Proxy random media with the token in the query, and bootstrap before listening !28
11 files changed, +121 −282
Layout: unified · split
SETUP.md −5
| @@ -35,11 +35,6 @@ below apply. A file named with `-c` must exist. | |||
| 35 | recently used entries are dropped past this. | 35 | recently used entries are dropped past this. |
| 36 | * `ttl` — How long a response is reused, in the time units below. Default | 36 | * `ttl` — How long a response is reused, in the time units below. Default |
| 37 | `5i`. | 37 | `5i`. |
| 38 | * `stale` — How long past `ttl` a response is kept to be served when | ||
| 39 | DeviantArt fails or blocks the instance, so a short ban does not take | ||
| 40 | the daily deviations, popular searches and feeds down. Default `1h`; | ||
| 41 | `0i` keeps nothing past `ttl`. After a block the instance also leaves | ||
| 42 | DeviantArt alone for a minute instead of retrying every request. | ||
| 43 | * `rate-limit` — Per-client budget for page, feed and API requests, so one | 38 | * `rate-limit` — Per-client budget for page, feed and API requests, so one |
| 44 | crawler cannot spend the whole upstream budget. Media, avatars and static | 39 | crawler cannot spend the whole upstream budget. Media, avatars and static |
| 45 | files are not counted. Over budget answers 429 with `Retry-After`. | 40 | files are not counted. Over budget answers 429 with `Retry-After`. |
app/api.go +15 −6
| @@ -3,6 +3,7 @@ package app | |||
| 3 | import ( | 3 | import ( |
| 4 | "encoding/json" | 4 | "encoding/json" |
| 5 | "math/rand" | 5 | "math/rand" |
| 6 | "net/url" | ||
| 6 | "strings" | 7 | "strings" |
| 7 | 8 | ||
| 8 | "github.com/krazywarez/devianter" | 9 | "github.com/krazywarez/devianter" |
| @@ -52,15 +53,23 @@ func (a API) sendMedia(d *devianter.Deviation) { | |||
| 52 | return | 53 | return |
| 53 | } | 54 | } |
| 54 | 55 | ||
| 55 | if CFG.Proxy { | 56 | if !CFG.Proxy { |
| 56 | mediaURL = mediaURL[21:] | ||
| 57 | dot := strings.Index(mediaURL, ".") | ||
| 58 | a.main.Writer.Header().Del("Content-Type") | ||
| 59 | a.main.DownloadAndSendMedia(mediaURL[:dot], mediaURL[dot+11:]) | ||
| 60 | } else { | ||
| 61 | a.main.Writer.Header().Add("Location", mediaURL) | 57 | a.main.Writer.Header().Add("Location", mediaURL) |
| 62 | a.main.Writer.WriteHeader(302) | 58 | a.main.Writer.WriteHeader(302) |
| 59 | return | ||
| 60 | } | ||
| 61 | |||
| 62 | // Parsed, not sliced: the signing token has to reach wixmp as a query | ||
| 63 | // parameter. Passing the raw tail as the path put "?token=..." inside | ||
| 64 | // the path, which wixmp answers with 401. | ||
| 65 | u, err := url.Parse(mediaURL) | ||
| 66 | if err != nil { | ||
| 67 | a.Error("bad media url", 502) | ||
| 68 | return | ||
| 63 | } | 69 | } |
| 70 | subdomain := strings.TrimSuffix(strings.TrimPrefix(u.Host, "images-wixmp-"), ".wixmp.com") | ||
| 71 | a.main.Writer.Header().Del("Content-Type") | ||
| 72 | a.main.downloadAndSendMedia(subdomain, strings.TrimPrefix(u.Path, "/"), u.Query().Get("token")) | ||
| 64 | } | 73 | } |
| 65 | 74 | ||
| 66 | // fetchDailyDeviations is devianter.GetDailyDeviations behind a variable so | 75 | // fetchDailyDeviations is devianter.GetDailyDeviations behind a variable so |
app/api_test.go +37
| @@ -1,7 +1,10 @@ | |||
| 1 | package app | 1 | package app |
| 2 | 2 | ||
| 3 | import ( | 3 | import ( |
| 4 | "net/http" | ||
| 4 | "net/http/httptest" | 5 | "net/http/httptest" |
| 6 | "net/url" | ||
| 7 | "strings" | ||
| 5 | "testing" | 8 | "testing" |
| 6 | 9 | ||
| 7 | "github.com/krazywarez/devianter" | 10 | "github.com/krazywarez/devianter" |
| @@ -108,3 +111,37 @@ func TestRandomHonoursNSFW(t *testing.T) { | |||
| 108 | t.Errorf("status %d, Location %q; want 404 and no media", w.Code, w.Header().Get("Location")) | 111 | t.Errorf("status %d, Location %q; want 404 and no media", w.Code, w.Header().Get("Location")) |
| 109 | } | 112 | } |
| 110 | } | 113 | } |
| 114 | |||
| 115 | // TestSendMediaProxiesWithTheTokenInTheQuery is the regression test for | ||
| 116 | // /api/random answering 401 with proxying on: the signing token was passed | ||
| 117 | // inside the path, so wixmp never saw it as a parameter. | ||
| 118 | func TestSendMediaProxiesWithTheTokenInTheQuery(t *testing.T) { | ||
| 119 | proxy, cache := CFG.Proxy, CFG.Cache.Enabled | ||
| 120 | CFG.Proxy, CFG.Cache.Enabled = true, false | ||
| 121 | defer func() { CFG.Proxy, CFG.Cache.Enabled = proxy, cache }() | ||
| 122 | |||
| 123 | var fetched string | ||
| 124 | orig := fetchMedia | ||
| 125 | fetchMedia = func(u string) Downloaded { | ||
| 126 | fetched = u | ||
| 127 | return Downloaded{Status: 200, Body: []byte("png"), Headers: http.Header{"Content-Type": {"image/png"}}} | ||
| 128 | } | ||
| 129 | defer func() { fetchMedia = orig }() | ||
| 130 | |||
| 131 | d := fullviewDeviation() | ||
| 132 | d.Media.Token = []string{"tok.en.sig"} | ||
| 133 | |||
| 134 | w := httptest.NewRecorder() | ||
| 135 | API{main: &skunkyart{Writer: w, Args: url.Values{}}}.sendMedia(d) | ||
| 136 | |||
| 137 | u, err := url.Parse(fetched) | ||
| 138 | if err != nil || u.Host != "images-wixmp-abc.wixmp.com" { | ||
| 139 | t.Fatalf("fetched %q, want a wixmp URL on the deviation's subdomain", fetched) | ||
| 140 | } | ||
| 141 | if strings.Contains(u.Path, "token") || u.Query().Get("token") == "" { | ||
| 142 | t.Errorf("token not passed as a query parameter: path %q query %q", u.Path, u.RawQuery) | ||
| 143 | } | ||
| 144 | if w.Code != 200 || w.Body.String() != "png" { | ||
| 145 | t.Errorf("response %d %q, want the proxied image", w.Code, w.Body.String()) | ||
| 146 | } | ||
| 147 | } | ||
app/apicache.go +28 −98
| @@ -21,31 +21,18 @@ import ( | |||
| 21 | type apiCache struct { | 21 | type apiCache struct { |
| 22 | maxBytes int64 | 22 | maxBytes int64 |
| 23 | ttl time.Duration | 23 | ttl time.Duration |
| 24 | // stale is how long past ttl an entry is kept to be served when upstream | 24 | now func() time.Time |
| 25 | // is blocked or unreachable. Zero keeps nothing past ttl. | ||
| 26 | stale time.Duration | ||
| 27 | now func() time.Time | ||
| 28 | 25 | ||
| 29 | mu sync.Mutex | 26 | mu sync.Mutex |
| 30 | entries map[string]*cacheEntry | 27 | entries map[string]*cacheEntry |
| 31 | lru *list.List // front is most recently used | 28 | lru *list.List // front is most recently used |
| 32 | held int64 | 29 | held int64 |
| 33 | hits int64 | 30 | hits int64 |
| 34 | misses int64 | 31 | misses int64 |
| 35 | staleHits int64 | ||
| 36 | |||
| 37 | // blockedUntil is set when DeviantArt answers with a block. Until then a | ||
| 38 | // miss with nothing stale to serve gets blockResp back without an | ||
| 39 | // upstream call, so a banned instance stops hammering the WAF. | ||
| 40 | blockedUntil time.Time | ||
| 41 | blockResp *cacheEntry | ||
| 42 | 32 | ||
| 43 | flight singleflight.Group | 33 | flight singleflight.Group |
| 44 | } | 34 | } |
| 45 | 35 | ||
| 46 | // blockBackoff is how long upstream is left alone after a block response. | ||
| 47 | const blockBackoff = time.Minute | ||
| 48 | |||
| 49 | // cacheEntry is one buffered response. header is a clone of the upstream | 36 | // cacheEntry is one buffered response. header is a clone of the upstream |
| 50 | // header; body is the whole body, read once. | 37 | // header; body is the whole body, read once. |
| 51 | type cacheEntry struct { | 38 | type cacheEntry struct { |
| @@ -57,11 +44,10 @@ type cacheEntry struct { | |||
| 57 | elem *list.Element | 44 | elem *list.Element |
| 58 | } | 45 | } |
| 59 | 46 | ||
| 60 | func newAPICache(maxBytes int64, ttl, stale time.Duration) *apiCache { | 47 | func newAPICache(maxBytes int64, ttl time.Duration) *apiCache { |
| 61 | return &apiCache{ | 48 | return &apiCache{ |
| 62 | maxBytes: maxBytes, | 49 | maxBytes: maxBytes, |
| 63 | ttl: ttl, | 50 | ttl: ttl, |
| 64 | stale: stale, | ||
| 65 | now: time.Now, | 51 | now: time.Now, |
| 66 | entries: map[string]*cacheEntry{}, | 52 | entries: map[string]*cacheEntry{}, |
| 67 | lru: list.New(), | 53 | lru: list.New(), |
| @@ -101,29 +87,16 @@ type cachedTransport struct { | |||
| 101 | base http.RoundTripper | 87 | base http.RoundTripper |
| 102 | } | 88 | } |
| 103 | 89 | ||
| 104 | // RoundTrip serves a fresh hit from memory. A miss is fetched once per key | 90 | // RoundTrip serves a hit from memory. A miss is fetched once per key however |
| 105 | // however many callers are waiting, buffered, stored if it is a 200, and | 91 | // many callers are waiting, buffered, stored if it is a 200, and handed to |
| 106 | // handed to every waiter as its own response. | 92 | // every waiter as its own response. |
| 107 | // | ||
| 108 | // When upstream fails or answers with a block, a stale entry is served | ||
| 109 | // instead if one is still held, so a short ban does not take the popular | ||
| 110 | // pages down. A block also starts a backoff during which misses with nothing | ||
| 111 | // stale get the block response back without an upstream call. | ||
| 112 | func (t *cachedTransport) RoundTrip(req *http.Request) (*http.Response, error) { | 93 | func (t *cachedTransport) RoundTrip(req *http.Request) (*http.Response, error) { |
| 113 | if !cacheable(req) { | 94 | if !cacheable(req) { |
| 114 | return t.base.RoundTrip(req) | 95 | return t.base.RoundTrip(req) |
| 115 | } | 96 | } |
| 116 | key := cacheKey(req) | 97 | key := cacheKey(req) |
| 117 | old, fresh := t.cache.get(key) | 98 | if e := t.cache.get(key); e != nil { |
| 118 | if fresh { | 99 | return e.response(req), nil |
| 119 | return old.response(req), nil | ||
| 120 | } | ||
| 121 | if blocked := t.cache.blockedResponse(); blocked != nil { | ||
| 122 | if old != nil { | ||
| 123 | t.cache.countStale() | ||
| 124 | return old.response(req), nil | ||
| 125 | } | ||
| 126 | return blocked.response(req), nil | ||
| 127 | } | 100 | } |
| 128 | 101 | ||
| 129 | v, err, _ := t.cache.flight.Do(key, func() (any, error) { | 102 | v, err, _ := t.cache.flight.Do(key, func() (any, error) { |
| @@ -137,57 +110,21 @@ func (t *cachedTransport) RoundTrip(req *http.Request) (*http.Response, error) { | |||
| 137 | return nil, err | 110 | return nil, err |
| 138 | } | 111 | } |
| 139 | e := &cacheEntry{key: key, status: resp.StatusCode, header: resp.Header.Clone(), body: body} | 112 | e := &cacheEntry{key: key, status: resp.StatusCode, header: resp.Header.Clone(), body: body} |
| 140 | switch e.status { | 113 | if e.status == http.StatusOK { |
| 141 | case http.StatusOK: | ||
| 142 | t.cache.put(e) | 114 | t.cache.put(e) |
| 143 | case http.StatusForbidden, http.StatusTooManyRequests: | ||
| 144 | t.cache.block(e) | ||
| 145 | } | 115 | } |
| 146 | return e, nil | 116 | return e, nil |
| 147 | }) | 117 | }) |
| 148 | if err != nil { | 118 | if err != nil { |
| 149 | if old != nil { | ||
| 150 | t.cache.countStale() | ||
| 151 | return old.response(req), nil | ||
| 152 | } | ||
| 153 | return nil, err | 119 | return nil, err |
| 154 | } | 120 | } |
| 155 | e, ok := v.(*cacheEntry) | 121 | e, ok := v.(*cacheEntry) |
| 156 | if !ok { | 122 | if !ok { |
| 157 | return nil, io.ErrUnexpectedEOF | 123 | return nil, io.ErrUnexpectedEOF |
| 158 | } | 124 | } |
| 159 | if e.status != http.StatusOK && old != nil { | ||
| 160 | t.cache.countStale() | ||
| 161 | return old.response(req), nil | ||
| 162 | } | ||
| 163 | return e.response(req), nil | 125 | return e.response(req), nil |
| 164 | } | 126 | } |
| 165 | 127 | ||
| 166 | // block records a block response and starts the backoff. | ||
| 167 | func (c *apiCache) block(e *cacheEntry) { | ||
| 168 | c.mu.Lock() | ||
| 169 | defer c.mu.Unlock() | ||
| 170 | c.blockedUntil = c.now().Add(blockBackoff) | ||
| 171 | c.blockResp = e | ||
| 172 | } | ||
| 173 | |||
| 174 | // blockedResponse returns the last block response while the backoff runs, | ||
| 175 | // or nil once it is over. | ||
| 176 | func (c *apiCache) blockedResponse() *cacheEntry { | ||
| 177 | c.mu.Lock() | ||
| 178 | defer c.mu.Unlock() | ||
| 179 | if c.blockResp != nil && c.now().Before(c.blockedUntil) { | ||
| 180 | return c.blockResp | ||
| 181 | } | ||
| 182 | return nil | ||
| 183 | } | ||
| 184 | |||
| 185 | func (c *apiCache) countStale() { | ||
| 186 | c.mu.Lock() | ||
| 187 | c.staleHits++ | ||
| 188 | c.mu.Unlock() | ||
| 189 | } | ||
| 190 | |||
| 191 | // response builds a fresh http.Response over the buffered body, so each | 128 | // response builds a fresh http.Response over the buffered body, so each |
| 192 | // caller can read and close its own. | 129 | // caller can read and close its own. |
| 193 | func (e *cacheEntry) response(req *http.Request) *http.Response { | 130 | func (e *cacheEntry) response(req *http.Request) *http.Response { |
| @@ -204,31 +141,25 @@ func (e *cacheEntry) response(req *http.Request) *http.Response { | |||
| 204 | } | 141 | } |
| 205 | } | 142 | } |
| 206 | 143 | ||
| 207 | // get returns the entry for key and whether it is still fresh. A fresh hit is | 144 | // get returns the live entry for key, marking it most recently used, or nil. |
| 208 | // marked most recently used. An entry past ttl but within the stale window is | 145 | // An expired entry is dropped on the way out. |
| 209 | // returned as not fresh, for the caller to fall back on; one past the stale | 146 | func (c *apiCache) get(key string) *cacheEntry { |
| 210 | // window is dropped. | ||
| 211 | func (c *apiCache) get(key string) (*cacheEntry, bool) { | ||
| 212 | c.mu.Lock() | 147 | c.mu.Lock() |
| 213 | defer c.mu.Unlock() | 148 | defer c.mu.Unlock() |
| 214 | 149 | ||
| 215 | e := c.entries[key] | 150 | e := c.entries[key] |
| 216 | if e == nil { | 151 | if e == nil { |
| 217 | c.misses++ | 152 | c.misses++ |
| 218 | return nil, false | 153 | return nil |
| 219 | } | 154 | } |
| 220 | now := c.now() | 155 | if !c.now().Before(e.expires) { |
| 221 | if now.Before(e.expires) { | 156 | c.remove(e) |
| 222 | c.lru.MoveToFront(e.elem) | 157 | c.misses++ |
| 223 | c.hits++ | 158 | return nil |
| 224 | return e, true | ||
| 225 | } | ||
| 226 | c.misses++ | ||
| 227 | if now.Before(e.expires.Add(c.stale)) { | ||
| 228 | return e, false | ||
| 229 | } | 159 | } |
| 230 | c.remove(e) | 160 | c.lru.MoveToFront(e.elem) |
| 231 | return nil, false | 161 | c.hits++ |
| 162 | return e | ||
| 232 | } | 163 | } |
| 233 | 164 | ||
| 234 | // put stores e, evicting from the least recently used end until it fits. A | 165 | // put stores e, evicting from the least recently used end until it fits. A |
| @@ -265,10 +196,9 @@ func (c *apiCache) remove(e *cacheEntry) { | |||
| 265 | c.held -= int64(len(e.body)) | 196 | c.held -= int64(len(e.body)) |
| 266 | } | 197 | } |
| 267 | 198 | ||
| 268 | // stats reports the counters for the hourly log line. stale counts the | 199 | // stats reports the counters for the hourly log line. |
| 269 | // misses that were answered from an expired entry because upstream failed. | 200 | func (c *apiCache) stats() (hits, misses int64, entries int, held int64) { |
| 270 | func (c *apiCache) stats() (hits, misses, stale int64, entries int, held int64) { | ||
| 271 | c.mu.Lock() | 201 | c.mu.Lock() |
| 272 | defer c.mu.Unlock() | 202 | defer c.mu.Unlock() |
| 273 | return c.hits, c.misses, c.staleHits, len(c.entries), c.held | 203 | return c.hits, c.misses, len(c.entries), c.held |
| 274 | } | 204 | } |
app/apicache_stale_test.go deleted −132
| @@ -1,132 +0,0 @@ | |||
| 1 | package app | ||
| 2 | |||
| 3 | import ( | ||
| 4 | "errors" | ||
| 5 | "net/http" | ||
| 6 | "testing" | ||
| 7 | "time" | ||
| 8 | ) | ||
| 9 | |||
| 10 | // failingRT is an upstream that can be switched between a scripted status and | ||
| 11 | // a transport error mid-test. | ||
| 12 | type failingRT struct { | ||
| 13 | fakeRT | ||
| 14 | err error | ||
| 15 | } | ||
| 16 | |||
| 17 | func (f *failingRT) RoundTrip(r *http.Request) (*http.Response, error) { | ||
| 18 | f.mu.Lock() | ||
| 19 | err := f.err | ||
| 20 | f.mu.Unlock() | ||
| 21 | if err != nil { | ||
| 22 | f.mu.Lock() | ||
| 23 | f.calls++ | ||
| 24 | f.mu.Unlock() | ||
| 25 | return nil, err | ||
| 26 | } | ||
| 27 | return f.fakeRT.RoundTrip(r) | ||
| 28 | } | ||
| 29 | |||
| 30 | func (f *failingRT) set(status int, body string, err error) { | ||
| 31 | f.mu.Lock() | ||
| 32 | defer f.mu.Unlock() | ||
| 33 | f.status, f.body, f.err = status, body, err | ||
| 34 | } | ||
| 35 | |||
| 36 | // staleCache returns a cache with a one minute TTL and one hour stale window | ||
| 37 | // over a controllable clock, warmed with one good response for puppyURL. | ||
| 38 | func staleCache(t *testing.T) (*apiCache, *failingRT, http.RoundTripper, *time.Time) { | ||
| 39 | t.Helper() | ||
| 40 | up := &failingRT{} | ||
| 41 | up.set(200, `{"good":1}`, nil) | ||
| 42 | c := newAPICache(1<<20, time.Minute, time.Hour) | ||
| 43 | now := time.Now() | ||
| 44 | c.now = func() time.Time { return now } | ||
| 45 | rt := c.transport(up) | ||
| 46 | get(t, rt, puppyURL) | ||
| 47 | return c, up, rt, &now | ||
| 48 | } | ||
| 49 | |||
| 50 | func TestStaleEntryServedWhenUpstreamBlocks(t *testing.T) { | ||
| 51 | c, up, rt, now := staleCache(t) | ||
| 52 | *now = now.Add(2 * time.Minute) // past ttl, inside the stale window | ||
| 53 | up.set(403, "<html>blocked</html>", nil) | ||
| 54 | |||
| 55 | status, body := get(t, rt, puppyURL) | ||
| 56 | |||
| 57 | if status != 200 || body != `{"good":1}` { | ||
| 58 | t.Errorf("got %d %q, want the stale 200 body", status, body) | ||
| 59 | } | ||
| 60 | if up.count() != 2 { | ||
| 61 | t.Errorf("upstream called %d times, want 2: one warm-up and one attempt that hit the block", up.count()) | ||
| 62 | } | ||
| 63 | if _, _, stale, _, _ := c.stats(); stale != 1 { | ||
| 64 | t.Errorf("stale counter is %d, want 1", stale) | ||
| 65 | } | ||
| 66 | } | ||
| 67 | |||
| 68 | func TestStaleEntryServedOnTransportError(t *testing.T) { | ||
| 69 | _, up, rt, now := staleCache(t) | ||
| 70 | *now = now.Add(2 * time.Minute) | ||
| 71 | up.set(0, "", errors.New("dial tcp: connection refused")) | ||
| 72 | |||
| 73 | status, body := get(t, rt, puppyURL) | ||
| 74 | |||
| 75 | if status != 200 || body != `{"good":1}` { | ||
| 76 | t.Errorf("got %d %q, want the stale 200 body", status, body) | ||
| 77 | } | ||
| 78 | } | ||
| 79 | |||
| 80 | func TestBlockBackoffSkipsUpstream(t *testing.T) { | ||
| 81 | _, up, rt, now := staleCache(t) | ||
| 82 | *now = now.Add(2 * time.Minute) | ||
| 83 | up.set(403, "<html>blocked</html>", nil) | ||
| 84 | get(t, rt, puppyURL) // triggers the block | ||
| 85 | calls := up.count() | ||
| 86 | |||
| 87 | other := "https://www.deviantart.com/_puppy/dabrowse/search/all?q=other" | ||
| 88 | status, body := get(t, rt, other) | ||
| 89 | if status != 403 || body != "<html>blocked</html>" { | ||
| 90 | t.Errorf("uncached key during backoff got %d %q, want the block response", status, body) | ||
| 91 | } | ||
| 92 | if up.count() != calls { | ||
| 93 | t.Errorf("upstream called during backoff (%d -> %d), want none", calls, up.count()) | ||
| 94 | } | ||
| 95 | |||
| 96 | *now = now.Add(blockBackoff + time.Second) | ||
| 97 | up.set(200, `{"back":1}`, nil) | ||
| 98 | if status, _ := get(t, rt, other); status != 200 || up.count() != calls+1 { | ||
| 99 | t.Errorf("after backoff: status %d, calls %d, want 200 and one more upstream call", status, up.count()) | ||
| 100 | } | ||
| 101 | } | ||
| 102 | |||
| 103 | func TestStaleEntryDroppedAfterTheWindow(t *testing.T) { | ||
| 104 | c, up, rt, now := staleCache(t) | ||
| 105 | *now = now.Add(time.Minute + time.Hour + time.Second) // past ttl and stale | ||
| 106 | up.set(403, "<html>blocked</html>", nil) | ||
| 107 | |||
| 108 | status, _ := get(t, rt, puppyURL) | ||
| 109 | |||
| 110 | if status != 403 { | ||
| 111 | t.Errorf("got %d, want the 403 passed through once nothing stale is held", status) | ||
| 112 | } | ||
| 113 | if _, _, _, entries, _ := c.stats(); entries != 0 { | ||
| 114 | t.Errorf("%d entries held, want 0", entries) | ||
| 115 | } | ||
| 116 | } | ||
| 117 | |||
| 118 | func TestZeroStaleKeepsNothingPastTTL(t *testing.T) { | ||
| 119 | up := &failingRT{} | ||
| 120 | up.set(200, "x", nil) | ||
| 121 | c := newAPICache(1<<20, time.Minute, 0) | ||
| 122 | now := time.Now() | ||
| 123 | c.now = func() time.Time { return now } | ||
| 124 | rt := c.transport(up) | ||
| 125 | get(t, rt, puppyURL) | ||
| 126 | now = now.Add(2 * time.Minute) | ||
| 127 | up.set(403, "blocked", nil) | ||
| 128 | |||
| 129 | if status, _ := get(t, rt, puppyURL); status != 403 { | ||
| 130 | t.Errorf("got %d with stale 0, want 403: nothing may be served past ttl", status) | ||
| 131 | } | ||
| 132 | } | ||
app/apicache_test.go +16 −18
| @@ -56,7 +56,7 @@ func get(t *testing.T, rt http.RoundTripper, url string) (int, string) { | |||
| 56 | 56 | ||
| 57 | func TestSecondRequestIsServedFromCache(t *testing.T) { | 57 | func TestSecondRequestIsServedFromCache(t *testing.T) { |
| 58 | up := &fakeRT{status: 200, body: `{"a":1}`} | 58 | up := &fakeRT{status: 200, body: `{"a":1}`} |
| 59 | rt := newAPICache(1<<20, time.Minute, 0).transport(up) | 59 | rt := newAPICache(1<<20, time.Minute).transport(up) |
| 60 | 60 | ||
| 61 | get(t, rt, puppyURL) | 61 | get(t, rt, puppyURL) |
| 62 | status, body := get(t, rt, puppyURL) | 62 | status, body := get(t, rt, puppyURL) |
| @@ -71,7 +71,7 @@ func TestSecondRequestIsServedFromCache(t *testing.T) { | |||
| 71 | 71 | ||
| 72 | func TestExpiredEntryIsRefetched(t *testing.T) { | 72 | func TestExpiredEntryIsRefetched(t *testing.T) { |
| 73 | up := &fakeRT{status: 200, body: `{}`} | 73 | up := &fakeRT{status: 200, body: `{}`} |
| 74 | c := newAPICache(1<<20, time.Minute, 0) | 74 | c := newAPICache(1<<20, time.Minute) |
| 75 | now := time.Now() | 75 | now := time.Now() |
| 76 | c.now = func() time.Time { return now } | 76 | c.now = func() time.Time { return now } |
| 77 | rt := c.transport(up) | 77 | rt := c.transport(up) |
| @@ -85,26 +85,24 @@ func TestExpiredEntryIsRefetched(t *testing.T) { | |||
| 85 | } | 85 | } |
| 86 | } | 86 | } |
| 87 | 87 | ||
| 88 | // A 500 here rather than a 403: a 403 is a block and starts the backoff, | ||
| 89 | // which is covered in apicache_stale_test.go. | ||
| 90 | func TestNon200IsNotStored(t *testing.T) { | 88 | func TestNon200IsNotStored(t *testing.T) { |
| 91 | up := &fakeRT{status: 500, body: "upstream broke"} | 89 | up := &fakeRT{status: 403, body: "blocked"} |
| 92 | rt := newAPICache(1<<20, time.Minute, 0).transport(up) | 90 | rt := newAPICache(1<<20, time.Minute).transport(up) |
| 93 | 91 | ||
| 94 | status, body := get(t, rt, puppyURL) | 92 | status, body := get(t, rt, puppyURL) |
| 95 | get(t, rt, puppyURL) | 93 | get(t, rt, puppyURL) |
| 96 | 94 | ||
| 97 | if status != 500 || body != "upstream broke" { | 95 | if status != 403 || body != "blocked" { |
| 98 | t.Errorf("first response is %d %q, want the upstream 500 passed through", status, body) | 96 | t.Errorf("first response is %d %q, want the upstream 403 passed through", status, body) |
| 99 | } | 97 | } |
| 100 | if up.count() != 2 { | 98 | if up.count() != 2 { |
| 101 | t.Errorf("upstream called %d times, want 2: a 500 must not be cached", up.count()) | 99 | t.Errorf("upstream called %d times, want 2: a 403 must not be cached", up.count()) |
| 102 | } | 100 | } |
| 103 | } | 101 | } |
| 104 | 102 | ||
| 105 | func TestBypassesSessionAndOtherHosts(t *testing.T) { | 103 | func TestBypassesSessionAndOtherHosts(t *testing.T) { |
| 106 | up := &fakeRT{status: 200, body: "x"} | 104 | up := &fakeRT{status: 200, body: "x"} |
| 107 | rt := newAPICache(1<<20, time.Minute, 0).transport(up) | 105 | rt := newAPICache(1<<20, time.Minute).transport(up) |
| 108 | 106 | ||
| 109 | for _, url := range []string{ | 107 | for _, url := range []string{ |
| 110 | "https://www.deviantart.com/_puppy", | 108 | "https://www.deviantart.com/_puppy", |
| @@ -121,7 +119,7 @@ func TestBypassesSessionAndOtherHosts(t *testing.T) { | |||
| 121 | 119 | ||
| 122 | func TestKeyIgnoresCSRFToken(t *testing.T) { | 120 | func TestKeyIgnoresCSRFToken(t *testing.T) { |
| 123 | up := &fakeRT{status: 200, body: "x"} | 121 | up := &fakeRT{status: 200, body: "x"} |
| 124 | rt := newAPICache(1<<20, time.Minute, 0).transport(up) | 122 | rt := newAPICache(1<<20, time.Minute).transport(up) |
| 125 | 123 | ||
| 126 | get(t, rt, puppyURL) | 124 | get(t, rt, puppyURL) |
| 127 | get(t, rt, strings.Replace(puppyURL, "csrf_token=abc", "csrf_token=def", 1)) | 125 | get(t, rt, strings.Replace(puppyURL, "csrf_token=abc", "csrf_token=def", 1)) |
| @@ -133,7 +131,7 @@ func TestKeyIgnoresCSRFToken(t *testing.T) { | |||
| 133 | 131 | ||
| 134 | func TestByteBoundEvictsLeastRecentlyUsed(t *testing.T) { | 132 | func TestByteBoundEvictsLeastRecentlyUsed(t *testing.T) { |
| 135 | up := &fakeRT{status: 200, body: strings.Repeat("x", 100)} | 133 | up := &fakeRT{status: 200, body: strings.Repeat("x", 100)} |
| 136 | rt := newAPICache(250, time.Minute, 0).transport(up) | 134 | rt := newAPICache(250, time.Minute).transport(up) |
| 137 | a := "https://www.deviantart.com/_puppy/a?p=1" | 135 | a := "https://www.deviantart.com/_puppy/a?p=1" |
| 138 | b := "https://www.deviantart.com/_puppy/b?p=1" | 136 | b := "https://www.deviantart.com/_puppy/b?p=1" |
| 139 | c := "https://www.deviantart.com/_puppy/c?p=1" | 137 | c := "https://www.deviantart.com/_puppy/c?p=1" |
| @@ -153,7 +151,7 @@ func TestByteBoundEvictsLeastRecentlyUsed(t *testing.T) { | |||
| 153 | 151 | ||
| 154 | func TestConcurrentMissesMakeOneUpstreamCall(t *testing.T) { | 152 | func TestConcurrentMissesMakeOneUpstreamCall(t *testing.T) { |
| 155 | up := &fakeRT{status: 200, body: "x", delay: 50 * time.Millisecond} | 153 | up := &fakeRT{status: 200, body: "x", delay: 50 * time.Millisecond} |
| 156 | rt := newAPICache(1<<20, time.Minute, 0).transport(up) | 154 | rt := newAPICache(1<<20, time.Minute).transport(up) |
| 157 | 155 | ||
| 158 | var wg sync.WaitGroup | 156 | var wg sync.WaitGroup |
| 159 | for range 20 { | 157 | for range 20 { |
| @@ -168,16 +166,16 @@ func TestConcurrentMissesMakeOneUpstreamCall(t *testing.T) { | |||
| 168 | 166 | ||
| 169 | func TestStatsCountHitsAndMisses(t *testing.T) { | 167 | func TestStatsCountHitsAndMisses(t *testing.T) { |
| 170 | up := &fakeRT{status: 200, body: "abc"} | 168 | up := &fakeRT{status: 200, body: "abc"} |
| 171 | c := newAPICache(1<<20, time.Minute, 0) | 169 | c := newAPICache(1<<20, time.Minute) |
| 172 | rt := c.transport(up) | 170 | rt := c.transport(up) |
| 173 | 171 | ||
| 174 | get(t, rt, puppyURL) | 172 | get(t, rt, puppyURL) |
| 175 | get(t, rt, puppyURL) | 173 | get(t, rt, puppyURL) |
| 176 | get(t, rt, puppyURL) | 174 | get(t, rt, puppyURL) |
| 177 | 175 | ||
| 178 | hits, misses, stale, entries, held := c.stats() | 176 | hits, misses, entries, held := c.stats() |
| 179 | if hits != 2 || misses != 1 || stale != 0 || entries != 1 || held != 3 { | 177 | if hits != 2 || misses != 1 || entries != 1 || held != 3 { |
| 180 | t.Errorf("stats = %d hits, %d misses, %d stale, %d entries, %d bytes; want 2, 1, 0, 1, 3", hits, misses, stale, entries, held) | 178 | t.Errorf("stats = %d hits, %d misses, %d entries, %d bytes; want 2, 1, 1, 3", hits, misses, entries, held) |
| 181 | } | 179 | } |
| 182 | } | 180 | } |
| 183 | 181 | ||
| @@ -187,7 +185,7 @@ func TestStatsCountHitsAndMisses(t *testing.T) { | |||
| 187 | func TestHitDoesNotConsumeAThrottleSlot(t *testing.T) { | 185 | func TestHitDoesNotConsumeAThrottleSlot(t *testing.T) { |
| 188 | up := &fakeRT{status: 200, body: "x"} | 186 | up := &fakeRT{status: 200, body: "x"} |
| 189 | th := &daThrottle{base: up, sem: make(chan struct{}, 1)} | 187 | th := &daThrottle{base: up, sem: make(chan struct{}, 1)} |
| 190 | rt := newAPICache(1<<20, time.Minute, 0).transport(th) | 188 | rt := newAPICache(1<<20, time.Minute).transport(th) |
| 191 | 189 | ||
| 192 | get(t, rt, puppyURL) // populate through the throttle | 190 | get(t, rt, puppyURL) // populate through the throttle |
| 193 | 191 | ||
app/cache.go +10 −3
| @@ -194,7 +194,14 @@ func buildMediaURL(subdomain, path, token string) (string, bool) { | |||
| 194 | // client, serving it from the on-disk or in-memory cache when enabled. It | 194 | // client, serving it from the on-disk or in-memory cache when enabled. It |
| 195 | // responds 403 when proxying is turned off for this instance. | 195 | // responds 403 when proxying is turned off for this instance. |
| 196 | func (s skunkyart) DownloadAndSendMedia(subdomain, path string) { | 196 | func (s skunkyart) DownloadAndSendMedia(subdomain, path string) { |
| 197 | mediaURL, ok := buildMediaURL(subdomain, path, s.Args.Get("token")) | 197 | s.downloadAndSendMedia(subdomain, path, s.Args.Get("token")) |
| 198 | } | ||
| 199 | |||
| 200 | // fetchMedia is Download behind a variable so tests can script the CDN. | ||
| 201 | var fetchMedia = Download | ||
| 202 | |||
| 203 | func (s skunkyart) downloadAndSendMedia(subdomain, path, token string) { | ||
| 204 | mediaURL, ok := buildMediaURL(subdomain, path, token) | ||
| 198 | if !ok { | 205 | if !ok { |
| 199 | s.ReturnHTTPError(400) | 206 | s.ReturnHTTPError(400) |
| 200 | return | 207 | return |
| @@ -225,7 +232,7 @@ func (s skunkyart) DownloadAndSendMedia(subdomain, path string) { | |||
| 225 | memPut(key, response) | 232 | memPut(key, response) |
| 226 | } | 233 | } |
| 227 | case CFG.Proxy: | 234 | case CFG.Proxy: |
| 228 | dwnld := Download(mediaURL) | 235 | dwnld := fetchMedia(mediaURL) |
| 229 | if dwnld.Status != 200 { | 236 | if dwnld.Status != 200 { |
| 230 | s.ReturnHTTPError(dwnld.Status) | 237 | s.ReturnHTTPError(dwnld.Status) |
| 231 | return | 238 | return |
| @@ -258,7 +265,7 @@ func (s skunkyart) loadOrFetchMedia(filePath, mediaURL string) ([]byte, bool) { | |||
| 258 | } | 265 | } |
| 259 | } | 266 | } |
| 260 | 267 | ||
| 261 | dwnld := Download(mediaURL) | 268 | dwnld := fetchMedia(mediaURL) |
| 262 | if dwnld.Status != 200 || !strings.HasPrefix(dwnld.Headers.Get("Content-Type"), "image") { | 269 | if dwnld.Status != 200 || !strings.HasPrefix(dwnld.Headers.Get("Content-Type"), "image") { |
| 263 | s.ReturnHTTPError(dwnld.Status) | 270 | s.ReturnHTTPError(dwnld.Status) |
| 264 | return nil, false | 271 | return nil, false |
app/config.go +2 −12
| @@ -32,7 +32,6 @@ type apiCacheConfig struct { | |||
| 32 | Enabled bool `json:"enabled"` | 32 | Enabled bool `json:"enabled"` |
| 33 | MaxSize int64 `json:"max-size"` | 33 | MaxSize int64 `json:"max-size"` |
| 34 | TTL string `json:"ttl"` | 34 | TTL string `json:"ttl"` |
| 35 | Stale string `json:"stale"` | ||
| 36 | } | 35 | } |
| 37 | 36 | ||
| 38 | type rateLimitConfig struct { | 37 | type rateLimitConfig struct { |
| @@ -76,7 +75,6 @@ var CFG = config{ | |||
| 76 | Enabled: true, | 75 | Enabled: true, |
| 77 | MaxSize: 64, | 76 | MaxSize: 64, |
| 78 | TTL: "5i", | 77 | TTL: "5i", |
| 79 | Stale: "1h", | ||
| 80 | }, | 78 | }, |
| 81 | RateLimit: rateLimitConfig{ | 79 | RateLimit: rateLimitConfig{ |
| 82 | PerMinute: 60, | 80 | PerMinute: 60, |
| @@ -90,9 +88,8 @@ var CFG = config{ | |||
| 90 | 88 | ||
| 91 | var lifetimeParsed int64 | 89 | var lifetimeParsed int64 |
| 92 | 90 | ||
| 93 | // apiCacheTTL and apiCacheStale are api-cache.ttl and api-cache.stale parsed, | 91 | // apiCacheTTL is api-cache.ttl parsed, set by ExecuteConfig. |
| 94 | // set by ExecuteConfig. | 92 | var apiCacheTTL time.Duration |
| 95 | var apiCacheTTL, apiCacheStale time.Duration | ||
| 96 | 93 | ||
| 97 | // parseLifetime reads a duration in the config's unit syntax: a number | 94 | // parseLifetime reads a duration in the config's unit syntax: a number |
| 98 | // followed by i (minutes), h (hours), d (days), w (weeks), m (30-day | 95 | // followed by i (minutes), h (hours), d (days), w (weeks), m (30-day |
| @@ -219,13 +216,6 @@ func ExecuteConfig() { | |||
| 219 | exit("config: api-cache.ttl: "+err.Error(), 1) | 216 | exit("config: api-cache.ttl: "+err.Error(), 1) |
| 220 | } | 217 | } |
| 221 | apiCacheTTL = d | 218 | apiCacheTTL = d |
| 222 | if CFG.APICache.Stale != "" { | ||
| 223 | d, err := parseLifetime(CFG.APICache.Stale) | ||
| 224 | if err != nil { | ||
| 225 | exit("config: api-cache.stale: "+err.Error(), 1) | ||
| 226 | } | ||
| 227 | apiCacheStale = d | ||
| 228 | } | ||
| 229 | } | 219 | } |
| 230 | 220 | ||
| 231 | // per-minute 0 turns the limit off; a burst below one token would | 221 | // per-minute 0 turns the limit off; a burst below one token would |
app/config_test.go +2 −2
| @@ -32,8 +32,8 @@ func TestParseLifetimeRejectsBadInput(t *testing.T) { | |||
| 32 | } | 32 | } |
| 33 | 33 | ||
| 34 | func TestAPICacheDefaults(t *testing.T) { | 34 | func TestAPICacheDefaults(t *testing.T) { |
| 35 | if !CFG.APICache.Enabled || CFG.APICache.MaxSize != 64 || CFG.APICache.TTL != "5i" || CFG.APICache.Stale != "1h" { | 35 | if !CFG.APICache.Enabled || CFG.APICache.MaxSize != 64 || CFG.APICache.TTL != "5i" { |
| 36 | t.Errorf("defaults are %+v, want enabled, 64 MB, 5i, stale 1h", CFG.APICache) | 36 | t.Errorf("defaults are %+v, want enabled, 64 MB, 5i", CFG.APICache) |
| 37 | } | 37 | } |
| 38 | } | 38 | } |
| 39 | 39 | ||
app/httpclient.go +3 −3
| @@ -100,8 +100,8 @@ func chain(base http.RoundTripper) http.RoundTripper { | |||
| 100 | func logCacheStatsForever(c *apiCache) { | 100 | func logCacheStatsForever(c *apiCache) { |
| 101 | for { | 101 | for { |
| 102 | time.Sleep(time.Hour) | 102 | time.Sleep(time.Hour) |
| 103 | hits, misses, stale, entries, held := c.stats() | 103 | hits, misses, entries, held := c.stats() |
| 104 | println("api cache:", hits, "hits,", misses, "misses,", stale, "served stale,", entries, "entries,", held>>20, "MB held") | 104 | println("api cache:", hits, "hits,", misses, "misses,", entries, "entries,", held>>20, "MB held") |
| 105 | } | 105 | } |
| 106 | } | 106 | } |
| 107 | 107 | ||
| @@ -111,7 +111,7 @@ func logCacheStatsForever(c *apiCache) { | |||
| 111 | func InstallDAThrottle() { | 111 | func InstallDAThrottle() { |
| 112 | baseTransport = tunedTransport() | 112 | baseTransport = tunedTransport() |
| 113 | if CFG.APICache.Enabled { | 113 | if CFG.APICache.Enabled { |
| 114 | daCache = newAPICache(CFG.APICache.MaxSize<<20, apiCacheTTL, apiCacheStale) | 114 | daCache = newAPICache(CFG.APICache.MaxSize<<20, apiCacheTTL) |
| 115 | go logCacheStatsForever(daCache) | 115 | go logCacheStatsForever(daCache) |
| 116 | } | 116 | } |
| 117 | http.DefaultTransport = chain(baseTransport) | 117 | http.DefaultTransport = chain(baseTransport) |
main.go +8 −3
| @@ -39,13 +39,18 @@ func main() { | |||
| 39 | // and let the request escape the throttle and the configured User-Agent. | 39 | // and let the request escape the throttle and the configured User-Agent. |
| 40 | go app.RefreshInstances() | 40 | go app.RefreshInstances() |
| 41 | 41 | ||
| 42 | // The first session bootstrap runs before the listener opens: requests | ||
| 43 | // that arrive before it finishes go to DeviantArt without a token and | ||
| 44 | // fail, which showed up as 502s for the first seconds after a restart. | ||
| 45 | if err := devianter.UpdateCSRF(); err != nil { | ||
| 46 | println(err.Error()) | ||
| 47 | } | ||
| 42 | go func() { | 48 | go func() { |
| 43 | for { | 49 | for { |
| 44 | err := devianter.UpdateCSRF() | 50 | time.Sleep(12 * time.Hour) |
| 45 | if err != nil { | 51 | if err := devianter.UpdateCSRF(); err != nil { |
| 46 | println(err.Error()) | 52 | println(err.Error()) |
| 47 | } | 53 | } |
| 48 | time.Sleep(12 * time.Hour) | ||
| 49 | } | 54 | } |
| 50 | }() | 55 | }() |
| 51 | 56 | ||