Files
RustProxy/scripts/tests/cache_age_transform_034_live.py
T
2026-10-17 23:37:43 +02:00

213 lines
11 KiB
Python
Executable File

#!/usr/bin/env python3
"""Loopback integration regression: transformed Age, HOT/COLD, diagnostics, isolation.
Run after compiling:
python3 scripts/tests/cache_age_transform_034_live.py --binary target/release/rust-proxy
To reproduce the original regression with the user-supplied 0.34.0 binary:
python3 scripts/tests/cache_age_transform_034_live.py --binary /path/to/old/rust-proxy --expect-old-age-bug
No internet and no user data; test config/data/passwords are in a disposable temporary dir.
"""
import argparse
import collections
import copy
import json
import sys
import threading
import time
from http.server import BaseHTTPRequestHandler, ThreadingHTTPServer
from pathlib import Path
sys.path.insert(0, str(Path(__file__).resolve().parents[1] / 'performance'))
from common import Fixture, http, sha256, write_json
BODY = b'/* test cache */\nbody { color: red; padding: 0px; }\n'
# Fixed sample from Brotli encoder, embedded to avoid Python dependency on brotli.
BROTLI_BODY = b"\x1b3\x00\xe8\x1d\x07\xee\xbe\r\xa2\xbc3'\x8cP\xe9\xad\xa0\xf2\xb2\xa7\xe4\xfa\x12\x84\xac\x0c\x1cr\xc0\xfe\xc0>\x0fR*a\x9c\x9e2\xe8\x08\xcb\x1cP\x881\xac\xdd\x88s5(\x9d\x01"
COUNTS = collections.Counter()
class Origin(BaseHTTPRequestHandler):
def log_message(self, *_):
pass
def do_GET(self):
path = self.path.split('?', 1)[0]
if path not in {'/aged.css', '/normal.css', '/private.css', '/no-store.css', '/other.css', '/br.css', '/aged-br.css'}:
self.send_error(404)
return
COUNTS[path] += 1
self.send_response(200)
if path in ('/aged.css', '/aged-br.css'):
policy = 'public, max-age=2592000'
elif path == '/private.css':
policy = 'private, max-age=2592000'
elif path == '/no-store.css':
policy = 'no-store'
else:
policy = 'public, max-age=3600'
self.send_header('Cache-Control', policy)
self.send_header('Content-Type', 'text/css')
payload = BROTLI_BODY if path in ('/br.css', '/aged-br.css') else BODY
self.send_header('Content-Length', str(len(payload)))
if path in ('/br.css', '/aged-br.css'):
self.send_header('Content-Encoding', 'br')
self.send_header('ETag', '"origin-css"')
self.send_header('Last-Modified', 'Thu, 07 Jan 2016 20:04:59 GMT')
self.send_header('Vary', 'Accept-Encoding')
if path in ('/aged.css', '/aged-br.css'):
self.send_header('Age', '1553483')
self.end_headers()
self.wfile.write(payload)
def request(port, path, encoding='identity'):
status, headers, body = http(port, path, headers={'Accept-Encoding': encoding})
return {'status': status, 'x_cache': headers.get('x-cache'),
'x_cache_hits': headers.get('x-cache-hits'), 'x_cache_resident_age': headers.get('x-cache-resident-age'),
'cache_status': headers.get('cache-status'), 'age': headers.get('age'),
'body': body.decode('utf-8', 'replace'), 'content_length': headers.get('content-length'),
'etag': headers.get('etag'), 'encoding': headers.get('content-encoding'), 'origin_count': COUNTS[path.split('?', 1)[0]]}
def dash(rust):
status, _, body = rust.admin('/api/cache?limit=200')
assert status == 200, ('cache dashboard', status, body[:200])
data = json.loads(body)['dashboard']
return {'objects': [{k: row.get(k) for k in ('vhost','path','tier','transformed',
'hits','age_seconds','resident_age_seconds','remaining_ttl_seconds')} for row in data['objects']],
'domains': data['domains']}
def find(d, path):
matches = [item for item in d['objects'] if item['path'] == path and item['vhost'] == 'reverse']
assert len(matches) == 1, (path, matches)
return matches[0]
def run(binary: Path, output: Path, expect_old_age_bug: bool):
outcome = {'binary_sha256': sha256(binary), 'expect_old_age_bug': expect_old_age_bug,
'tests': [], 'passed': False}
output.mkdir(parents=True, exist_ok=True)
with Fixture(output, workers=2) as fixture:
origin = ThreadingHTTPServer(('127.0.0.1', fixture.origin_port), Origin)
thread = threading.Thread(target=origin.serve_forever, daemon=True)
thread.start()
try:
rust = fixture.rust(binary, 'cache-age-034')
rust.login()
cfg = copy.deepcopy(rust.base)
cfg['cache'] = {'enabled': True, 'max_ttl_seconds': 86_400,
'expose_status': True, 'expose_hits': True,
'expose_cache_status': True, 'expose_resident_age': True}
cfg['transform'] = {'enabled': True, 'css': True}
for vhost in cfg['vhosts']:
vhost['transform'] = {'enabled': True, 'css': True}
rust.apply(cfg, 'fresh-cache-transform')
for path in ('/normal.css', '/aged.css', '/br.css', '/aged-br.css', '/private.css', '/no-store.css'):
before = COUNTS[path]
encoding = 'br' if path in ('/br.css', '/aged-br.css') else 'identity'
first, second = request(rust.reverse_port, path, encoding), request(rust.reverse_port, path, encoding)
row = {'path': path, 'first': first, 'second': second,
'origin_delta': COUNTS[path] - before}
outcome['tests'].append(row)
if path in ('/aged.css', '/aged-br.css') and expect_old_age_bug:
assert first['x_cache'] == second['x_cache'] == 'MISS', row
assert row['origin_delta'] == 2, row
elif path in ('/normal.css', '/aged.css', '/br.css', '/aged-br.css'):
assert first['x_cache'] == 'MISS' and second['x_cache'] == 'HIT', row
assert row['origin_delta'] == 1, row
assert second['cache_status'] and 'transformed' in second['cache_status'], row
assert first['etag'] == second['etag'] and first['body'] == second['body'], row
else:
assert first['x_cache'] == second['x_cache'] == 'MISS', row
print(path, first['x_cache'], second['x_cache'], 'origin+', row['origin_delta'], flush=True)
snapshot = dash(rust)
outcome['tests'].append({'dashboard': snapshot})
assert find(snapshot, '/normal.css')['transformed'], snapshot
aged = find(snapshot, '/aged.css')
if expect_old_age_bug:
assert aged['remaining_ttl_seconds'] == 0, aged
else:
assert aged['transformed'] and aged['remaining_ttl_seconds'] >= 86_000, aged
assert aged['age_seconds'] >= 1_553_483, aged
assert aged['resident_age_seconds'] < 10, aged
aged_br = find(snapshot, '/aged-br.css')
assert aged_br['transformed'] and aged_br['remaining_ttl_seconds'] >= 86_000, aged_br
# Query-string bypass is an explicit administrator rule. Even a
# transformed CSS with public max-age stays BYPASS while it is ON,
# then the *same* URL must become MISS -> HIT after disabling it.
# This is not a transform regression: it is a request policy.
cfg['cache']['bypass_query'] = True
rust.apply(cfg, 'query-bypass-enabled')
query_path = '/normal.css?v=transformed-public'
before = COUNTS['/normal.css']
q_bypass = [request(rust.reverse_port, query_path) for _ in range(2)]
assert [row['x_cache'] for row in q_bypass] == ['BYPASS', 'BYPASS'], q_bypass
assert COUNTS['/normal.css'] - before == 2, q_bypass
cfg['cache']['bypass_query'] = False
rust.apply(cfg, 'query-bypass-disabled')
q_cached = [request(rust.reverse_port, query_path) for _ in range(2)]
assert [row['x_cache'] for row in q_cached] == ['MISS', 'HIT'], q_cached
assert q_cached[1]['cache_status'] and 'transformed' in q_cached[1]['cache_status']
assert COUNTS['/normal.css'] - before == 3, q_cached
outcome['tests'].append({'query_bypass_policy': {'bypass': q_bypass, 'cached': q_cached}})
# Operational LB tuning must not invalidate a cached representation.
before = COUNTS['/normal.css']
cfg['vhosts'][0]['load_balancing']['keepalive'] = {'max_idle_per_host': 16}
rust.apply(cfg, 'only-backend-keepalive-changed')
normal = request(rust.reverse_port, '/normal.css')
assert normal['x_cache'] == 'HIT' and COUNTS['/normal.css'] == before
outcome['tests'].append({'keepalive_update_without_purge': normal})
# Diagnostics can be intentionally hidden on MISS and remain visible on HIT.
cfg['cache']['expose_only_hits'] = True
rust.apply(cfg, 'only-show-hit-headers')
before = COUNTS['/other.css']
first, second = request(rust.reverse_port, '/other.css'), request(rust.reverse_port, '/other.css')
assert first['x_cache'] is None and second['x_cache'] == 'HIT'
assert COUNTS['/other.css'] - before == 1
outcome['tests'].append({'only_hits': [first, second]})
if not expect_old_age_bug:
# Cold path on a 1-entry HOT tier: demotion and on-disk reuse.
cfg['cache'].update({'max_entries': 1, 'cold_enabled': True, 'cold_on_lru': True,
'cold_path': 'cache/age-regression-cold'})
cfg['cache'].pop('expose_only_hits', None)
rust.apply(cfg, 'cold-enabled-one-hot-entry')
assert request(rust.reverse_port, '/aged.css')['x_cache'] == 'MISS'
assert request(rust.reverse_port, '/aged.css')['x_cache'] == 'HIT'
assert request(rust.reverse_port, '/other.css')['x_cache'] == 'MISS'
deadline = time.monotonic() + 4
while True:
snapshot = dash(rust)
matches = [x for x in snapshot['objects'] if x['path'] == '/aged.css' and x['tier'] == 'cold']
if matches:
assert matches[0]['remaining_ttl_seconds'] >= 86_000, matches
break
if time.monotonic() > deadline:
raise AssertionError('aged transformed CSS never demoted to COLD: ' + str(snapshot))
time.sleep(.05)
cold_hit = request(rust.reverse_port, '/aged.css')
assert cold_hit['x_cache'] == 'HIT' and 'transformed' in cold_hit['cache_status'], cold_hit
outcome['tests'].append({'cold': {'row': matches[0], 'hit': cold_hit}})
outcome['passed'] = True
finally:
origin.shutdown()
origin.server_close()
write_json(output / 'cache-age-transform-result.json', outcome)
return outcome
if __name__ == '__main__':
parser = argparse.ArgumentParser(description=__doc__)
parser.add_argument('--binary', type=Path, required=True)
parser.add_argument('--output', type=Path, default=Path('/tmp/rustproxy-cache-age-034-live'))
parser.add_argument('--expect-old-age-bug', action='store_true')
args = parser.parse_args()
try:
report = run(args.binary.resolve(), args.output.resolve(), args.expect_old_age_bug)
except Exception as error:
print('FAIL', type(error).__name__, str(error)[:1000], file=sys.stderr)
raise
print('PASS', len(report['tests']), 'scenarios; see', args.output / 'cache-age-transform-result.json')