest_last_success_timestamp"] == 0.0
|
||||||
|
|
||||||
|
def test_ollama_down_sets_embed_skipped(self, tmp_path):
|
||||||
|
result = self._result(failed=False, ollama_up=False)
|
||||||
|
result["chunk_embed"] = {"stats": None, "failed": False, "skipped": True, "error": None}
|
||||||
|
metrics = build_metrics(result, now_ts=1000.0, prom_path=tmp_path / "x.prom")
|
||||||
|
assert metrics["kb_ingest_embed_skipped"] == 1
|
||||||
|
assert metrics["kb_ingest_chunks_inserted"] == 0
|
||||||
|
|
||||||
|
def test_embed_backlog_none_is_omitted(self, tmp_path):
|
||||||
|
metrics = build_metrics(
|
||||||
|
self._result(failed=False, embed_backlog=None), now_ts=1000.0, prom_path=tmp_path / "x.prom",
|
||||||
|
)
|
||||||
|
assert "kb_ingest_embed_backlog" not in metrics
|
||||||
|
|
||||||
|
|
||||||
|
# ---------------------------------------------------------------------------
|
||||||
|
# main() — CLI-level guardrails and exit-code propagation
|
||||||
|
# ---------------------------------------------------------------------------
|
||||||
|
|
||||||
|
class TestMainGuards:
|
||||||
|
def test_missing_dsn_exits_1(self, monkeypatch):
|
||||||
|
monkeypatch.setattr(sys, "argv", ["documents-ingest-cyclic",
|
||||||
|
"--paperless-token", "tok"])
|
||||||
|
monkeypatch.delenv("KB_DSN", raising=False)
|
||||||
|
with pytest.raises(SystemExit) as exc:
|
||||||
|
main()
|
||||||
|
assert exc.value.code == 1
|
||||||
|
|
||||||
|
def test_missing_paperless_token_exits_1(self, monkeypatch):
|
||||||
|
monkeypatch.setattr(sys, "argv", ["documents-ingest-cyclic", "--dsn", "dsn"])
|
||||||
|
monkeypatch.delenv("PAPERLESS_API_TOKEN", raising=False)
|
||||||
|
with pytest.raises(SystemExit) as exc:
|
||||||
|
main()
|
||||||
|
assert exc.value.code == 1
|
||||||
|
|
||||||
|
def test_apply_without_anthropic_key_exits_1(self, monkeypatch):
|
||||||
|
monkeypatch.setattr(sys, "argv", [
|
||||||
|
"documents-ingest-cyclic", "--dsn", "dsn", "--paperless-token", "tok", "--apply",
|
||||||
|
])
|
||||||
|
monkeypatch.delenv("ANTHROPIC_API_KEY", raising=False)
|
||||||
|
with pytest.raises(SystemExit) as exc:
|
||||||
|
main()
|
||||||
|
assert exc.value.code == 1
|
||||||
|
|
||||||
|
|
||||||
|
class TestMainExitCodePropagation:
|
||||||
|
def test_exit_0_on_success(self, monkeypatch, tmp_path):
|
||||||
|
_patch_stages(monkeypatch, ollama_up=True)
|
||||||
|
monkeypatch.setattr(sys, "argv", [
|
||||||
|
"documents-ingest-cyclic", "--dsn", "dsn", "--paperless-token", "tok",
|
||||||
|
"--anthropic-api-key", "key", "--prom-path", str(tmp_path / "kb-ingest.prom"),
|
||||||
|
])
|
||||||
|
with pytest.raises(SystemExit) as exc:
|
||||||
|
main()
|
||||||
|
assert exc.value.code == 0
|
||||||
|
assert (tmp_path / "kb-ingest.prom").exists()
|
||||||
|
|
||||||
|
def test_exit_1_on_stage_failure(self, monkeypatch, tmp_path):
|
||||||
|
_patch_stages(monkeypatch, ollama_up=True, adapter=dict(ADAPTER_OK, errors=1))
|
||||||
|
monkeypatch.setattr(sys, "argv", [
|
||||||
|
"documents-ingest-cyclic", "--dsn", "dsn", "--paperless-token", "tok",
|
||||||
|
"--anthropic-api-key", "key", "--prom-path", str(tmp_path / "kb-ingest.prom"),
|
||||||
|
])
|
||||||
|
with pytest.raises(SystemExit) as exc:
|
||||||
|
main()
|
||||||
|
assert exc.value.code == 1
|
||||||
|
|
||||||
|
def test_exit_0_when_ollama_down_apply_still_true(self, monkeypatch, tmp_path):
|
||||||
|
_patch_stages(monkeypatch, ollama_up=False)
|
||||||
|
monkeypatch.setattr(sys, "argv", [
|
||||||
|
"documents-ingest-cyclic", "--dsn", "dsn", "--paperless-token", "tok",
|
||||||
|
"--anthropic-api-key", "key", "--apply",
|
||||||
|
"--prom-path", str(tmp_path / "kb-ingest.prom"),
|
||||||
|
])
|
||||||
|
with pytest.raises(SystemExit) as exc:
|
||||||
|
main()
|
||||||
|
assert exc.value.code == 0
|
||||||
|
prom_content = (tmp_path / "kb-ingest.prom").read_text()
|
||||||
|
assert "kb_ingest_embed_skipped 1" in prom_content
|
||||||
38
services/fleet-prometheus/rules/kb-ingest.yml
Normal file
38
services/fleet-prometheus/rules/kb-ingest.yml
Normal file
|
|
@ -0,0 +1,38 @@
|
||||||
|
# fleet-prometheus kb-ingest rules — module 5 phase 3 step 5
|
||||||
|
# (docs/kb/modules/05-faza3-plan.md §7.2).
|
||||||
|
#
|
||||||
|
# Same delivery convention as liveness.yml: no Alertmanager, these rules only make alerts
|
||||||
|
# FIRING (visible at GET /api/v1/alerts on this Prometheus instance); brain-watchdog@PIHA
|
||||||
|
# polls that API and forwards to Telegram. Do NOT add Alertmanager config here.
|
||||||
|
#
|
||||||
|
# Metrics come from documents-ingest-cyclic's Prometheus textfile-collector output
|
||||||
|
# (jobs/documents-ingest/systemd, `/opt/homelab/state/node-exporter/kb-ingest.prom`),
|
||||||
|
# scraped via node_exporter's textfile collector on PIHA — every series from that target
|
||||||
|
# already carries `node="piha"` (fleet-node job's static_configs label), same as liveness.yml.
|
||||||
|
|
||||||
|
groups:
|
||||||
|
- name: kb-ingest
|
||||||
|
rules:
|
||||||
|
- alert: KbIngestStale
|
||||||
|
# 2 missed daily runs (OnCalendar=*-*-* 03:30) — the timer itself down, or every
|
||||||
|
# run since has hard-failed (Ollama-down alone never blocks this: the wrapper only
|
||||||
|
# skips its two embed stages then, still updating last_success_timestamp).
|
||||||
|
expr: time() - kb_ingest_last_success_timestamp{node="piha"} > 172800
|
||||||
|
for: 5m
|
||||||
|
labels:
|
||||||
|
severity: critical
|
||||||
|
annotations:
|
||||||
|
summary: "kb-ingest stale on {{ $labels.node }}"
|
||||||
|
description: "kb-ingest.timer has not completed a successful run in over 48h on {{ $labels.node }} — check `journalctl -u kb-ingest.service` and the latest /opt/homelab/logs/kb-ingest/run-*.log."
|
||||||
|
|
||||||
|
- alert: KbEmbedBacklogGrowing
|
||||||
|
# SOLARIA/Ollama has been unreachable (or embedding has been failing) for 3 straight
|
||||||
|
# days of ticks — the backlog itself isn't an incident (Ollama sleeps by design,
|
||||||
|
# plan §1.3), sustained non-zero for this long is.
|
||||||
|
expr: kb_ingest_embed_backlog{node="piha"} > 0
|
||||||
|
for: 72h
|
||||||
|
labels:
|
||||||
|
severity: warning
|
||||||
|
annotations:
|
||||||
|
summary: "kb-ingest embed backlog growing on {{ $labels.node }}"
|
||||||
|
description: "kb_ingest_embed_backlog on {{ $labels.node }} has been > 0 for 72h straight — Ollama@SOLARIA may not have come back up in that window, or chunk/summary embedding has been failing on live Ollama."
|
||||||
|
|
@ -1,5 +1,10 @@
|
||||||
name: node_exporter
|
name: node_exporter
|
||||||
owner_node: vps
|
# Deployed per-host: vps and piha (each with its own hosts/<node>/services.yaml entry +
|
||||||
|
# hosts/<node>/runtime/node_exporter/docker-compose.override.yml). Was owner_node: vps only
|
||||||
|
# until KB module 5 phase 3 step 5 (docs/kb/modules/05-faza3-plan.md §7.2) needed the
|
||||||
|
# textfile collector on PIHA — closes the docs/backlog.md "stability-agent / node_exporter
|
||||||
|
# owner_node single, biegaja wielomiejscowo -> per-host" item for node_exporter's half.
|
||||||
|
owner_node: per-host
|
||||||
role: metrics-exporter
|
role: metrics-exporter
|
||||||
description: >
|
description: >
|
||||||
Prometheus Node Exporter — exposes host-level metrics (CPU, memory, disk,
|
Prometheus Node Exporter — exposes host-level metrics (CPU, memory, disk,
|
||||||
|
|
|
||||||
Loading…
Reference in a new issue