Compare commits

..

7 Commits

Author SHA1 Message Date
unkinben 9e7944835e fix: cross-link tag manifests to digest keys and add fetch lock to prevent thundering herd
ci/woodpecker/pr/pre-commit Pipeline failed
ci/woodpecker/pr/test Pipeline was successful
ci/woodpecker/pr/build Pipeline was successful
Tag manifests (e.g. library/nginx/manifests/latest) and their sha256-addressed
counterparts were stored at separate S3 keys with no cross-reference, so a
sha256 manifest request always missed cache even when the identical content had
just been stored under the tag key.

After serving any mutable (tag) manifest, compute the sha256 of the response
body and write it under the digest key (manifests/sha256:<hex>) if absent. The
next sha256-addressed pull hits cache immediately.

Also adds a short-lived Redis distributed lock (SET NX EX 30) around upstream
fetches so that concurrent pods racing for the same cold key poll storage for
up to 5 s before issuing a duplicate upstream request, eliminating the
thundering herd on deploy events.

Includes unit tests for both the lock primitives (acquire/release, fail-open
when Redis is unavailable) and the docker proxy behaviour (cross-link written
on tag hit, not written for sha256 requests, lock acquired/released, poll path
serves from cache without upstream fetch, fallback fetch when poll times out).
2026-05-10 21:21:24 +10:00
unkinben 8a7f26b193 feat: cache parsed member indexes as msgpack to skip YAML re-parse on rebuild (#40)
ci/woodpecker/tag/docker Pipeline was successful
Closes #36

## Summary

- After fetching a member's `index.yaml` (from upstream or S3), the handler now parses it and stores a compact msgpack file (`index.msgpack`) alongside the raw YAML in S3
- On subsequent virtual rebuilds (member caches valid, virtual TTL expired), the handler loads the msgpack file instead of re-parsing raw YAML — eliminating the costliest phase
- `_entries_to_msgpack_safe()` converts datetime/date objects to ISO strings before packing (msgpack cannot natively serialize Python datetimes)
- `_merge_helm_indexes()` accepts `list[dict | None]` as pre-parsed entries; falls back to raw YAML parse when msgpack is unavailable
- `_VirtualHandler.merge()` protocol updated to pass pre-parsed entries to all future handler implementations
- Broken msgpack is detected and rebuilt from raw YAML automatically

## Performance

Phase breakdown (19-member helm-all virtual, 14 MB total):

| Phase | Time | % |
|---|---|---|
| YAML parse (eliminated) | 6314 ms | 60% |
| URL rewrite + dedup | 33 ms | 0.3% |
| YAML dump | 4124 ms | 39% |

| Scenario | Before (CSafeLoader only, #34) | After |
|---|---|---|
| Cold rebuild (upstream fetch) | ~21s | ~26s (+5s for msgpack build, one-time) |
| **Warm rebuild (S3 hit, virtual expired)** | **~9.6s** | **~5.9s (38% faster)** |
| Virtual cache hit | ~0.03s | ~0.03s |

Log line confirms msgpack hits: `msgpack=19/19`

## Test plan

- 297 tests pass
- `TestEntriesToMsgpackSafe`: datetime/date serialization, empty input, round-trip
- `TestMergeHelmIndexesWithParsed`: pre-parsed path produces identical output to raw-bytes path
- `TestGetMemberIndexMsgpack`: msgpack hit, cold-build, broken msgpack fallback, upstream failure
- Docker warm-rebuild measured at 5.9s vs 9.6s baseline

Reviewed-on: #40
2026-05-02 17:15:31 +10:00
unkinben 15f934cd0b perf: use yaml.CSafeLoader/CDumper for 4x faster virtual index merge (#39)
Closes #34

## Summary

- At module load time, a `try/except` selects `yaml.CSafeLoader` / `yaml.CDumper` (C extensions) when libyaml is available, otherwise falls back to `yaml.SafeLoader` / `yaml.Dumper`
- `_HelmDumper` inherits from whichever dumper base was selected — custom datetime/date representers are registered the same way as before
- `_merge_helm_indexes` uses `yaml.load(raw_data, Loader=_YamlLoader)` instead of `yaml.safe_load`
- No change to `yaml.dump(...)` call — it already passes `Dumper=_HelmDumper`, which now inherits from the C base when available
- Five new tests in `TestYamlExtensionSelection` cover: loader/dumper base are classes, `_HelmDumper` inherits from the selected base, C extensions used when available, loader can parse YAML

## Measured performance gain

19-member `helm-all` virtual repo, real upstream data, Docker (AlmaLinux 9):

| | `merge=` time |
|---|---|
| Before (SafeLoader + Dumper) | **38,877ms** |
| After (CSafeLoader + CDumper) | **9,625ms** |
| Speedup | **4.0×** |

Local microbenchmark (500 charts × 10 versions × 19 members, 3 runs avg):
- Before: **40.8s** → After: **6.1s** (**6.7×** faster)

## Test plan

- [x] 283 unit tests pass (`make test`)
- [x] Wheel builds cleanly (`uv build --wheel`)
- [x] C extension confirmed available in AlmaLinux 9 container: `yaml.CSafeLoader: <class 'yaml.cyaml.CSafeLoader'>`
- [x] Baseline Docker timing measured with pure-Python path forced: merge=38,877ms
- [x] After Docker timing measured with C extension path: merge=9,625ms

Reviewed-on: #39
2026-05-02 11:51:00 +10:00
unkinben 7b6c69b70f perf: offload virtual repo merge to thread pool via asyncio.to_thread (#38)
Closes #35

## Summary

- Wraps `handler.merge(...)` in `await asyncio.to_thread(...)` so the CPU-bound YAML parse/merge/dump runs in the thread pool instead of blocking the event loop
- Change is at the generic `handle()` dispatch site — applies to all current and future `_VirtualHandler` implementations without modification
- Also fixes a pre-existing bug in `examples/single-file/remotes.yaml` where `base_url` and `package` keys were merged onto a single line, preventing `docker-compose up` from starting the app

## Measured performance gain

19-member `helm-all` virtual repo, single uvicorn worker, cache miss (38s merge):

| | Concurrent `/health` latency |
|---|---|
| Before (blocking) | **37,721ms** for first request (stalled) |
| After (thread pool) | **8–63ms** for all requests |

## Test plan

- [x] 278 unit tests pass (`make test`)
- [x] Live concurrency test: cache miss merge started in background, 5 concurrent `/health` checks measured — all <65ms
- [x] Baseline comparison: same test with blocking call — first health check stalled 37.7s

Reviewed-on: #38
2026-05-02 01:35:45 +10:00
unkinben 624d858062 fix: rewrite helm index.yaml URLs post-parse to handle relative URLs (#37)
Closes #33

## Summary

- `_merge_helm_indexes` now parses each member's raw YAML first, then rewrites `urls` entries in-place via the new `_rewrite_urls` helper
- **Relative URLs** (e.g. `rancher-2.13.1.tgz`) are prepended with `{proxy_base}/api/v1/remote/{member_name}/`
- **Absolute URLs** matching `base_url` are rewritten to the proxy path (existing behaviour, now correct)
- **Absolute URLs** with a different prefix are left unchanged
- Removes the `_helm.resolve_content` raw-bytes detour from the virtual merge path; `remote/helm.py` is unchanged (still used for direct remote proxying)

## Test plan

- [x] 278 unit tests pass (`make test`)
- [x] New `TestRewriteUrls` class covering relative, absolute-match, absolute-no-match, leading-slash, and multi-URL cases
- [x] New `test_relative_urls_rewritten_to_proxy` in `TestMergeHelmIndexes`
- [x] Updated `test_first_member_wins_on_duplicate_name_and_version` to assert on proxy remote name (not upstream hostname)
- [x] Live Docker test: Rancher `index.yaml` relative URLs rewritten correctly to `http://localhost:8000/api/v1/remote/rancher-stable/rancher-2.14.1.tgz` etc.
- [x] `helm-all` virtual (19 members) returns HTTP 200 with 395k-line merged index on cache miss

Reviewed-on: #37
2026-05-02 01:22:16 +10:00
unkinben 1656664dfa refactor: split config into remotes/virtuals/locals sections (#31)
ci/woodpecker/tag/docker Pipeline was successful
Repository types now live under dedicated top-level keys instead of a
shared remotes: block distinguished by a type field:

  remotes:   caching proxy remotes (no type field needed)
  virtuals:  virtual merged-index repositories
  locals:    local upload repositories

Routes for local repos move from /api/v1/remote/ to /api/v1/local/.
config.py gains get_virtual_config() and get_local_config() lookups.
Root endpoint now reports all three sections. Drop root conf.d/ (was
an exact duplicate of examples/conf.d-method/).

Reviewed-on: #31
2026-04-30 23:50:20 +10:00
unkinben c7baae8d0d feat: add virtual repository support for unified index merging (#30)
Adds a new virtual repo type that merges indexes from multiple member remotes
of the same package type. Currently supports helm (index.yaml merge with URL
rewriting). Member fetches run in parallel; merged index is Redis-cached at
min(mutable_ttl) across members.

Reviewed-on: #30
2026-04-29 23:01:14 +10:00
22 changed files with 803 additions and 316 deletions
+26 -23
View File
@@ -70,10 +70,11 @@ src/artifactapi/
| Method | Path | Description |
|---|---|---|
| `GET` | `/api/v1/remote/{remote}/{path}` | Fetch artifact (auto-cache on miss) |
| `PUT` | `/api/v1/remote/{remote}/{path}` | Upload to local remote |
| `HEAD` | `/api/v1/remote/{remote}/{path}` | Check existence (local remotes) |
| `DELETE` | `/api/v1/remote/{remote}/{path}` | Delete from local remote |
| `GET` | `/api/v1/virtual/{virtual}/{path}` | Fetch from virtual (merged) repository |
| `GET` | `/api/v1/local/{local}/{path}` | Download from local repository |
| `PUT` | `/api/v1/local/{local}/{path}` | Upload to local repository |
| `HEAD` | `/api/v1/local/{local}/{path}` | Check existence (local) |
| `DELETE` | `/api/v1/local/{local}/{path}` | Delete from local repository |
| `GET` | `/v2/{remote}/{path}` | Docker Registry v2 proxy |
| `PUT` | `/cache/flush` | Flush cache entries |
| `GET` | `/health` | Health check |
@@ -120,13 +121,14 @@ config_dir: conf.d # or an absolute path
remotes: {} # optional base remotes
```
### remotes.yaml Structure
### Configuration structure
Repositories are declared under three top-level keys matching their type:
```yaml
remotes:
remotes: # proxy (caching) remotes
remote-name:
base_url: "https://example.com"
type: "remote" # "remote", "local", or "virtual"
package: "generic" # generic, alpine, rpm, docker, pypi, npm, helm
description: "..."
immutable_patterns: # regex — cached forever
@@ -137,6 +139,20 @@ remotes:
cache:
immutable_ttl: 0 # 0 = indefinitely
mutable_ttl: 3600
virtuals: # virtual (merged-index) repositories
virtual-name:
package: "helm"
members:
- remote-a
- remote-b
locals: # local upload repositories (no base_url)
local-name:
package: "generic"
cache:
immutable_ttl: 0
mutable_ttl: 0
```
## Remote Types
@@ -149,7 +165,6 @@ Arbitrary HTTP file servers — GitHub releases, HashiCorp, custom servers.
remotes:
github:
base_url: "https://github.com"
type: "remote"
package: "generic"
immutable_patterns:
- "gruntwork-io/terragrunt/.*terragrunt_linux_amd64.*"
@@ -158,7 +173,6 @@ remotes:
github-archive:
base_url: "https://github.com"
type: "remote"
package: "generic"
immutable_patterns:
- ".*/archive/refs/tags/.*\\.tar\\.gz$" # tag archives never change
@@ -178,7 +192,6 @@ Access: `GET /api/v1/remote/github/owner/repo/releases/download/v1.0/binary.tar.
remotes:
alpine:
base_url: "https://dl-cdn.alpinelinux.org"
type: "remote"
package: "alpine"
immutable_patterns:
- ".*/x86_64/.*\\.apk$"
@@ -195,7 +208,6 @@ remotes:
remotes:
almalinux:
base_url: "https://mirror.example.com/almalinux"
type: "remote"
package: "rpm"
immutable_patterns:
- ".*/x86_64/.*\\.rpm$"
@@ -213,7 +225,6 @@ remotes:
remotes:
dockerhub:
base_url: "https://registry-1.docker.io"
type: "remote"
package: "docker"
# username / password optional for public images
cache:
@@ -222,7 +233,6 @@ remotes:
ghcr:
base_url: "https://ghcr.io"
type: "remote"
package: "docker"
username: "your-github-username"
password: "ghp_your_pat" # read:packages scope
@@ -255,7 +265,6 @@ mirrors:
remotes:
pypi:
base_url: "https://files.pythonhosted.org"
type: "remote"
package: "pypi"
check_mutable_updates: true
immutable_patterns:
@@ -287,7 +296,6 @@ default = true
remotes:
npm:
base_url: "https://registry.npmjs.org"
type: "remote"
package: "npm"
check_mutable_updates: true
immutable_patterns:
@@ -314,7 +322,6 @@ registry=https://artifacts.example.com/api/v1/remote/npm/
remotes:
hashicorp-helm:
base_url: "https://helm.releases.hashicorp.com"
type: "remote"
package: "helm"
check_mutable_updates: true
immutable_patterns:
@@ -343,7 +350,6 @@ All members must share the same `package` type as the virtual repo. Currently su
remotes:
helm-hashicorp:
base_url: "https://helm.releases.hashicorp.com"
type: "remote"
package: "helm"
immutable_patterns:
- "\\.tgz$"
@@ -353,7 +359,6 @@ remotes:
helm-bitnami:
base_url: "https://charts.bitnami.com/bitnami"
type: "remote"
package: "helm"
immutable_patterns:
- "\\.tgz$"
@@ -361,8 +366,8 @@ remotes:
immutable_ttl: 0
mutable_ttl: 3600
virtuals:
helm-all:
type: "virtual"
package: "helm"
members:
- helm-hashicorp # listed first = highest priority
@@ -385,7 +390,7 @@ If a member is unreachable and has no cached index, it is skipped and a warning
**Caching:**
The merged index is cached using `min(mutable_ttl)` across all members. Each member's raw index is cached in S3 under its own remote key by the normal proxy rules; the virtual handler reuses those copies when available.
The merged index is cached using `min(mutable_ttl)` across all members. Each member's raw index is cached in S3 under its own remote key; the virtual handler reuses those copies when available. On rebuild, each member's parsed index is also stored as a compact msgpack file (`index.msgpack`) alongside the raw YAML, eliminating the YAML parse cost on subsequent rebuilds.
**Helm example:**
@@ -399,9 +404,8 @@ Chart tarball URLs in the merged `index.yaml` are rewritten to point at the indi
### local
```yaml
remotes:
locals:
local-generic:
type: "local"
package: "generic"
description: "Local file repository"
cache:
@@ -409,7 +413,7 @@ remotes:
mutable_ttl: 0
```
No `base_url`. Files are uploaded via `PUT` and served via `GET`.
No `base_url`. Files are uploaded via `PUT /api/v1/local/{name}/{path}` and downloaded via `GET /api/v1/local/{name}/{path}`.
## Caching Model
@@ -451,7 +455,6 @@ Set `quarantine_new: true` and `quarantine_days: N` on a remote to block immutab
remotes:
pypi:
base_url: "https://files.pythonhosted.org"
type: "remote"
package: "pypi"
quarantine_new: true
quarantine_days: 3 # block packages published in the last 3 days
-11
View File
@@ -1,11 +0,0 @@
remotes:
alpine:
base_url: "https://dl-cdn.alpinelinux.org"
type: "remote"
package: "alpine"
description: "Alpine Linux APK package repository"
immutable_patterns:
- ".*/x86_64/.*\\.apk$"
cache:
immutable_ttl: 0
mutable_ttl: 7200
-12
View File
@@ -1,12 +0,0 @@
remotes:
github:
base_url: "https://github.com"
type: "remote"
package: "generic"
description: "GitHub releases and files"
immutable_patterns:
- "gruntwork-io/terragrunt/.*terragrunt_linux_amd64.*"
- "prometheus/node_exporter/.*/node_exporter-.*\\.linux-amd64\\.tar\\.gz$"
cache:
immutable_ttl: 0
mutable_ttl: 0
-17
View File
@@ -1,17 +0,0 @@
remotes:
pypi:
base_url: "https://files.pythonhosted.org"
type: "remote"
package: "pypi"
description: "Python Package Index"
check_mutable_updates: true
quarantine_new: true
quarantine_days: 3
immutable_patterns:
- "packages/.*\\.whl$"
- "packages/.*\\.whl\\.metadata$"
- "packages/.*\\.tar\\.gz$"
- "packages/.*\\.zip$"
cache:
immutable_ttl: 0
mutable_ttl: 600
-1
View File
@@ -1,7 +1,6 @@
remotes:
alpine:
base_url: "https://dl-cdn.alpinelinux.org"
type: "remote"
package: "alpine"
description: "Alpine Linux APK package repository"
immutable_patterns:
-1
View File
@@ -1,7 +1,6 @@
remotes:
github:
base_url: "https://github.com"
type: "remote"
package: "generic"
description: "GitHub releases and files"
immutable_patterns:
-1
View File
@@ -1,7 +1,6 @@
remotes:
pypi:
base_url: "https://files.pythonhosted.org"
type: "remote"
package: "pypi"
description: "Python Package Index"
check_mutable_updates: true
+3 -34
View File
@@ -35,7 +35,6 @@
remotes:
github:
base_url: "https://github.com"
type: "remote"
package: "generic"
description: "GitHub releases and files"
immutable_patterns:
@@ -67,7 +66,6 @@ remotes:
github-archive:
base_url: "https://github.com"
type: "remote"
package: "generic"
description: "GitHub repository archive tarballs"
immutable_patterns:
@@ -87,7 +85,6 @@ remotes:
gitea-dl:
base_url: "https://dl.gitea.com"
type: "remote"
package: "generic"
description: "Gitea download site"
immutable_patterns:
@@ -98,7 +95,6 @@ remotes:
hashicorp-releases:
base_url: "https://releases.hashicorp.com"
type: "remote"
package: "generic"
description: "HashiCorp product releases"
immutable_patterns:
@@ -119,7 +115,6 @@ remotes:
alpine:
base_url: "https://dl-cdn.alpinelinux.org"
type: "remote"
package: "alpine"
description: "Alpine Linux APK package repository"
immutable_patterns:
@@ -133,7 +128,6 @@ remotes:
almalinux:
base_url: "https://gsl-syd.mm.fcix.net/almalinux"
type: "remote"
package: "rpm"
description: "AlmaLinux RPM package repository"
immutable_patterns:
@@ -150,7 +144,6 @@ remotes:
epel:
base_url: "http://mirror.aarnet.edu.au/pub/epel"
type: "remote"
package: "rpm"
description: "EPEL (Extra Packages for Enterprise Linux)"
immutable_patterns:
@@ -165,7 +158,6 @@ remotes:
fedora:
base_url: "https://gsl-syd.mm.fcix.net/fedora/linux"
type: "remote"
package: "rpm"
description: "Fedora Linux RPM package repository"
immutable_patterns:
@@ -180,7 +172,6 @@ remotes:
ghcr:
base_url: "https://ghcr.io"
type: "remote"
package: "docker"
description: "GitHub Container Registry"
# username: "your-github-username"
@@ -194,7 +185,6 @@ remotes:
dockerhub:
base_url: "https://registry-1.docker.io"
type: "remote"
package: "docker"
description: "Docker Hub registry"
cache:
@@ -203,7 +193,6 @@ remotes:
pypi:
base_url: "https://files.pythonhosted.org"
type: "remote"
package: "pypi"
description: "Python Package Index — simple index and package files via a single remote"
# simple/ requests are transparently fetched from pypi.org; package files come from
@@ -226,7 +215,6 @@ remotes:
pypi-gitea:
base_url: "https://gitea.example.com/api/packages/myorg/pypi"
type: "remote"
package: "pypi"
description: "Private Gitea PyPI registry — simple index and files at the same host"
# username: "your-gitea-username"
@@ -244,7 +232,6 @@ remotes:
npm:
base_url: "https://registry.npmjs.org"
type: "remote"
package: "npm"
description: "npm registry — package metadata with tarball URL rewriting"
check_mutable_updates: true
@@ -258,7 +245,6 @@ remotes:
hashicorp-helm:
base_url: "https://helm.releases.hashicorp.com"
type: "remote"
package: "helm"
description: "HashiCorp Helm chart repository (Vault, Consul, Nomad, etc.)"
check_mutable_updates: true
@@ -270,7 +256,6 @@ remotes:
metallb:
base_url: "https://metallb.github.io/metallb"
type: "remote"
package: "helm"
description: "MetalLB load balancer Helm charts"
check_mutable_updates: true
@@ -282,7 +267,6 @@ remotes:
jetstack:
base_url: "https://charts.jetstack.io"
type: "remote"
package: "helm"
description: "Jetstack Helm charts (cert-manager)"
check_mutable_updates: true
@@ -294,7 +278,6 @@ remotes:
rancher-stable:
base_url: "https://releases.rancher.com/server-charts/stable"
type: "remote"
package: "helm"
description: "Rancher stable Helm charts"
check_mutable_updates: true
@@ -306,7 +289,6 @@ remotes:
purelb:
base_url: "https://gitlab.com/api/v4/projects/20400619/packages/helm/stable"
type: "remote"
package: "helm"
description: "PureLB load balancer Helm charts"
check_mutable_updates: true
@@ -318,7 +300,6 @@ remotes:
istio:
base_url: "https://istio-release.storage.googleapis.com/charts"
type: "remote"
package: "helm"
description: "Istio service mesh Helm charts"
check_mutable_updates: true
@@ -330,7 +311,6 @@ remotes:
cnpg:
base_url: "https://cloudnative-pg.github.io/charts"
type: "remote"
package: "helm"
description: "CloudNativePG operator Helm charts"
check_mutable_updates: true
@@ -342,7 +322,6 @@ remotes:
ceph-csi:
base_url: "https://ceph.github.io/csi-charts"
type: "remote"
package: "helm"
description: "Ceph CSI driver Helm charts"
check_mutable_updates: true
@@ -354,7 +333,6 @@ remotes:
external-dns:
base_url: "https://kubernetes-sigs.github.io/external-dns/"
type: "remote"
package: "helm"
description: "ExternalDNS Helm charts"
check_mutable_updates: true
@@ -366,7 +344,6 @@ remotes:
intel-helm:
base_url: "https://intel.github.io/helm-charts/"
type: "remote"
package: "helm"
description: "Intel Helm charts"
check_mutable_updates: true
@@ -378,7 +355,6 @@ remotes:
elastic:
base_url: "https://helm.elastic.co"
type: "remote"
package: "helm"
description: "Elastic stack Helm charts"
check_mutable_updates: true
@@ -390,7 +366,6 @@ remotes:
k8up-io:
base_url: "https://k8up-io.github.io/k8up"
type: "remote"
package: "helm"
description: "K8up backup operator Helm charts"
check_mutable_updates: true
@@ -402,7 +377,6 @@ remotes:
victoriametrics:
base_url: "https://victoriametrics.github.io/helm-charts/"
type: "remote"
package: "helm"
description: "VictoriaMetrics observability Helm charts"
check_mutable_updates: true
@@ -414,7 +388,6 @@ remotes:
grafana:
base_url: "https://grafana.github.io/helm-charts"
type: "remote"
package: "helm"
description: "Grafana observability Helm charts"
check_mutable_updates: true
@@ -426,7 +399,6 @@ remotes:
helm-openldap:
base_url: "https://jp-gouin.github.io/helm-openldap/"
type: "remote"
package: "helm"
description: "OpenLDAP Helm charts"
check_mutable_updates: true
@@ -438,7 +410,6 @@ remotes:
woodpecker:
base_url: "https://woodpecker-ci.org/"
type: "remote"
package: "helm"
description: "Woodpecker CI Helm charts"
check_mutable_updates: true
@@ -450,7 +421,6 @@ remotes:
stakater:
base_url: "https://stakater.github.io/stakater-charts"
type: "remote"
package: "helm"
description: "Stakater Helm charts"
check_mutable_updates: true
@@ -462,7 +432,6 @@ remotes:
jfrog:
base_url: "https://charts.jfrog.io/"
type: "remote"
package: "helm"
description: "JFrog Helm charts"
check_mutable_updates: true
@@ -474,7 +443,6 @@ remotes:
openvox:
base_url: "https://openvoxproject.github.io/openvox-helm-chart"
type: "remote"
package: "helm"
description: "OpenVox Helm charts"
check_mutable_updates: true
@@ -484,8 +452,9 @@ remotes:
immutable_ttl: 0
mutable_ttl: 3600
virtuals:
helm-all:
type: "virtual"
package: "helm"
description: "Virtual repository merging all helm remotes — member order is priority order for duplicate chart+version"
members:
@@ -509,8 +478,8 @@ remotes:
- jfrog
- openvox
locals:
local-generic:
type: "local"
package: "generic"
description: "Local generic file repository"
cache:
+1
View File
@@ -14,6 +14,7 @@ dependencies = [
"lxml>=4.9.0",
"prometheus-client>=0.19.0",
"python-multipart>=0.0.6",
"msgpack>=1.0.0",
]
requires-python = ">=3.11"
readme = "README.md"
+40 -15
View File
@@ -1,3 +1,4 @@
import asyncio
import hashlib
import json
import logging
@@ -47,23 +48,39 @@ async def proxy(request: Request, remote_name: str, path: str, storage, cache, c
if not await _proxy.handle_expired_mutable(remote_name, path, remote_url, config, cache, storage):
cached_key = None
lock_acquired = False
if not cached_key:
lock_acquired = cache.acquire_fetch_lock(remote_name, path)
if not lock_acquired:
# Another pod is already fetching — poll storage briefly before issuing a duplicate upstream request
for _ in range(10):
await asyncio.sleep(0.5)
probe_key = storage.get_object_key(remote_name, path)
if storage.exists(probe_key):
cached_key = probe_key
break
if not cached_key:
logger.info(f"Cache MISS: {remote_name}/{path} - fetching from remote: {remote_url}")
result = await _proxy.cache_single_artifact(remote_url, remote_name, path, storage, remote_config)
if result["status"] == "error":
raise HTTPException(status_code=502, detail=f"Failed to fetch: {result['error']}")
if result["status"] == "cached" and is_mutable:
cache_config = config.get_cache_config(remote_name)
mutable_ttl = cache_config.get("mutable_ttl", 3600)
cache.mark_index_cached(remote_name, path, mutable_ttl)
logger.info(f"Mutable file cached with TTL: {remote_name}/{path} (ttl: {mutable_ttl}s)")
if result.get("etag") or result.get("last_modified"):
cache.store_mutable_meta(remote_name, path, result.get("etag"), result.get("last_modified"))
if not is_mutable:
published = result.get("last_modified")
if published:
cache.store_artifact_published(remote_name, path, published)
_proxy._check_quarantine(remote_name, published, config)
try:
result = await _proxy.cache_single_artifact(remote_url, remote_name, path, storage, remote_config)
if result["status"] == "error":
raise HTTPException(status_code=502, detail=f"Failed to fetch: {result['error']}")
if result["status"] == "cached" and is_mutable:
cache_config = config.get_cache_config(remote_name)
mutable_ttl = cache_config.get("mutable_ttl", 3600)
cache.mark_index_cached(remote_name, path, mutable_ttl)
logger.info(f"Mutable file cached with TTL: {remote_name}/{path} (ttl: {mutable_ttl}s)")
if result.get("etag") or result.get("last_modified"):
cache.store_mutable_meta(remote_name, path, result.get("etag"), result.get("last_modified"))
if not is_mutable:
published = result.get("last_modified")
if published:
cache.store_artifact_published(remote_name, path, published)
_proxy._check_quarantine(remote_name, published, config)
finally:
if lock_acquired:
cache.release_fetch_lock(remote_name, path)
elif not is_mutable:
published = cache.get_artifact_published(remote_name, path)
if not published:
@@ -90,6 +107,14 @@ async def proxy(request: Request, remote_name: str, path: str, storage, cache, c
content_type = "application/vnd.oci.image.manifest.v1+json"
digest = f"sha256:{hashlib.sha256(artifact_data).hexdigest()}"
# Cross-link tag manifests to their sha256 digest key so digest-addressed pulls hit cache
if is_mutable and "/manifests/" in path:
digest_path = re.sub(r"/manifests/[^/]+$", f"/manifests/{digest}", path)
digest_key = storage.get_object_key(remote_name, digest_path)
if not storage.exists(digest_key):
storage.upload(digest_key, artifact_data)
headers = {
"Docker-Distribution-Api-Version": "registry/2.0",
"Docker-Content-Digest": digest,
+21 -16
View File
@@ -1,5 +1,6 @@
import hashlib
import logging
import os
from fastapi import HTTPException, Response, UploadFile
from fastapi.responses import JSONResponse
@@ -7,12 +8,23 @@ from fastapi.responses import JSONResponse
logger = logging.getLogger(__name__)
def download(remote_name: str, path: str, storage, database, config) -> Response:
if not config.get_local_config(remote_name):
raise HTTPException(status_code=404, detail=f"Local repository '{remote_name}' not configured")
metadata = database.get_local_file_metadata(remote_name, path)
if not metadata:
raise HTTPException(status_code=404, detail="File not found")
content = storage.download_object(metadata["s3_key"])
return Response(
content=content,
media_type=metadata.get("content_type", "application/octet-stream"),
headers={"Content-Disposition": f"attachment; filename={os.path.basename(path)}"},
)
async def upload(remote_name: str, path: str, file: UploadFile, storage, database, config) -> JSONResponse:
remote_config = config.get_remote_config(remote_name)
if not remote_config:
raise HTTPException(status_code=404, detail=f"Remote '{remote_name}' not configured")
if remote_config.get("type") != "local":
raise HTTPException(status_code=400, detail="Upload only supported for local repositories")
if not config.get_local_config(remote_name):
raise HTTPException(status_code=404, detail=f"Local repository '{remote_name}' not configured")
try:
content = await file.read()
@@ -59,12 +71,8 @@ async def upload(remote_name: str, path: str, file: UploadFile, storage, databas
def check_exists(remote_name: str, path: str, database, config) -> Response:
remote_config = config.get_remote_config(remote_name)
if not remote_config:
raise HTTPException(status_code=404, detail=f"Remote '{remote_name}' not configured")
if remote_config.get("type") != "local":
raise HTTPException(status_code=405, detail="HEAD method only supported for local repositories")
if not config.get_local_config(remote_name):
raise HTTPException(status_code=404, detail=f"Local repository '{remote_name}' not configured")
try:
metadata = database.get_local_file_metadata(remote_name, path)
@@ -87,11 +95,8 @@ def check_exists(remote_name: str, path: str, database, config) -> Response:
def delete(remote_name: str, path: str, storage, database, config) -> JSONResponse:
remote_config = config.get_remote_config(remote_name)
if not remote_config:
raise HTTPException(status_code=404, detail=f"Remote '{remote_name}' not configured")
if remote_config.get("type") != "local":
raise HTTPException(status_code=400, detail="Delete only supported for local repositories")
if not config.get_local_config(remote_name):
raise HTTPException(status_code=404, detail=f"Local repository '{remote_name}' not configured")
try:
s3_key = database.delete_local_file(remote_name, path)
-13
View File
@@ -218,19 +218,6 @@ async def handle(request: Request, remote_name: str, path: str, storage, cache,
if not remote_config:
raise HTTPException(status_code=404, detail=f"Remote '{remote_name}' not configured")
if remote_config.get("type") == "local":
metadata = database.get_local_file_metadata(remote_name, path)
if not metadata:
raise HTTPException(status_code=404, detail="File not found")
content = storage.download_object(metadata["s3_key"])
if content is None:
raise HTTPException(status_code=500, detail="File not accessible")
return Response(
content=content,
media_type=metadata.get("content_type", "application/octet-stream"),
headers={"Content-Disposition": f"attachment; filename={os.path.basename(path)}"},
)
path_parts = path.split("/")
if len(path_parts) >= 2:
repo_path = f"{path_parts[0]}/{path_parts[1]}"
+115 -25
View File
@@ -6,15 +6,21 @@ from datetime import UTC, date, datetime
from typing import Protocol, runtime_checkable
import httpx
import msgpack as _msgpack
import yaml
from fastapi import HTTPException, Request, Response
from ..remote import helm as _helm
logger = logging.getLogger(__name__)
try:
_YamlLoader = yaml.CSafeLoader
_YamlDumperBase = yaml.CDumper
except AttributeError:
_YamlLoader = yaml.SafeLoader
_YamlDumperBase = yaml.Dumper
class _HelmDumper(yaml.Dumper):
class _HelmDumper(_YamlDumperBase):
"""YAML dumper that serializes datetime/date objects back to ISO 8601 strings.
yaml.safe_load converts timestamp-shaped YAML scalars (e.g. chart `created`
@@ -37,21 +43,43 @@ _HelmDumper.add_representer(datetime, _repr_datetime)
_HelmDumper.add_representer(date, _repr_date)
def _entries_to_msgpack_safe(entries: dict) -> dict:
"""Convert datetime/date values to ISO strings for msgpack serialization."""
result = {}
for chart, versions in entries.items():
safe_versions = []
for v in versions:
safe_v = {}
for k, val in v.items():
if isinstance(val, datetime):
safe_v[k] = val.isoformat()
elif isinstance(val, date):
safe_v[k] = val.isoformat()
else:
safe_v[k] = val
safe_versions.append(safe_v)
result[chart] = safe_versions
return result
async def _get_member_index(
member_name: str,
member_cfg: dict,
path: str,
storage,
cache,
) -> tuple[str, dict, int, bytes | None]:
) -> tuple[str, dict, int, bytes | None, dict | None]:
"""Fetch or retrieve cached index.yaml for one member remote.
Returns (member_name, member_cfg, ttl, raw_bytes).
Returns (member_name, member_cfg, ttl, raw_bytes, parsed_entries).
raw_bytes is None if the member is unreachable and not in S3.
parsed_entries is the pre-parsed entries dict (from msgpack cache), or None.
"""
member_ttl = member_cfg.get("cache", {}).get("mutable_ttl", 3600)
s3_key = storage.get_object_key(member_name, path)
msgpack_key = storage.get_object_key(member_name, "index.msgpack")
raw_data: bytes | None = None
parsed_entries: dict | None = None
if storage.exists(s3_key) and cache.is_index_valid(member_name, path):
try:
@@ -59,6 +87,13 @@ async def _get_member_index(
logger.info(f"Virtual: cache hit for member '{member_name}'")
except Exception:
raw_data = None
if raw_data is not None and storage.exists(msgpack_key):
try:
packed = storage.download_object(msgpack_key)
parsed_entries = _msgpack.unpackb(packed, raw=False)
logger.debug(f"Virtual: msgpack hit for member '{member_name}'")
except Exception:
parsed_entries = None
if raw_data is None:
base_url = member_cfg.get("base_url", "").rstrip("/")
@@ -76,35 +111,74 @@ async def _get_member_index(
raw_data = response.content
except Exception as e:
logger.warning(f"Virtual: failed to fetch index.yaml from member '{member_name}': {e}")
return member_name, member_cfg, member_ttl, None
return member_name, member_cfg, member_ttl, None, None
try:
storage.upload(s3_key, raw_data)
cache.mark_index_cached(member_name, path, member_ttl)
except Exception as e:
logger.warning(f"Virtual: failed to cache index.yaml for member '{member_name}': {e}")
return member_name, member_cfg, member_ttl, raw_data
if parsed_entries is None and raw_data is not None:
try:
index = yaml.load(raw_data, Loader=_YamlLoader)
safe_entries = _entries_to_msgpack_safe(index.get("entries") or {})
storage.upload(msgpack_key, _msgpack.packb(safe_entries, use_bin_type=True))
parsed_entries = safe_entries
except Exception as e:
logger.warning(f"Virtual: failed to build msgpack cache for '{member_name}': {e}")
return member_name, member_cfg, member_ttl, raw_data, parsed_entries
def _merge_helm_indexes(raw_indexes: list[bytes], member_names: list[str], member_configs: list[dict], proxy_base: str) -> bytes:
def _rewrite_urls(urls: list, base_url: str, proxy_base: str, member_name: str) -> list:
proxy_remote = f"{proxy_base}/api/v1/remote/{member_name}"
rewritten = []
for url in urls:
if url.startswith(("http://", "https://")):
if base_url and url.startswith(base_url):
url = proxy_remote + url[len(base_url) :]
else:
url = f"{proxy_remote}/{url.lstrip('/')}"
rewritten.append(url)
return rewritten
def _merge_helm_indexes(
raw_indexes: list[bytes],
parsed_entries_list: list[dict | None],
member_names: list[str],
member_configs: list[dict],
proxy_base: str,
) -> bytes:
"""Merge helm index.yaml files with per-member URL rewriting.
Priority is determined by position in member_names: earlier members win
when the same chart name + version appears in multiple remotes.
Uses pre-parsed msgpack entries when available to skip YAML parsing.
"""
merged_entries: dict[str, list] = {}
for raw_data, member_name, member_cfg in zip(raw_indexes, member_names, member_configs):
for raw_data, pre_parsed, member_name, member_cfg in zip(raw_indexes, parsed_entries_list, member_names, member_configs):
base_url = member_cfg.get("base_url", "").rstrip("/")
rewritten, _ = _helm.resolve_content(raw_data, "index.yaml", "index.yaml", base_url, proxy_base, member_name)
try:
index = yaml.safe_load(rewritten)
except Exception as e:
logger.warning(f"Virtual: failed to parse index.yaml from member '{member_name}': {e}")
continue
if pre_parsed is not None:
entries = pre_parsed
else:
try:
index = yaml.load(raw_data, Loader=_YamlLoader)
except Exception as e:
logger.warning(f"Virtual: failed to parse index.yaml from member '{member_name}': {e}")
continue
entries = index.get("entries") or {}
for chart_name, versions in (index.get("entries") or {}).items():
for chart_name, versions in entries.items():
for version_entry in versions:
version_entry["urls"] = _rewrite_urls(
version_entry.get("urls") or [],
base_url,
proxy_base,
member_name,
)
if chart_name not in merged_entries:
merged_entries[chart_name] = list(versions)
else:
@@ -126,7 +200,14 @@ def _merge_helm_indexes(raw_indexes: list[bytes], member_names: list[str], membe
@runtime_checkable
class _VirtualHandler(Protocol):
def accepts_path(self, path: str) -> bool: ...
def merge(self, raw_indexes: list[bytes], member_names: list[str], member_configs: list[dict], proxy_base: str) -> bytes: ...
def merge(
self,
raw_indexes: list[bytes],
parsed_entries: list[dict | None],
member_names: list[str],
member_configs: list[dict],
proxy_base: str,
) -> bytes: ...
def path_error(self) -> str: ...
@@ -134,8 +215,15 @@ class _HelmHandler:
def accepts_path(self, path: str) -> bool:
return path == "index.yaml"
def merge(self, raw_indexes: list[bytes], member_names: list[str], member_configs: list[dict], proxy_base: str) -> bytes:
return _merge_helm_indexes(raw_indexes, member_names, member_configs, proxy_base)
def merge(
self,
raw_indexes: list[bytes],
parsed_entries: list[dict | None],
member_names: list[str],
member_configs: list[dict],
proxy_base: str,
) -> bytes:
return _merge_helm_indexes(raw_indexes, parsed_entries, member_names, member_configs, proxy_base)
def path_error(self) -> str:
return "Virtual helm repositories only serve index.yaml; chart tarballs are served directly by member remotes"
@@ -147,11 +235,9 @@ _HANDLERS: dict[str, _VirtualHandler] = {
async def handle(request: Request, virtual_name: str, path: str, storage, cache, config) -> Response:
virtual_cfg = config.get_remote_config(virtual_name)
virtual_cfg = config.get_virtual_config(virtual_name)
if not virtual_cfg:
raise HTTPException(status_code=404, detail=f"Virtual repository '{virtual_name}' not configured")
if virtual_cfg.get("type") != "virtual":
raise HTTPException(status_code=400, detail=f"'{virtual_name}' is not a virtual repository")
package = virtual_cfg.get("package")
handler = _HANDLERS.get(package)
@@ -188,17 +274,19 @@ async def handle(request: Request, virtual_name: str, path: str, storage, cache,
fetch_ms = int((time.perf_counter() - t_fetch) * 1000)
raw_indexes: list[bytes] = []
used_parsed: list[dict | None] = []
used_members: list[str] = []
used_configs: list[dict] = []
min_ttl: int | None = None
for member_name, member_cfg, member_ttl, raw_data in results:
for member_name, member_cfg, member_ttl, raw_data, parsed_entries in results:
if min_ttl is None or member_ttl < min_ttl:
min_ttl = member_ttl
if raw_data is None:
logger.warning(f"Virtual '{virtual_name}': skipping unreachable member '{member_name}'")
continue
raw_indexes.append(raw_data)
used_parsed.append(parsed_entries)
used_members.append(member_name)
used_configs.append(member_cfg)
@@ -209,7 +297,7 @@ async def handle(request: Request, virtual_name: str, path: str, storage, cache,
min_ttl = 3600
t_merge = time.perf_counter()
merged = handler.merge(raw_indexes, used_members, used_configs, proxy_base)
merged = await asyncio.to_thread(handler.merge, raw_indexes, used_parsed, used_members, used_configs, proxy_base)
merge_ms = int((time.perf_counter() - t_merge) * 1000)
try:
@@ -217,9 +305,11 @@ async def handle(request: Request, virtual_name: str, path: str, storage, cache,
storage.upload(virtual_key, merged)
cache.mark_index_cached(virtual_name, path, min_ttl)
store_ms = int((time.perf_counter() - t_store) * 1000)
msgpack_hits = sum(1 for p in used_parsed if p is not None)
logger.info(
f"Virtual MISS: {virtual_name}/{path} rebuilt from {used_members} "
f"(fetch={fetch_ms}ms merge={merge_ms}ms store={store_ms}ms ttl={min_ttl}s)"
f"(fetch={fetch_ms}ms merge={merge_ms}ms store={store_ms}ms ttl={min_ttl}s "
f"msgpack={msgpack_hits}/{len(used_members)})"
)
except Exception as e:
logger.warning(f"Virtual: failed to store merged index for '{virtual_name}': {e}")
+19
View File
@@ -99,6 +99,25 @@ class RedisCache:
except Exception:
return None
def acquire_fetch_lock(self, remote_name: str, path: str, ttl: int = 30) -> bool:
"""Try to acquire a short-lived fetch lock. Returns True if acquired, False if held by another caller."""
if not self.available:
return True # fail open: no Redis → behave as if we always hold the lock
key = f"fetchlock:{remote_name}:{hashlib.sha256(path.encode()).hexdigest()[:16]}"
try:
return bool(self.client.set(key, 1, nx=True, ex=ttl))
except Exception:
return True
def release_fetch_lock(self, remote_name: str, path: str) -> None:
if not self.available:
return
key = f"fetchlock:{remote_name}:{hashlib.sha256(path.encode()).hexdigest()[:16]}"
try:
self.client.delete(key)
except Exception:
pass
def cleanup_expired_index(self, storage, remote_name: str, path: str) -> None:
if not self.available:
return
+12 -4
View File
@@ -50,8 +50,8 @@ class ConfigManager:
def _merge(base: dict, overlay: dict) -> dict:
result = {**base}
for key, value in overlay.items():
if key == "remotes" and isinstance(base.get("remotes"), dict) and isinstance(value, dict):
result["remotes"] = {**base.get("remotes", {}), **value}
if key in ("remotes", "virtuals", "locals") and isinstance(base.get(key), dict) and isinstance(value, dict):
result[key] = {**base.get(key, {}), **value}
else:
result[key] = value
return result
@@ -67,11 +67,11 @@ class ConfigManager:
self._config_dir = None
if os.path.isdir(self.config_path):
return self._load_from_dir(self.config_path) or {"remotes": {}}
return self._load_from_dir(self.config_path) or {"remotes": {}, "virtuals": {}, "locals": {}}
config = self._load_single_file(self.config_path)
if not config:
return {"remotes": {}}
return {"remotes": {}, "virtuals": {}, "locals": {}}
config_dir = config.pop("config_dir", None)
if config_dir:
@@ -119,6 +119,14 @@ class ConfigManager:
self._check_reload()
return self.config.get("remotes", {}).get(remote_name)
def get_virtual_config(self, virtual_name: str) -> dict | None:
self._check_reload()
return self.config.get("virtuals", {}).get(virtual_name)
def get_local_config(self, local_name: str) -> dict | None:
self._check_reload()
return self.config.get("locals", {}).get(local_name)
def get_immutable_patterns(self, remote_name: str, repo_path: str = "") -> list[str]:
remote_config = self.get_remote_config(remote_name)
if not remote_config:
+21 -10
View File
@@ -49,7 +49,13 @@ class ArtifactRequest(BaseModel):
@app.get("/")
def read_root():
config._check_reload()
return {"message": "Artifact Storage API", "version": app.version, "remotes": list(config.config.get("remotes", {}).keys())}
return {
"message": "Artifact Storage API",
"version": app.version,
"remotes": list(config.config.get("remotes", {}).keys()),
"virtuals": list(config.config.get("virtuals", {}).keys()),
"locals": list(config.config.get("locals", {}).keys()),
}
@app.get("/health")
@@ -99,19 +105,24 @@ async def get_artifact(request: Request, remote_name: str, path: str):
return await proxy.handle(request, remote_name, path, storage, cache, config, database, metrics)
@app.put("/api/v1/remote/{remote_name}/{path:path}")
async def upload_file(remote_name: str, path: str, file: UploadFile = File(...)):
return await local.upload(remote_name, path, file, storage, database, config)
@app.get("/api/v1/local/{local_name}/{path:path}")
def get_local_artifact(local_name: str, path: str):
return local.download(local_name, path, storage, database, config)
@app.head("/api/v1/remote/{remote_name}/{path:path}")
def check_file_exists(remote_name: str, path: str):
return local.check_exists(remote_name, path, database, config)
@app.put("/api/v1/local/{local_name}/{path:path}")
async def upload_local_file(local_name: str, path: str, file: UploadFile = File(...)):
return await local.upload(local_name, path, file, storage, database, config)
@app.delete("/api/v1/remote/{remote_name}/{path:path}")
def delete_file(remote_name: str, path: str):
return local.delete(remote_name, path, storage, database, config)
@app.head("/api/v1/local/{local_name}/{path:path}")
def check_local_file_exists(local_name: str, path: str):
return local.check_exists(local_name, path, database, config)
@app.delete("/api/v1/local/{local_name}/{path:path}")
def delete_local_file(local_name: str, path: str):
return local.delete(local_name, path, storage, database, config)
@app.post("/api/v1/artifacts/cache")
+13 -7
View File
@@ -87,9 +87,10 @@ class MetricsManager:
# Get from database if available
db_sizes = self.database_manager.get_storage_by_remote()
if db_sizes:
# Initialize all configured remotes to 0
# Initialize all configured remotes and locals to 0
remote_sizes = {}
for remote in config_manager.config.get("remotes", {}).keys():
all_names = list(config_manager.config.get("remotes", {}).keys()) + list(config_manager.config.get("locals", {}).keys())
for remote in all_names:
remote_sizes[remote] = db_sizes.get(remote, 0)
# Update Prometheus gauges
@@ -101,10 +102,10 @@ class MetricsManager:
# Fallback to S3 scanning if database not available
try:
remote_sizes = {}
remotes = config_manager.config.get("remotes", {}).keys()
all_names = list(config_manager.config.get("remotes", {}).keys()) + list(config_manager.config.get("locals", {}).keys())
# Initialize all remotes to 0
for remote in remotes:
# Initialize all remotes and locals to 0
for remote in all_names:
remote_sizes[remote] = 0
paginator = storage.client.get_paginator("list_objects_v2")
@@ -174,8 +175,13 @@ class MetricsManager:
metrics["requests"]["cache_hit_ratio"] = cache_hits / total_requests if total_requests > 0 else 0.0
metrics["bandwidth"]["saved_bytes"] = bandwidth_saved
# Get per-remote metrics
for remote in config_manager.config.get("remotes", {}).keys():
# Get per-repo metrics
all_repos = {
**config_manager.config.get("remotes", {}),
**config_manager.config.get("virtuals", {}),
**config_manager.config.get("locals", {}),
}
for remote in all_repos.keys():
remote_cache_hits = int(self.redis_client.client.get(f"metrics:cache_hits:{remote}") or 0)
remote_cache_misses = int(self.redis_client.client.get(f"metrics:cache_misses:{remote}") or 0)
remote_total = remote_cache_hits + remote_cache_misses
+9 -22
View File
@@ -20,61 +20,48 @@ TEST_REMOTES = {
"remotes": {
"alpine-test": {
"base_url": "https://dl-cdn.alpinelinux.org",
"type": "remote",
"package": "alpine",
"immutable_patterns": [".*/x86_64/.*\\.apk$"],
"cache": {"immutable_ttl": 0, "mutable_ttl": 3600},
},
"rpm-test": {
"base_url": "https://example.com/rpm",
"type": "remote",
"package": "rpm",
"immutable_patterns": [".*/x86_64/.*\\.rpm$", ".*/repodata/.*$"],
"cache": {"immutable_ttl": 0, "mutable_ttl": 3600},
},
"docker-test": {
"base_url": "https://registry.example.com",
"type": "remote",
"package": "docker",
"cache": {"immutable_ttl": 0, "mutable_ttl": 300},
},
"docker-restricted": {
"base_url": "https://registry.example.com",
"type": "remote",
"package": "docker",
"immutable_patterns": ["^library/nginx"],
"cache": {"immutable_ttl": 0, "mutable_ttl": 300},
},
"generic-test": {
"base_url": "https://releases.example.com",
"type": "remote",
"package": "generic",
"immutable_patterns": [".*\\.tar\\.gz$"],
"cache": {"immutable_ttl": 0, "mutable_ttl": 0},
},
"custom-index-test": {
"base_url": "https://example.com",
"type": "remote",
"package": "generic",
"mutable_patterns": ["metadata\\.json$"],
"cache": {"immutable_ttl": 0, "mutable_ttl": 600},
},
"check-mutable-test": {
"base_url": "https://example.com",
"type": "remote",
"package": "generic",
"mutable_patterns": ["metadata\\.json$"],
"check_mutable_updates": True,
"cache": {"immutable_ttl": 0, "mutable_ttl": 600},
},
"local-test": {
"type": "local",
"package": "generic",
"cache": {"immutable_ttl": 0, "mutable_ttl": 0},
},
"pypi-test": {
"base_url": "https://files.pythonhosted.org",
"type": "remote",
"package": "pypi",
"immutable_patterns": [
r"packages/.*\.whl$",
@@ -85,7 +72,6 @@ TEST_REMOTES = {
},
"npm-test": {
"base_url": "https://registry.npmjs.org",
"type": "remote",
"package": "npm",
"immutable_patterns": [r"\.tgz$"],
"mutable_patterns": [r"^(?!.*\.tgz$).*"],
@@ -93,14 +79,12 @@ TEST_REMOTES = {
},
"helm-test": {
"base_url": "https://helm.releases.hashicorp.com",
"type": "remote",
"package": "helm",
"immutable_patterns": [r"\.tgz$"],
"cache": {"immutable_ttl": 0, "mutable_ttl": 3600},
},
"quarantine-test": {
"base_url": "https://releases.example.com",
"type": "remote",
"package": "generic",
"immutable_patterns": [r".*\.tar\.gz$"],
"quarantine_new": True,
@@ -109,7 +93,6 @@ TEST_REMOTES = {
},
"quarantine-disabled": {
"base_url": "https://releases.example.com",
"type": "remote",
"package": "generic",
"immutable_patterns": [r".*\.tar\.gz$"],
"quarantine_new": False,
@@ -118,27 +101,31 @@ TEST_REMOTES = {
},
"helm-member-2": {
"base_url": "https://charts.example.com",
"type": "remote",
"package": "helm",
"immutable_patterns": [r"\.tgz$"],
"cache": {"immutable_ttl": 0, "mutable_ttl": 1800},
},
},
"locals": {
"local-test": {
"package": "generic",
"cache": {"immutable_ttl": 0, "mutable_ttl": 0},
},
},
"virtuals": {
"helm-virtual-test": {
"type": "virtual",
"package": "helm",
"members": ["helm-test", "helm-member-2"],
},
"unsupported-virtual-test": {
"type": "virtual",
"package": "rpm",
"members": ["rpm-test"],
},
"empty-virtual-test": {
"type": "virtual",
"package": "helm",
"members": [],
},
}
},
}
# ---------------------------------------------------------------------------
+71
View File
@@ -327,3 +327,74 @@ class TestArtifactPublished:
def test_get_returns_none_when_unavailable(self, unavailable_cache):
assert unavailable_cache.get_artifact_published("remote", "path") is None
# ---------------------------------------------------------------------------
# fetch lock (thundering-herd deduplication)
# ---------------------------------------------------------------------------
class TestFetchLock:
def test_acquire_returns_true_when_lock_obtained(self, cache_with_redis, mock_redis_client):
mock_redis_client.set.return_value = True
result = cache_with_redis.acquire_fetch_lock("myremote", "library/nginx/manifests/latest")
assert result is True
def test_acquire_calls_set_nx_with_ttl(self, cache_with_redis, mock_redis_client):
mock_redis_client.set.return_value = True
cache_with_redis.acquire_fetch_lock("myremote", "library/nginx/manifests/latest", ttl=15)
_, kwargs = mock_redis_client.set.call_args
assert kwargs.get("nx") is True
assert kwargs.get("ex") == 15
def test_acquire_returns_false_when_lock_already_held(self, cache_with_redis, mock_redis_client):
mock_redis_client.set.return_value = None # Redis SET NX → None when key exists
result = cache_with_redis.acquire_fetch_lock("myremote", "library/nginx/manifests/latest")
assert result is False
def test_acquire_fails_open_when_unavailable(self, unavailable_cache):
# caller must be allowed to proceed when Redis is down
assert unavailable_cache.acquire_fetch_lock("myremote", "some/path") is True
def test_acquire_fails_open_on_redis_exception(self, cache_with_redis, mock_redis_client):
mock_redis_client.set.side_effect = Exception("connection reset")
assert cache_with_redis.acquire_fetch_lock("myremote", "some/path") is True
def test_lock_key_embeds_path_hash(self, cache_with_redis, mock_redis_client):
mock_redis_client.set.return_value = True
path = "library/nginx/manifests/latest"
cache_with_redis.acquire_fetch_lock("myremote", path)
args, _ = mock_redis_client.set.call_args
expected_hash = hashlib.sha256(path.encode()).hexdigest()[:16]
assert args[0] == f"fetchlock:myremote:{expected_hash}"
def test_lock_key_hash_is_16_chars(self, cache_with_redis, mock_redis_client):
mock_redis_client.set.return_value = True
cache_with_redis.acquire_fetch_lock("myremote", "some/long/path/file.tar.gz")
args, _ = mock_redis_client.set.call_args
# key format: fetchlock:<remote>:<16-char hash>
parts = args[0].split(":")
assert len(parts) == 3
assert len(parts[2]) == 16
def test_different_paths_produce_different_lock_keys(self, cache_with_redis, mock_redis_client):
mock_redis_client.set.return_value = True
cache_with_redis.acquire_fetch_lock("myremote", "path/a/manifests/latest")
key_a = mock_redis_client.set.call_args[0][0]
mock_redis_client.set.reset_mock()
cache_with_redis.acquire_fetch_lock("myremote", "path/b/manifests/latest")
key_b = mock_redis_client.set.call_args[0][0]
assert key_a != key_b
def test_release_deletes_correct_key(self, cache_with_redis, mock_redis_client):
path = "library/nginx/manifests/latest"
cache_with_redis.release_fetch_lock("myremote", path)
expected_hash = hashlib.sha256(path.encode()).hexdigest()[:16]
mock_redis_client.delete.assert_called_once_with(f"fetchlock:myremote:{expected_hash}")
def test_release_no_op_when_unavailable(self, unavailable_cache):
unavailable_cache.release_fetch_lock("myremote", "some/path") # must not raise
def test_release_no_op_on_redis_exception(self, cache_with_redis, mock_redis_client):
mock_redis_client.delete.side_effect = Exception("timeout")
cache_with_redis.release_fetch_lock("myremote", "some/path") # must not raise
+19 -19
View File
@@ -27,24 +27,24 @@ def make_config(tmp_path):
class TestGetMutablePatterns:
def test_alpine_returns_package_defaults(self, make_config):
cfg = make_config({"r": {"type": "remote", "package": "alpine", "base_url": "https://x.com"}})
cfg = make_config({"r": {"package": "alpine", "base_url": "https://x.com"}})
patterns = cfg.get_mutable_patterns("r")
assert r"APKINDEX\.tar\.gz$" in patterns
def test_rpm_returns_package_defaults(self, make_config):
cfg = make_config({"r": {"type": "remote", "package": "rpm", "base_url": "https://x.com"}})
cfg = make_config({"r": {"package": "rpm", "base_url": "https://x.com"}})
patterns = cfg.get_mutable_patterns("r")
assert r"repomd\.xml$" in patterns
assert any("repodata" in p for p in patterns)
def test_docker_returns_package_defaults(self, make_config):
cfg = make_config({"r": {"type": "remote", "package": "docker", "base_url": "https://x.com"}})
cfg = make_config({"r": {"package": "docker", "base_url": "https://x.com"}})
patterns = cfg.get_mutable_patterns("r")
assert any("manifests" in p for p in patterns)
assert any("tags/list" in p for p in patterns)
def test_generic_returns_empty_list(self, make_config):
cfg = make_config({"r": {"type": "remote", "package": "generic", "base_url": "https://x.com"}})
cfg = make_config({"r": {"package": "generic", "base_url": "https://x.com"}})
assert cfg.get_mutable_patterns("r") == []
def test_unknown_remote_returns_empty_list(self, make_config):
@@ -52,12 +52,12 @@ class TestGetMutablePatterns:
assert cfg.get_mutable_patterns("nonexistent") == []
def test_missing_package_field_defaults_to_generic(self, make_config):
cfg = make_config({"r": {"type": "remote", "base_url": "https://x.com"}})
cfg = make_config({"r": {"base_url": "https://x.com"}})
assert cfg.get_mutable_patterns("r") == []
def test_unknown_package_type_returns_empty_list(self, make_config):
# A mis-spelled package type silently returns [] — this is a known footgun
cfg = make_config({"r": {"type": "remote", "package": "deb", "base_url": "https://x.com"}})
cfg = make_config({"r": {"package": "deb", "base_url": "https://x.com"}})
assert cfg.get_mutable_patterns("r") == []
def test_extra_patterns_appended_after_defaults(self, make_config):
@@ -134,7 +134,7 @@ class TestGetMutablePatterns:
assert r"custom-meta\.xml$" in patterns
def test_npm_has_no_package_defaults(self, make_config):
cfg = make_config({"r": {"type": "remote", "package": "npm", "base_url": "https://x.com"}})
cfg = make_config({"r": {"package": "npm", "base_url": "https://x.com"}})
assert cfg.get_mutable_patterns("r") == []
def test_npm_explicit_mutable_pattern_matches_metadata(self, make_config):
@@ -155,14 +155,14 @@ class TestGetMutablePatterns:
assert any(re.search(p, "@babel/core") for p in patterns)
def test_helm_returns_index_yaml_as_mutable(self, make_config):
cfg = make_config({"r": {"type": "remote", "package": "helm", "base_url": "https://helm.example.com"}})
cfg = make_config({"r": {"package": "helm", "base_url": "https://helm.example.com"}})
patterns = cfg.get_mutable_patterns("r")
assert r"index\.yaml$" in patterns
def test_helm_chart_tarballs_not_mutable_by_default(self, make_config):
import re
cfg = make_config({"r": {"type": "remote", "package": "helm", "base_url": "https://helm.example.com"}})
cfg = make_config({"r": {"package": "helm", "base_url": "https://helm.example.com"}})
patterns = cfg.get_mutable_patterns("r")
# Only index.yaml is mutable; .tgz chart tarballs are not
assert not any(re.search(p, "vault-0.29.1.tgz") for p in patterns)
@@ -210,7 +210,7 @@ class TestGetImmutablePatterns:
assert cfg.get_immutable_patterns("nonexistent") == []
def test_returns_empty_when_no_patterns_configured(self, make_config):
cfg = make_config({"r": {"type": "remote", "package": "generic", "base_url": "https://x.com"}})
cfg = make_config({"r": {"package": "generic", "base_url": "https://x.com"}})
assert cfg.get_immutable_patterns("r") == []
def test_multiple_patterns_returned(self, make_config):
@@ -281,7 +281,7 @@ class TestGetUserMutablePatterns:
def test_excludes_package_defaults(self, make_config):
# Package defaults (APKINDEX etc.) must NOT appear here
cfg = make_config({"r": {"type": "remote", "package": "alpine", "base_url": "https://x.com"}})
cfg = make_config({"r": {"package": "alpine", "base_url": "https://x.com"}})
assert cfg.get_user_mutable_patterns("r") == []
def test_returns_empty_for_missing_remote(self, make_config):
@@ -289,7 +289,7 @@ class TestGetUserMutablePatterns:
assert cfg.get_user_mutable_patterns("nonexistent") == []
def test_returns_empty_when_key_absent(self, make_config):
cfg = make_config({"r": {"type": "remote", "package": "generic", "base_url": "https://x.com"}})
cfg = make_config({"r": {"package": "generic", "base_url": "https://x.com"}})
assert cfg.get_user_mutable_patterns("r") == []
@@ -317,7 +317,7 @@ class TestGetCacheConfig:
assert cfg.get_cache_config("nonexistent") == {}
def test_returns_empty_dict_when_no_cache_key(self, make_config):
cfg = make_config({"r": {"type": "remote", "package": "generic", "base_url": "https://x.com"}})
cfg = make_config({"r": {"package": "generic", "base_url": "https://x.com"}})
assert cfg.get_cache_config("r") == {}
@@ -329,11 +329,11 @@ class TestGetCacheConfig:
class TestConfigReload:
def test_reloads_when_file_mtime_advances(self, tmp_path):
cfg_file = tmp_path / "remotes.yaml"
cfg_file.write_text(yaml.dump({"remotes": {"repo-a": {"type": "remote", "package": "generic", "base_url": "https://x.com"}}}))
cfg_file.write_text(yaml.dump({"remotes": {"repo-a": {"package": "generic", "base_url": "https://x.com"}}}))
cfg = ConfigManager(str(cfg_file))
assert "repo-a" in cfg.config["remotes"]
cfg_file.write_text(yaml.dump({"remotes": {"repo-b": {"type": "remote", "package": "generic", "base_url": "https://y.com"}}}))
cfg_file.write_text(yaml.dump({"remotes": {"repo-b": {"package": "generic", "base_url": "https://y.com"}}}))
future_mtime = cfg._last_modified + 1
os.utime(str(cfg_file), (future_mtime, future_mtime))
@@ -344,7 +344,7 @@ class TestConfigReload:
def test_no_reload_when_file_unchanged(self, tmp_path):
cfg_file = tmp_path / "remotes.yaml"
cfg_file.write_text(yaml.dump({"remotes": {"repo-a": {"type": "remote", "package": "generic", "base_url": "https://x.com"}}}))
cfg_file.write_text(yaml.dump({"remotes": {"repo-a": {"package": "generic", "base_url": "https://x.com"}}}))
cfg = ConfigManager(str(cfg_file))
# Call check_reload without touching the file — should not reload
@@ -360,7 +360,7 @@ class TestConfigReload:
class TestGetQuarantineConfig:
def test_returns_false_zero_when_not_configured(self, make_config):
cfg = make_config({"r": {"type": "remote", "package": "generic", "base_url": "https://x.com"}})
cfg = make_config({"r": {"package": "generic", "base_url": "https://x.com"}})
enabled, days = cfg.get_quarantine_config("r")
assert enabled is False
assert days == 0
@@ -426,7 +426,7 @@ class TestGetQuarantineConfig:
def _remote(base_url: str = "https://x.com") -> dict:
return {"type": "remote", "package": "generic", "base_url": base_url}
return {"package": "generic", "base_url": base_url}
class TestConfigDirMode:
@@ -445,7 +445,7 @@ class TestConfigDirMode:
def test_empty_directory_returns_empty_remotes(self, tmp_path):
cfg = ConfigManager(str(tmp_path))
assert cfg.config == {"remotes": {}}
assert cfg.config == {"remotes": {}, "virtuals": {}, "locals": {}}
def test_ignores_non_yaml_files(self, tmp_path):
(tmp_path / "notes.txt").write_text("not yaml")
+140 -26
View File
@@ -260,6 +260,135 @@ class TestDockerProxy:
mock_fetch.assert_called_once()
assert response.status_code == 200
# --- Issue 1: sha256 digest cross-linking ---
def test_tag_manifest_is_stored_under_digest_key_on_cache_hit(self, client, patched_deps):
# When serving a cached tag manifest the handler must also write the content
# under the sha256 digest key so subsequent sha256-addressed pulls hit cache.
deps = patched_deps
manifest = json.dumps({"mediaType": "application/vnd.oci.image.manifest.v1+json", "layers": []}).encode()
# First exists call (tag manifest): hit. Second (digest key): miss → triggers upload.
deps["storage"].exists.side_effect = [True, False]
deps["storage"].download_object.return_value = manifest
deps["cache"].is_mutable_file.return_value = True
deps["cache"].is_index_valid.return_value = True
response = client.get("/v2/docker-test/library/nginx/manifests/v1.25.3")
assert response.status_code == 200
deps["storage"].upload.assert_called_once_with(
deps["storage"].get_object_key.return_value, manifest
)
def test_tag_manifest_digest_key_not_written_when_already_exists(self, client, patched_deps):
# When the digest key already exists in storage upload must not be called.
deps = patched_deps
manifest = json.dumps({"mediaType": "application/vnd.oci.image.manifest.v1+json", "layers": []}).encode()
# Both the tag key and the digest key already present.
deps["storage"].exists.return_value = True
deps["storage"].download_object.return_value = manifest
deps["cache"].is_mutable_file.return_value = True
deps["cache"].is_index_valid.return_value = True
client.get("/v2/docker-test/library/nginx/manifests/v1.25.3")
deps["storage"].upload.assert_not_called()
def test_sha256_manifest_request_is_not_cross_linked(self, client, patched_deps):
# sha256-addressed manifests are immutable — the cross-link logic must not apply.
deps = patched_deps
manifest = json.dumps({"mediaType": "application/vnd.oci.image.manifest.v1+json", "layers": []}).encode()
deps["storage"].exists.return_value = True
deps["storage"].download_object.return_value = manifest
deps["cache"].is_mutable_file.return_value = False # sha256 manifest is immutable
with patch("artifactapi.artifact.proxy._fetch_last_modified", new_callable=AsyncMock, return_value=None):
client.get("/v2/docker-test/library/nginx/manifests/sha256:" + "a" * 64)
deps["storage"].upload.assert_not_called()
# --- Issue 2: thundering herd distributed lock ---
def test_lock_acquired_and_released_on_upstream_fetch(self, client, patched_deps):
deps = patched_deps
manifest = json.dumps({"mediaType": "application/vnd.oci.image.manifest.v1+json", "layers": []}).encode()
deps["storage"].exists.side_effect = [False, False] # initial miss; digest key also absent
deps["storage"].download_object.return_value = manifest
deps["cache"].is_mutable_file.return_value = True
deps["cache"].acquire_fetch_lock.return_value = True
with patch(
"artifactapi.artifact.proxy.cache_single_artifact",
new_callable=AsyncMock,
return_value={"status": "cached"},
):
response = client.get("/v2/docker-test/library/nginx/manifests/latest")
deps["cache"].acquire_fetch_lock.assert_called_once()
deps["cache"].release_fetch_lock.assert_called_once()
assert response.status_code == 200
def test_lock_released_even_when_fetch_returns_error(self, client, patched_deps):
deps = patched_deps
deps["storage"].exists.return_value = False
deps["cache"].is_mutable_file.return_value = True
deps["cache"].acquire_fetch_lock.return_value = True
with patch(
"artifactapi.artifact.proxy.cache_single_artifact",
new_callable=AsyncMock,
return_value={"status": "error", "error": "upstream down"},
):
response = client.get("/v2/docker-test/library/nginx/manifests/latest")
deps["cache"].release_fetch_lock.assert_called_once()
assert response.status_code == 502
def test_thundering_herd_polls_storage_when_lock_not_acquired(self, client, patched_deps):
# When the lock is held by another pod the handler must poll storage and serve
# from cache once the competing fetch completes, without issuing its own upstream request.
deps = patched_deps
manifest = json.dumps({"mediaType": "application/vnd.oci.image.manifest.v1+json", "layers": []}).encode()
# Initial cache check: miss. First poll iteration: another pod has written it.
# Third call is for the digest cross-link check (is_mutable=True path); digest key exists.
deps["storage"].exists.side_effect = [False, True, True]
deps["storage"].download_object.return_value = manifest
deps["cache"].is_mutable_file.return_value = True
deps["cache"].is_index_valid.return_value = True
deps["cache"].acquire_fetch_lock.return_value = False # lock held by peer
with patch("artifactapi.artifact.docker.asyncio.sleep", new_callable=AsyncMock):
with patch(
"artifactapi.artifact.proxy.cache_single_artifact",
new_callable=AsyncMock,
) as mock_fetch:
response = client.get("/v2/docker-test/library/nginx/manifests/latest")
mock_fetch.assert_not_called()
assert response.status_code == 200
def test_thundering_herd_falls_through_to_fetch_if_poll_times_out(self, client, patched_deps):
# If the item never appears in storage during the poll window the handler must
# still issue its own upstream fetch as a fallback.
deps = patched_deps
manifest = json.dumps({"mediaType": "application/vnd.oci.image.manifest.v1+json", "layers": []}).encode()
# All exists calls return False — item never appears during polling.
deps["storage"].exists.return_value = False
deps["storage"].download_object.return_value = manifest
deps["cache"].is_mutable_file.return_value = True
deps["cache"].acquire_fetch_lock.return_value = False # lock held by peer
with patch("artifactapi.artifact.docker.asyncio.sleep", new_callable=AsyncMock):
with patch(
"artifactapi.artifact.proxy.cache_single_artifact",
new_callable=AsyncMock,
return_value={"status": "cached"},
) as mock_fetch:
response = client.get("/v2/docker-test/library/nginx/manifests/latest")
mock_fetch.assert_called_once()
assert response.status_code == 200
# ---------------------------------------------------------------------------
# Generic artifact route /api/v1/remote/{remote}/{path}
@@ -523,68 +652,53 @@ class TestGenericArtifactRoute:
deps["database"].get_local_file_metadata.return_value = None
deps["database"].available = True
response = client.get("/api/v1/remote/local-test/path/to/nonexistent.bin")
response = client.get("/api/v1/local/local-test/path/to/nonexistent.bin")
assert response.status_code == 404
# ---------------------------------------------------------------------------
# Upload route PUT /api/v1/remote/{remote}/{path}
# Upload route PUT /api/v1/local/{local}/{path}
# ---------------------------------------------------------------------------
class TestUploadRoute:
def test_unknown_remote_returns_404(self, client, patched_deps):
def test_unknown_local_returns_404(self, client, patched_deps):
response = client.put(
"/api/v1/remote/nonexistent/path/to/file.tar.gz",
"/api/v1/local/nonexistent/path/to/file.tar.gz",
files={"file": ("file.tar.gz", b"content", "application/octet-stream")},
)
assert response.status_code == 404
def test_non_local_remote_returns_400(self, client, patched_deps):
response = client.put(
"/api/v1/remote/generic-test/path/to/file.tar.gz",
files={"file": ("file.tar.gz", b"content", "application/octet-stream")},
)
assert response.status_code == 400
# ---------------------------------------------------------------------------
# HEAD route HEAD /api/v1/remote/{remote}/{path}
# HEAD route HEAD /api/v1/local/{local}/{path}
# ---------------------------------------------------------------------------
class TestHeadRoute:
def test_non_local_remote_returns_405(self, client, patched_deps):
response = client.head("/api/v1/remote/generic-test/path/to/file.tar.gz")
assert response.status_code == 405
def test_local_repo_file_not_found_returns_404(self, client, patched_deps):
deps = patched_deps
deps["database"].get_local_file_metadata.return_value = None
deps["database"].available = True
response = client.head("/api/v1/remote/local-test/path/to/nonexistent.bin")
response = client.head("/api/v1/local/local-test/path/to/nonexistent.bin")
assert response.status_code == 404
def test_unknown_remote_returns_404(self, client, patched_deps):
response = client.head("/api/v1/remote/nonexistent/path/to/file.bin")
def test_unknown_local_returns_404(self, client, patched_deps):
response = client.head("/api/v1/local/nonexistent/path/to/file.bin")
assert response.status_code == 404
# ---------------------------------------------------------------------------
# DELETE route DELETE /api/v1/remote/{remote}/{path}
# DELETE route DELETE /api/v1/local/{local}/{path}
# ---------------------------------------------------------------------------
class TestDeleteRoute:
def test_unknown_remote_returns_404(self, client, patched_deps):
response = client.delete("/api/v1/remote/nonexistent/path/to/file.tar.gz")
def test_unknown_local_returns_404(self, client, patched_deps):
response = client.delete("/api/v1/local/nonexistent/path/to/file.tar.gz")
assert response.status_code == 404
def test_non_local_remote_returns_400(self, client, patched_deps):
response = client.delete("/api/v1/remote/generic-test/path/to/file.tar.gz")
assert response.status_code == 400
# ---------------------------------------------------------------------------
# Cache flush PUT /cache/flush
+293 -59
View File
@@ -8,11 +8,15 @@ import yaml
from artifactapi.artifact.virtual import (
_HANDLERS,
_entries_to_msgpack_safe,
_get_member_index,
_HelmDumper,
_HelmHandler,
_merge_helm_indexes,
_rewrite_urls,
_VirtualHandler,
_YamlDumperBase,
_YamlLoader,
)
# ---------------------------------------------------------------------------
@@ -66,12 +70,47 @@ entries:
generated: "2023-01-01T00:00:00.000Z"
"""
_INDEX_RELATIVE = b"""\
apiVersion: v1
entries:
rancher:
- name: rancher
version: "2.13.1"
urls:
- rancher-2.13.1.tgz
generated: "2023-01-01T00:00:00.000Z"
"""
_CFG_A = {"base_url": "https://helm.releases.hashicorp.com", "cache": {"mutable_ttl": 3600}}
_CFG_B = {"base_url": "https://charts.example.com", "cache": {"mutable_ttl": 1800}}
def _identity_resolve(data, *args, **kwargs):
return data, None
# ---------------------------------------------------------------------------
# _YamlLoader / _YamlDumperBase — C extension selection
# ---------------------------------------------------------------------------
class TestYamlExtensionSelection:
def test_loader_is_a_class(self):
assert isinstance(_YamlLoader, type)
def test_dumper_base_is_a_class(self):
assert isinstance(_YamlDumperBase, type)
def test_helm_dumper_uses_selected_base(self):
assert issubclass(_HelmDumper, _YamlDumperBase)
def test_c_extensions_used_when_available(self):
try:
assert _YamlLoader is yaml.CSafeLoader
assert _YamlDumperBase is yaml.CDumper
except AttributeError:
assert _YamlLoader is yaml.SafeLoader
assert _YamlDumperBase is yaml.Dumper
def test_loader_can_parse_yaml(self):
result = yaml.load(b"key: value", Loader=_YamlLoader)
assert result == {"key": "value"}
# ---------------------------------------------------------------------------
@@ -135,14 +174,13 @@ class TestHelmHandler:
assert isinstance(msg, str) and len(msg) > 0
def test_merge_returns_bytes(self):
with patch("artifactapi.artifact.virtual._helm.resolve_content", side_effect=_identity_resolve):
result = self.handler.merge([_INDEX_A], ["member-a"], [_CFG_A], "http://proxy.example.com")
result = self.handler.merge([_INDEX_A], [None], ["member-a"], [_CFG_A], "http://proxy.example.com")
assert isinstance(result, bytes)
def test_merge_delegates_to_merge_helm_indexes(self):
with patch("artifactapi.artifact.virtual._merge_helm_indexes", return_value=b"merged") as mock_fn:
result = self.handler.merge([b"data"], ["m"], [{}], "http://proxy")
mock_fn.assert_called_once_with([b"data"], ["m"], [{}], "http://proxy")
result = self.handler.merge([b"data"], [None], ["m"], [{}], "http://proxy")
mock_fn.assert_called_once_with([b"data"], [None], ["m"], [{}], "http://proxy")
assert result == b"merged"
@@ -160,6 +198,41 @@ class TestHandlersRegistry:
assert isinstance(_HANDLERS["helm"], _VirtualHandler)
# ---------------------------------------------------------------------------
# _rewrite_urls
# ---------------------------------------------------------------------------
class TestRewriteUrls:
def _rewrite(self, urls, base_url="https://upstream.example.com", proxy_base="http://proxy.example.com", member_name="my-remote"):
return _rewrite_urls(urls, base_url, proxy_base, member_name)
def test_absolute_url_matching_base_is_rewritten(self):
result = self._rewrite(["https://upstream.example.com/chart-1.0.0.tgz"])
assert result == ["http://proxy.example.com/api/v1/remote/my-remote/chart-1.0.0.tgz"]
def test_relative_url_is_prepended_with_proxy_remote(self):
result = self._rewrite(["chart-1.0.0.tgz"])
assert result == ["http://proxy.example.com/api/v1/remote/my-remote/chart-1.0.0.tgz"]
def test_relative_url_with_leading_slash(self):
result = self._rewrite(["/chart-1.0.0.tgz"])
assert result == ["http://proxy.example.com/api/v1/remote/my-remote/chart-1.0.0.tgz"]
def test_absolute_url_not_matching_base_is_unchanged(self):
result = self._rewrite(["https://other.example.com/chart-1.0.0.tgz"])
assert result == ["https://other.example.com/chart-1.0.0.tgz"]
def test_empty_url_list_returns_empty(self):
assert self._rewrite([]) == []
def test_multiple_urls_all_rewritten(self):
urls = ["https://upstream.example.com/a-1.0.0.tgz", "b-2.0.0.tgz"]
result = self._rewrite(urls)
assert result[0] == "http://proxy.example.com/api/v1/remote/my-remote/a-1.0.0.tgz"
assert result[1] == "http://proxy.example.com/api/v1/remote/my-remote/b-2.0.0.tgz"
# ---------------------------------------------------------------------------
# _merge_helm_indexes
# ---------------------------------------------------------------------------
@@ -167,8 +240,7 @@ class TestHandlersRegistry:
class TestMergeHelmIndexes:
def _merge(self, raw_indexes, member_names, member_configs, proxy_base="http://proxy.example.com"):
with patch("artifactapi.artifact.virtual._helm.resolve_content", side_effect=_identity_resolve):
return _merge_helm_indexes(raw_indexes, member_names, member_configs, proxy_base)
return _merge_helm_indexes(raw_indexes, [None] * len(raw_indexes), member_names, member_configs, proxy_base)
def _parse(self, raw):
return yaml.safe_load(raw)
@@ -187,7 +259,18 @@ class TestMergeHelmIndexes:
def test_first_member_wins_on_duplicate_name_and_version(self):
index = self._parse(self._merge([_INDEX_A, _INDEX_B], ["member-a", "member-b"], [_CFG_A, _CFG_B]))
v027 = next(e for e in index["entries"]["vault"] if e["version"] == "0.27.0")
assert "helm.releases.hashicorp.com" in v027["urls"][0]
assert "member-a" in v027["urls"][0]
def test_absolute_urls_rewritten_to_proxy(self):
index = self._parse(self._merge([_INDEX_A], ["member-a"], [_CFG_A]))
url = index["entries"]["vault"][0]["urls"][0]
assert url == "http://proxy.example.com/api/v1/remote/member-a/vault-0.27.0.tgz"
def test_relative_urls_rewritten_to_proxy(self):
cfg = {"base_url": "https://releases.rancher.com/server-charts/stable", "cache": {"mutable_ttl": 3600}}
index = self._parse(self._merge([_INDEX_RELATIVE], ["rancher-stable"], [cfg]))
url = index["entries"]["rancher"][0]["urls"][0]
assert url == "http://proxy.example.com/api/v1/remote/rancher-stable/rancher-2.13.1.tgz"
def test_different_versions_of_same_chart_both_included(self):
index = self._parse(self._merge([_INDEX_A, _INDEX_B], ["member-a", "member-b"], [_CFG_A, _CFG_B]))
@@ -260,7 +343,7 @@ class TestGetMemberIndex:
storage.exists.return_value = True
cache.is_index_valid.return_value = True
_, _, _, raw_data = await _get_member_index("m", member_cfg, "index.yaml", storage, cache)
_, _, _, raw_data, _ = await _get_member_index("m", member_cfg, "index.yaml", storage, cache)
assert raw_data == b"cached bytes"
@@ -283,7 +366,7 @@ class TestGetMemberIndex:
mock_cls.return_value.__aenter__.return_value = mock_client
mock_client.get.return_value = self._fake_response(b"fresh bytes")
_, _, _, raw_data = await _get_member_index("m", member_cfg, "index.yaml", storage, cache)
_, _, _, raw_data, _ = await _get_member_index("m", member_cfg, "index.yaml", storage, cache)
assert raw_data == b"fresh bytes"
@@ -293,7 +376,7 @@ class TestGetMemberIndex:
mock_cls.return_value.__aenter__.return_value = mock_client
mock_client.get.return_value = self._fake_response()
_, _, _, raw_data = await _get_member_index("m", member_cfg, "index.yaml", storage, cache)
_, _, _, raw_data, _ = await _get_member_index("m", member_cfg, "index.yaml", storage, cache)
assert raw_data == b"upstream bytes"
@@ -352,7 +435,7 @@ class TestGetMemberIndex:
mock_cls.return_value.__aenter__.return_value = mock_client
mock_client.get.side_effect = Exception("connection refused")
_, _, _, raw_data = await _get_member_index("m", member_cfg, "index.yaml", storage, cache)
_, _, _, raw_data, _ = await _get_member_index("m", member_cfg, "index.yaml", storage, cache)
assert raw_data is None
@@ -364,7 +447,7 @@ class TestGetMemberIndex:
mock_cls.return_value.__aenter__.return_value = mock_client
mock_client.get.return_value = self._fake_response()
_, _, _, raw_data = await _get_member_index("m", member_cfg, "index.yaml", storage, cache)
_, _, _, raw_data, _ = await _get_member_index("m", member_cfg, "index.yaml", storage, cache)
assert raw_data == b"upstream bytes"
@@ -375,7 +458,7 @@ class TestGetMemberIndex:
mock_cls.return_value.__aenter__.return_value = mock_client
mock_client.get.return_value = self._fake_response()
_, _, ttl, _ = await _get_member_index("m", cfg, "index.yaml", storage, cache)
_, _, ttl, _, _ = await _get_member_index("m", cfg, "index.yaml", storage, cache)
assert ttl == 900
@@ -386,7 +469,7 @@ class TestGetMemberIndex:
mock_cls.return_value.__aenter__.return_value = mock_client
mock_client.get.return_value = self._fake_response()
_, _, ttl, _ = await _get_member_index("m", cfg, "index.yaml", storage, cache)
_, _, ttl, _, _ = await _get_member_index("m", cfg, "index.yaml", storage, cache)
assert ttl == 3600
@@ -430,10 +513,10 @@ class TestVirtualRoute:
response = client.get("/api/v1/virtual/no-such-virtual/index.yaml")
assert response.status_code == 404
def test_non_virtual_type_returns_400(self, client, patched_virtual_deps):
# helm-test is type "remote", not "virtual"
def test_non_virtual_name_returns_404(self, client, patched_virtual_deps):
# helm-test is in remotes, not virtuals
response = client.get("/api/v1/virtual/helm-test/index.yaml")
assert response.status_code == 400
assert response.status_code == 404
def test_unsupported_package_returns_400(self, client, patched_virtual_deps):
# unsupported-virtual-test has package "rpm"
@@ -484,22 +567,16 @@ class TestVirtualRoute:
mock_get.assert_not_called()
def test_cache_miss_returns_200_with_yaml_content_type(self, client, patched_virtual_deps):
with (
patch("artifactapi.artifact.virtual._get_member_index", new_callable=AsyncMock) as mock_get,
patch("artifactapi.artifact.virtual._helm.resolve_content", side_effect=_identity_resolve),
):
mock_get.return_value = ("helm-test", _CFG_A, 3600, _INDEX_SIMPLE)
with patch("artifactapi.artifact.virtual._get_member_index", new_callable=AsyncMock) as mock_get:
mock_get.return_value = ("helm-test", _CFG_A, 3600, _INDEX_SIMPLE, None)
response = client.get("/api/v1/virtual/helm-virtual-test/index.yaml")
assert response.status_code == 200
assert "text/yaml" in response.headers["content-type"]
def test_cache_miss_response_contains_merged_entries(self, client, patched_virtual_deps):
with (
patch("artifactapi.artifact.virtual._get_member_index", new_callable=AsyncMock) as mock_get,
patch("artifactapi.artifact.virtual._helm.resolve_content", side_effect=_identity_resolve),
):
mock_get.return_value = ("helm-test", _CFG_A, 3600, _INDEX_SIMPLE)
with patch("artifactapi.artifact.virtual._get_member_index", new_callable=AsyncMock) as mock_get:
mock_get.return_value = ("helm-test", _CFG_A, 3600, _INDEX_SIMPLE, None)
response = client.get("/api/v1/virtual/helm-virtual-test/index.yaml")
index = yaml.safe_load(response.content)
@@ -507,35 +584,26 @@ class TestVirtualRoute:
def test_cache_miss_stores_result_in_s3(self, client, patched_virtual_deps):
deps = patched_virtual_deps
with (
patch("artifactapi.artifact.virtual._get_member_index", new_callable=AsyncMock) as mock_get,
patch("artifactapi.artifact.virtual._helm.resolve_content", side_effect=_identity_resolve),
):
mock_get.return_value = ("helm-test", _CFG_A, 3600, _INDEX_SIMPLE)
with patch("artifactapi.artifact.virtual._get_member_index", new_callable=AsyncMock) as mock_get:
mock_get.return_value = ("helm-test", _CFG_A, 3600, _INDEX_SIMPLE, None)
client.get("/api/v1/virtual/helm-virtual-test/index.yaml")
deps["storage"].upload.assert_called_once()
def test_cache_miss_marks_index_cached(self, client, patched_virtual_deps):
deps = patched_virtual_deps
with (
patch("artifactapi.artifact.virtual._get_member_index", new_callable=AsyncMock) as mock_get,
patch("artifactapi.artifact.virtual._helm.resolve_content", side_effect=_identity_resolve),
):
mock_get.return_value = ("helm-test", _CFG_A, 3600, _INDEX_SIMPLE)
with patch("artifactapi.artifact.virtual._get_member_index", new_callable=AsyncMock) as mock_get:
mock_get.return_value = ("helm-test", _CFG_A, 3600, _INDEX_SIMPLE, None)
client.get("/api/v1/virtual/helm-virtual-test/index.yaml")
deps["cache"].mark_index_cached.assert_called_once()
def test_cache_miss_uses_min_ttl_across_members(self, client, patched_virtual_deps):
deps = patched_virtual_deps
with (
patch("artifactapi.artifact.virtual._get_member_index", new_callable=AsyncMock) as mock_get,
patch("artifactapi.artifact.virtual._helm.resolve_content", side_effect=_identity_resolve),
):
with patch("artifactapi.artifact.virtual._get_member_index", new_callable=AsyncMock) as mock_get:
mock_get.side_effect = [
("helm-test", _CFG_A, 3600, _INDEX_SIMPLE),
("helm-member-2", _CFG_B, 1800, _INDEX_SIMPLE),
("helm-test", _CFG_A, 3600, _INDEX_SIMPLE, None),
("helm-member-2", _CFG_B, 1800, _INDEX_SIMPLE, None),
]
client.get("/api/v1/virtual/helm-virtual-test/index.yaml")
@@ -544,19 +612,16 @@ class TestVirtualRoute:
def test_all_members_unreachable_returns_502(self, client, patched_virtual_deps):
with patch("artifactapi.artifact.virtual._get_member_index", new_callable=AsyncMock) as mock_get:
mock_get.return_value = ("helm-test", _CFG_A, 3600, None)
mock_get.return_value = ("helm-test", _CFG_A, 3600, None, None)
response = client.get("/api/v1/virtual/helm-virtual-test/index.yaml")
assert response.status_code == 502
def test_one_member_unreachable_still_returns_200(self, client, patched_virtual_deps):
with (
patch("artifactapi.artifact.virtual._get_member_index", new_callable=AsyncMock) as mock_get,
patch("artifactapi.artifact.virtual._helm.resolve_content", side_effect=_identity_resolve),
):
with patch("artifactapi.artifact.virtual._get_member_index", new_callable=AsyncMock) as mock_get:
mock_get.side_effect = [
("helm-test", _CFG_A, 3600, _INDEX_SIMPLE),
("helm-member-2", _CFG_B, 1800, None),
("helm-test", _CFG_A, 3600, _INDEX_SIMPLE, None),
("helm-member-2", _CFG_B, 1800, None, None),
]
response = client.get("/api/v1/virtual/helm-virtual-test/index.yaml")
@@ -572,10 +637,9 @@ class TestVirtualRoute:
with (
patch("artifactapi.artifact.virtual._get_member_index", new_callable=AsyncMock) as mock_get,
patch("artifactapi.artifact.virtual._helm.resolve_content", side_effect=_identity_resolve),
patch.object(main_mod.config, "get_remote_config", side_effect=patched_get),
):
mock_get.return_value = ("helm-test", _CFG_A, 3600, _INDEX_SIMPLE)
mock_get.return_value = ("helm-test", _CFG_A, 3600, _INDEX_SIMPLE, None)
response = client.get("/api/v1/virtual/helm-virtual-test/index.yaml")
# only helm-test was available — should succeed
@@ -586,11 +650,181 @@ class TestVirtualRoute:
deps = patched_virtual_deps
deps["storage"].upload.side_effect = Exception("S3 write error")
with (
patch("artifactapi.artifact.virtual._get_member_index", new_callable=AsyncMock) as mock_get,
patch("artifactapi.artifact.virtual._helm.resolve_content", side_effect=_identity_resolve),
):
mock_get.return_value = ("helm-test", _CFG_A, 3600, _INDEX_SIMPLE)
with patch("artifactapi.artifact.virtual._get_member_index", new_callable=AsyncMock) as mock_get:
mock_get.return_value = ("helm-test", _CFG_A, 3600, _INDEX_SIMPLE, None)
response = client.get("/api/v1/virtual/helm-virtual-test/index.yaml")
assert response.status_code == 200
# ---------------------------------------------------------------------------
# _entries_to_msgpack_safe
# ---------------------------------------------------------------------------
class TestEntriesToMsgpackSafe:
def test_plain_string_values_pass_through(self):
entries = {"chart": [{"name": "chart", "version": "1.0.0", "urls": ["http://x/c.tgz"]}]}
result = _entries_to_msgpack_safe(entries)
assert result["chart"][0]["version"] == "1.0.0"
def test_datetime_converted_to_iso_string(self):
dt = datetime(2023, 6, 15, 12, 0, 0, tzinfo=UTC)
entries = {"chart": [{"name": "chart", "version": "1.0.0", "created": dt}]}
result = _entries_to_msgpack_safe(entries)
assert isinstance(result["chart"][0]["created"], str)
assert "2023-06-15" in result["chart"][0]["created"]
def test_date_converted_to_iso_string(self):
entries = {"chart": [{"name": "chart", "version": "1.0.0", "created": date(2023, 6, 15)}]}
result = _entries_to_msgpack_safe(entries)
assert result["chart"][0]["created"] == "2023-06-15"
def test_empty_entries_returns_empty_dict(self):
assert _entries_to_msgpack_safe({}) == {}
def test_multiple_versions_all_converted(self):
dt = datetime(2023, 1, 1, tzinfo=UTC)
entries = {
"chart": [
{"name": "chart", "version": "1.0.0", "created": dt},
{"name": "chart", "version": "2.0.0", "created": dt},
]
}
result = _entries_to_msgpack_safe(entries)
for v in result["chart"]:
assert isinstance(v["created"], str)
def test_result_is_msgpack_serializable(self):
import msgpack
dt = datetime(2023, 6, 15, 12, 0, 0, tzinfo=UTC)
entries = {"chart": [{"name": "chart", "version": "1.0.0", "created": dt, "urls": ["http://x/c.tgz"]}]}
safe = _entries_to_msgpack_safe(entries)
packed = msgpack.packb(safe, use_bin_type=True)
unpacked = msgpack.unpackb(packed, raw=False)
assert unpacked["chart"][0]["created"] == safe["chart"][0]["created"]
# ---------------------------------------------------------------------------
# _merge_helm_indexes — pre-parsed entries path
# ---------------------------------------------------------------------------
class TestMergeHelmIndexesWithParsed:
"""Verify that pre-parsed entries (from msgpack) produce the same output as raw YAML."""
def _parse_entries(self, raw: bytes) -> dict:
index = yaml.safe_load(raw)
return index.get("entries") or {}
def test_parsed_entries_produce_same_charts_as_raw(self):
parsed = self._parse_entries(_INDEX_A)
raw_result = yaml.safe_load(_merge_helm_indexes([_INDEX_A], [None], ["member-a"], [_CFG_A], "http://proxy.example.com"))
parsed_result = yaml.safe_load(_merge_helm_indexes([_INDEX_A], [parsed], ["member-a"], [_CFG_A], "http://proxy.example.com"))
assert set(raw_result["entries"].keys()) == set(parsed_result["entries"].keys())
def test_parsed_entries_urls_are_rewritten(self):
parsed = self._parse_entries(_INDEX_A)
result = yaml.safe_load(_merge_helm_indexes([_INDEX_A], [parsed], ["member-a"], [_CFG_A], "http://proxy.example.com"))
url = result["entries"]["vault"][0]["urls"][0]
assert "member-a" in url
assert "proxy.example.com" in url
def test_none_parsed_falls_back_to_raw_bytes(self):
result = yaml.safe_load(_merge_helm_indexes([_INDEX_A], [None], ["member-a"], [_CFG_A], "http://proxy.example.com"))
assert "vault" in result["entries"]
def test_mixed_parsed_and_raw_merge_correctly(self):
parsed_a = self._parse_entries(_INDEX_A)
result = yaml.safe_load(
_merge_helm_indexes(
[_INDEX_A, _INDEX_B],
[parsed_a, None],
["member-a", "member-b"],
[_CFG_A, _CFG_B],
"http://proxy.example.com",
)
)
assert "vault" in result["entries"]
assert "nginx" in result["entries"]
# ---------------------------------------------------------------------------
# _get_member_index — msgpack cache behaviour
# ---------------------------------------------------------------------------
class TestGetMemberIndexMsgpack:
@pytest.fixture
def storage(self):
m = MagicMock()
m.get_object_key.side_effect = lambda name, path: f"{name}/{path}"
m.exists.return_value = False
m.download_object.return_value = _INDEX_SIMPLE
return m
@pytest.fixture
def cache(self):
m = MagicMock()
m.is_index_valid.return_value = False
return m
@pytest.fixture
def member_cfg(self):
return {"base_url": "https://helm.releases.hashicorp.com", "cache": {"mutable_ttl": 3600}}
def _fake_response(self, content=_INDEX_SIMPLE):
r = MagicMock()
r.content = content
r.raise_for_status = MagicMock()
return r
async def test_cache_hit_with_msgpack_returns_parsed_entries(self, storage, cache, member_cfg):
import msgpack
entries = {"mychart": [{"name": "mychart", "version": "1.0.0", "urls": ["http://x/c.tgz"]}]}
packed = msgpack.packb(entries, use_bin_type=True)
storage.exists.side_effect = lambda key: True
cache.is_index_valid.return_value = True
storage.download_object.side_effect = lambda key: packed if key.endswith("index.msgpack") else _INDEX_SIMPLE
_, _, _, raw_data, parsed = await _get_member_index("m", member_cfg, "index.yaml", storage, cache)
assert parsed == entries
async def test_cache_miss_builds_msgpack_and_returns_parsed(self, storage, cache, member_cfg):
with patch("artifactapi.artifact.virtual.httpx.AsyncClient") as mock_cls:
mock_client = AsyncMock()
mock_cls.return_value.__aenter__.return_value = mock_client
mock_client.get.return_value = self._fake_response()
_, _, _, raw_data, parsed = await _get_member_index("m", member_cfg, "index.yaml", storage, cache)
assert raw_data == _INDEX_SIMPLE
assert isinstance(parsed, dict)
assert "mychart" in parsed
async def test_broken_msgpack_rebuilds_from_raw_yaml(self, storage, cache, member_cfg):
storage.exists.side_effect = lambda key: True
cache.is_index_valid.return_value = True
storage.download_object.side_effect = lambda key: b"not-valid-msgpack" if key.endswith("index.msgpack") else _INDEX_SIMPLE
_, _, _, raw_data, parsed = await _get_member_index("m", member_cfg, "index.yaml", storage, cache)
assert raw_data == _INDEX_SIMPLE
# Falls back to YAML parse and rebuilds msgpack — entries are returned
assert isinstance(parsed, dict)
assert "mychart" in parsed
async def test_upstream_failure_returns_none_for_both(self, storage, cache, member_cfg):
with patch("artifactapi.artifact.virtual.httpx.AsyncClient") as mock_cls:
mock_client = AsyncMock()
mock_cls.return_value.__aenter__.return_value = mock_client
mock_client.get.side_effect = Exception("timeout")
_, _, _, raw_data, parsed = await _get_member_index("m", member_cfg, "index.yaml", storage, cache)
assert raw_data is None
assert parsed is None