// which is one of "src", "dest", or "thirdleg"
func (c *syncCmd) storageFromParam(which storageType, val string) (blobserver.Storage, error) {
if val == "" {
switch which {
case storageThird:
return nil, nil
case storageSource:
discl := c.discoClient()
discl.SetLogger(c.logger)
src, err := discl.BlobRoot()
if err != nil {
return nil, fmt.Errorf("Failed to discover source server's blob path: %v", err)
}
val = src
}
if val == "" {
return nil, cmdmain.UsageError("No --" + string(which) + " flag value specified")
}
}
if which == storageDest && val == "stdout" {
return nil, nil
}
if looksLikePath(val) {
disk, err := localdisk.New(val)
if err != nil {
return nil, fmt.Errorf("Interpreted --%v=%q as a local disk path, but got error: %v", val, err)
}
return disk, nil
}
cl := client.New(val)
// TODO(mpl): probably needs the transport setup for trusted certs here.
cl.SetupAuth()
cl.SetLogger(c.logger)
return noHub{cl}, nil
}
// killReindex starts a reindexing in a new process, and kills that process
// after killTime. It then (naively for now ?) verifies that the kv store file is
// not corrupted by reinitializing an (possibly incomplete) index (with a corpus)
// with it. If the indexing was completed before we could kill the process, it
// returns true, false otherwise.
func killReindex(b *testing.B, dbfile string, killTime time.Duration,
sortedProvider func(dbfile string) (sorted.KeyValue, error)) bool {
cmd := exec.Command(os.Args[0], "-test.run=TestChildIndexer")
cmd.Env = append(cmd.Env, "TEST_BE_CHILD=1", "TEST_BE_CHILD_DBFILE="+dbfile)
var stdout, stderr bytes.Buffer
cmd.Stdout = &stdout
cmd.Stderr = &stderr
if err := cmd.Start(); err != nil {
b.Fatal(err)
}
waitc := make(chan error)
go func() {
waitc <- cmd.Wait()
}()
fullIndex := false
select {
case err := <-waitc:
if err == nil {
// indexer finished before we killed it
fullIndex = true
b.Logf("Finished indexing before being killed at %v", killTime)
break
}
// TODO(mpl): do better
if err.Error() != "signal: killed" {
b.Fatalf("unexpected (not killed) error from indexer process: %v %v %v", err, stdout.String(), stderr.String())
}
case <-time.After(killTime):
if err := cmd.Process.Kill(); err != nil {
b.Fatal(err)
}
err := <-waitc
// TODO(mpl): do better
if err != nil && err.Error() != "signal: killed" {
b.Fatalf("unexpected (not killed) error from indexer process: %v %v %v", err, stdout.String(), stderr.String())
}
}
kv, err := sortedProvider(dbfile)
if err != nil {
b.Fatal(err)
}
idx, err := index.New(kv)
if err != nil {
b.Fatal(err)
}
bs, err := localdisk.New(filepath.Join(filepath.Dir(dbfile), "bs"))
if err != nil {
b.Fatal(err)
}
idx.InitBlobSource(bs)
if _, err := idx.KeepInMemory(); err != nil {
b.Fatal(err)
}
if err := idx.Close(); err != nil {
b.Fatal(err)
}
return fullIndex
}
// TODO(rh): tame copy/paste code from cammount
func main() {
client.AddFlags()
flag.Parse()
cacheDir, err := ioutil.TempDir("", "camlicache")
if err != nil {
log.Fatalf("Error creating temp cache directory: %v", err)
}
defer os.RemoveAll(cacheDir)
diskcache, err := localdisk.New(cacheDir)
if err != nil {
log.Fatalf("Error setting up local disk cache: %v", err)
}
if flag.NArg() != 1 {
log.Fatal("usage: camwebdav <blobref>")
}
br := blobref.Parse(flag.Arg(0))
if br == nil {
log.Fatalf("%s was not a valid blobref.", flag.Arg(0))
}
client := client.NewOrFail()
fetcher := cacher.NewCachingFetcher(diskcache, client)
f = fs.NewCamliFileSystem(fetcher, br)
http.HandleFunc("/", webdav)
err = http.ListenAndServe(*davaddr, nil)
if err != nil {
log.Fatalf("Error starting WebDAV server: %v", err)
}
}
func reindex(b *testing.B, dbfile string,
sortedProvider func(dbfile string) (sorted.KeyValue, error)) *index.Index {
b.Logf("reindexing")
if err := os.RemoveAll(dbfile); err != nil {
b.Fatal(err)
}
kv, err := sortedProvider(dbfile)
if err != nil {
b.Fatal(err)
}
bs, err := localdisk.New(filepath.Join(filepath.Dir(dbfile), "bs"))
if err != nil {
b.Fatal(err)
}
idx, err := index.New(kv)
if err != nil {
b.Fatal(err)
}
idx.InitBlobSource(bs)
b.ResetTimer()
if err := idx.Reindex(); err != nil {
b.Fatal(err)
}
return idx
}
func enumerateMeta(b *testing.B, dbfile string,
sortedProvider func(dbfile string) (sorted.KeyValue, error)) int {
b.Logf("enumerating meta blobs")
kv, err := sortedProvider(dbfile)
if err != nil {
b.Fatal(err)
}
bs, err := localdisk.New(filepath.Join(filepath.Dir(dbfile), "bs"))
if err != nil {
b.Fatal(err)
}
idx, err := index.New(kv)
if err != nil {
b.Fatal(err)
}
idx.InitBlobSource(bs)
defer idx.Close()
ch := make(chan camtypes.BlobMeta, 100)
go func() {
if err := idx.EnumerateBlobMeta(nil, ch); err != nil {
b.Fatal(err)
}
}()
n := 0
for range ch {
n++
}
b.Logf("Enumerated %d meta blobs", n)
return n
}
func main() {
// Scans the arg list and sets up flags
debug := flag.Bool("debug", false, "print debugging messages.")
client.AddFlags()
flag.Parse()
errorf := func(msg string, args ...interface{}) {
fmt.Fprintf(os.Stderr, msg, args...)
os.Exit(2)
}
if n := flag.NArg(); n < 1 || n > 2 {
errorf("usage: cammount <mountpoint> [<root-blobref>]\n")
}
mountPoint := flag.Arg(0)
client := client.NewOrFail() // automatic from flags
cacheDir, err := ioutil.TempDir("", "camlicache")
if err != nil {
errorf("Error creating temp cache directory: %v\n", err)
}
defer os.RemoveAll(cacheDir)
diskcache, err := localdisk.New(cacheDir)
if err != nil {
errorf("Error setting up local disk cache: %v", err)
}
fetcher := cacher.NewCachingFetcher(diskcache, client)
var camfs *fs.CamliFileSystem
if flag.NArg() == 2 {
root := blobref.Parse(flag.Arg(1))
if root == nil {
errorf("Error parsing root blobref: %q\n", root)
}
var err error
camfs, err = fs.NewRootedCamliFileSystem(fetcher, root)
if err != nil {
errorf("Error creating root with %v: %v", root, err)
}
} else {
camfs = fs.NewCamliFileSystem(fetcher)
log.Printf("starting with fs %#v", camfs)
}
if *debug {
// TODO: set fs's logger
}
conn, err := fuse.Mount(mountPoint)
if err != nil {
log.Fatalf("Mount: %v", err)
}
err = conn.Serve(camfs)
if err != nil {
log.Fatalf("Serve: %v", err)
}
log.Printf("fuse process ending.")
}
// which is one of "src", "dest", or "thirdleg"
func (c *syncCmd) storageFromParam(which storageType, val string) (blobserver.Storage, error) {
var httpClient *http.Client
if val == "" {
switch which {
case storageThird:
return nil, nil
case storageSource:
discl := c.discoClient()
discl.SetLogger(c.logger)
src, err := discl.BlobRoot()
if err != nil {
return nil, fmt.Errorf("Failed to discover source server's blob path: %v", err)
}
val = src
httpClient = discl.HTTPClient()
}
if val == "" {
return nil, cmdmain.UsageError("No --" + string(which) + " flag value specified")
}
}
if which == storageDest && val == "stdout" {
return nil, nil
}
if looksLikePath(val) {
disk, err := localdisk.New(val)
if err != nil {
return nil, fmt.Errorf("Interpreted --%v=%q as a local disk path, but got error: %v", which, val, err)
}
c.oneIsDisk = true
return disk, nil
}
cl := client.New(val)
cl.InsecureTLS = c.insecureTLS
if httpClient == nil {
httpClient = &http.Client{
Transport: cl.TransportForConfig(nil),
}
}
cl.SetHTTPClient(httpClient)
if err := cl.SetupAuth(); err != nil {
return nil, fmt.Errorf("could not setup auth for connecting to %v: %v", val, err)
}
cl.SetLogger(c.logger)
serverKeyID, err := cl.ServerKeyID()
if err != nil && err != client.ErrNoSigning {
fmt.Fprintf(cmdmain.Stderr, "Failed to discover keyId for server %v: %v", val, err)
} else {
if which == storageSource {
c.srcKeyID = serverKeyID
} else if which == storageDest {
c.destKeyID = serverKeyID
}
}
return cl, nil
}
// New returns a new blobserver Storage implementation, storing blobs in the provided dir.
// If dir has an index.kv file, a diskpacked implementation is returnd.
func New(dir string) (blobserver.Storage, error) {
if v, err := diskpacked.IsDir(dir); err != nil {
return nil, err
} else if v {
return diskpacked.New(dir)
}
if v, err := localdisk.IsDir(dir); err != nil {
return nil, err
} else if v {
return localdisk.New(dir)
}
return diskpacked.New(dir)
}
// NewDiskCache returns a new DiskCache from a StreamingFetcher, which
// is usually the pkg/client HTTP client (which typically has much
// higher latency and lower bandwidth than local disk).
func NewDiskCache(fetcher blob.StreamingFetcher) (*DiskCache, error) {
// TODO: max disk size, keep LRU of access, smarter cleaning,
// persistent directory per-user, etc.
cacheDir, err := ioutil.TempDir("", "camlicache")
if err != nil {
return nil, err
}
diskcache, err := localdisk.New(cacheDir)
if err != nil {
return nil, err
}
dc := &DiskCache{
CachingFetcher: NewCachingFetcher(diskcache, fetcher),
Root: cacheDir,
}
return dc, nil
}
func TestChildIndexer(t *testing.T) {
if os.Getenv("TEST_BE_CHILD") != "1" {
t.Skip("not a real test; used as a child process by the benchmarks")
}
dbfile := os.Getenv("TEST_BE_CHILD_DBFILE")
if dbfile == "" {
log.Fatal("empty TEST_BE_CHILD_DBFILE")
}
if err := os.RemoveAll(dbfile); err != nil {
log.Fatal(err)
}
var kv sorted.KeyValue
var err error
switch {
case strings.HasSuffix(dbfile, "leveldb.db"):
kv, err = leveldb.NewStorage(dbfile)
case strings.HasSuffix(dbfile, "kvfile.db"):
kv, err = kvfile.NewStorage(dbfile)
case strings.HasSuffix(dbfile, "sqlite.db"):
kv, err = sqlite.NewStorage(dbfile)
default:
log.Fatalf("unknown sorted provider for %v", dbfile)
}
if err != nil {
log.Fatal(err)
}
bs, err := localdisk.New(filepath.Join(filepath.Dir(dbfile), "bs"))
if err != nil {
log.Fatal(err)
}
idx, err := index.New(kv)
if err != nil {
log.Fatal(err)
}
idx.InitBlobSource(bs)
defer func() {
if err := idx.Close(); err != nil {
log.Fatal(err)
}
}()
if err := idx.Reindex(); err != nil {
log.Fatal(err)
}
}
// which is one of "src", "dest", or "thirdleg"
func (c *syncCmd) storageFromParam(which storageType, val string) (blobserver.Storage, error) {
var httpClient *http.Client
if val == "" {
switch which {
case storageThird:
return nil, nil
case storageSource:
discl := c.discoClient()
discl.SetLogger(c.logger)
src, err := discl.BlobRoot()
if err != nil {
return nil, fmt.Errorf("Failed to discover source server's blob path: %v", err)
}
val = src
httpClient = discl.HTTPClient()
}
if val == "" {
return nil, cmdmain.UsageError("No --" + string(which) + " flag value specified")
}
}
if which == storageDest && val == "stdout" {
return nil, nil
}
if looksLikePath(val) {
disk, err := localdisk.New(val)
if err != nil {
return nil, fmt.Errorf("Interpreted --%v=%q as a local disk path, but got error: %v", which, val, err)
}
return disk, nil
}
cl := client.New(val)
cl.InsecureTLS = c.insecureTLS
if httpClient == nil {
httpClient = &http.Client{
Transport: cl.TransportForConfig(nil),
}
}
cl.SetHTTPClient(httpClient)
cl.SetupAuth()
cl.SetLogger(c.logger)
return cl, nil
}
// dataStores returns the blobserver that stores the instances configurations, and the kv
// store for the instances states.
func dataStores() (blobserver.Storage, sorted.KeyValue, error) {
dataDir := os.Getenv("CAMLI_GCE_DATA")
if dataDir == "" {
dataDir = "camli-gce-data"
log.Printf("data dir not provided as env var CAMLI_GCE_DATA, so defaulting to %v", dataDir)
}
blobsDir := filepath.Join(dataDir, "instance-conf")
if err := os.MkdirAll(blobsDir, 0700); err != nil {
return nil, nil, err
}
instConf, err := localdisk.New(blobsDir)
if err != nil {
return nil, nil, err
}
instState, err := leveldb.NewStorage(filepath.Join(dataDir, "instance-state"))
if err != nil {
return nil, nil, err
}
return instConf, instState, nil
}
// NewDiskCache returns a new DiskCache from a Fetcher, which
// is usually the pkg/client HTTP client (which typically has much
// higher latency and lower bandwidth than local disk).
func NewDiskCache(fetcher blob.Fetcher) (*DiskCache, error) {
cacheDir := filepath.Join(osutil.CacheDir(), "blobs")
if !osutil.DirExists(cacheDir) {
if err := os.Mkdir(cacheDir, 0700); err != nil {
log.Printf("Warning: failed to make %s: %v; using tempdir instead", cacheDir, err)
cacheDir, err = ioutil.TempDir("", "camlicache")
if err != nil {
return nil, err
}
}
}
// TODO: max disk size, keep LRU of access, smarter cleaning, etc
// TODO: use diskpacked instead? harder to clean, though.
diskcache, err := localdisk.New(cacheDir)
if err != nil {
return nil, err
}
dc := &DiskCache{
CachingFetcher: NewCachingFetcher(diskcache, fetcher),
Root: cacheDir,
}
return dc, nil
}
// Populates the bs, and the index at the same time through the sync handler
func populate(b *testing.B, dbfile string,
sortedProvider func(dbfile string) (sorted.KeyValue, error)) *index.Index {
b.Logf("populating %v", dbfile)
kv, err := sortedProvider(dbfile)
if err != nil {
b.Fatal(err)
}
bsRoot := filepath.Join(filepath.Dir(dbfile), "bs")
if err := os.MkdirAll(bsRoot, 0700); err != nil {
b.Fatal(err)
}
dataDir, err := os.Open("testdata")
if err != nil {
b.Fatal(err)
}
fis, err := dataDir.Readdir(-1)
if err != nil {
b.Fatal(err)
}
if len(fis) == 0 {
b.Fatalf("no files in %s dir", "testdata")
}
ks := doKeyStuff(b)
bs, err := localdisk.New(bsRoot)
if err != nil {
b.Fatal(err)
}
if _, err := blobserver.Receive(bs, ks.pubKeyRef, strings.NewReader(ks.pubKey)); err != nil {
b.Fatal(err)
}
idx, err := index.New(kv)
if err != nil {
b.Fatal(err)
}
idx.InitBlobSource(bs)
sh := server.NewSyncHandler("/bs/", "/index/", bs, idx, sorted.NewMemoryKeyValue())
b.ResetTimer()
for _, v := range fis {
f, err := os.Open(filepath.Join(dataDir.Name(), v.Name()))
if err != nil {
b.Fatal(err)
}
td := &trackDigestReader{r: f}
fm := schema.NewFileMap(v.Name())
fm.SetModTime(v.ModTime())
fileRef, err := schema.WriteFileMap(bs, fm, td)
if err != nil {
b.Fatal(err)
}
f.Close()
unsigned := schema.NewPlannedPermanode(td.Sum())
unsigned.SetSigner(ks.pubKeyRef)
sr := &jsonsign.SignRequest{
UnsignedJSON: unsigned.Blob().JSON(),
// TODO(mpl): if we make a bs that discards, replace this with a memory bs that has only the pubkey
Fetcher: bs,
EntityFetcher: ks.entityFetcher,
SignatureTime: time.Unix(0, 0),
}
signed, err := sr.Sign()
if err != nil {
b.Fatal("problem signing: " + err.Error())
}
pn := blob.SHA1FromString(signed)
// N.B: use blobserver.Receive so that the blob hub gets notified, and the blob gets enqueued into the index
if _, err := blobserver.Receive(bs, pn, strings.NewReader(signed)); err != nil {
b.Fatal(err)
}
contentAttr := schema.NewSetAttributeClaim(pn, "camliContent", fileRef.String())
claimTime, ok := fm.ModTime()
if !ok {
b.Fatal(err)
}
contentAttr.SetClaimDate(claimTime)
contentAttr.SetSigner(ks.pubKeyRef)
sr = &jsonsign.SignRequest{
UnsignedJSON: contentAttr.Blob().JSON(),
// TODO(mpl): if we make a bs that discards, replace this with a memory bs that has only the pubkey
Fetcher: bs,
EntityFetcher: ks.entityFetcher,
SignatureTime: claimTime,
}
signed, err = sr.Sign()
if err != nil {
b.Fatal("problem signing: " + err.Error())
}
cl := blob.SHA1FromString(signed)
if _, err := blobserver.Receive(bs, cl, strings.NewReader(signed)); err != nil {
b.Fatal(err)
}
}
sh.IdleWait()
return idx
}
func newPublishHandler(conf *config) *publishHandler {
cl, err := app.Client()
if err != nil {
logger.Fatalf("could not get a client for the publish handler %v", err)
}
if conf.RootName == "" {
logger.Fatal("camliRoot not found in the app configuration")
}
maxResizeBytes := conf.MaxResizeBytes
if maxResizeBytes == 0 {
maxResizeBytes = constants.DefaultMaxResizeMem
}
var CSSFiles, JSDeps []string
if conf.SourceRoot != "" {
appRoot := filepath.Join(conf.SourceRoot, "app", "publisher")
Files = &fileembed.Files{
DirFallback: appRoot,
}
// TODO(mpl): Can I readdir by listing with "/" on Files, even with DirFallBack?
// Apparently not, but retry later.
dir, err := os.Open(appRoot)
if err != nil {
logger.Fatal(err)
}
defer dir.Close()
names, err := dir.Readdirnames(-1)
if err != nil {
logger.Fatal(err)
}
for _, v := range names {
if strings.HasSuffix(v, ".css") {
CSSFiles = append(CSSFiles, v)
continue
}
// TODO(mpl): document or fix (use a map?) the ordering
// problem: i.e. jquery.js must be sourced before
// publisher.js. For now, just cheat by sorting the
// slice.
if strings.HasSuffix(v, ".js") {
JSDeps = append(JSDeps, v)
}
}
sort.Strings(JSDeps)
} else {
Files.Listable = true
dir, err := Files.Open("/")
if err != nil {
logger.Fatal(err)
}
defer dir.Close()
fis, err := dir.Readdir(-1)
if err != nil {
logger.Fatal(err)
}
for _, v := range fis {
name := v.Name()
if strings.HasSuffix(name, ".css") {
CSSFiles = append(CSSFiles, name)
continue
}
if strings.HasSuffix(name, ".js") {
JSDeps = append(JSDeps, name)
}
}
sort.Strings(JSDeps)
}
// TODO(mpl): add all htmls found in Files to the template if none specified?
if conf.GoTemplate == "" {
logger.Fatal("a go template is required in the app configuration")
}
goTemplate, err := goTemplate(Files, conf.GoTemplate)
if err != nil {
logger.Fatal(err)
}
var cache blobserver.Storage
var thumbMeta *server.ThumbMeta
if conf.CacheRoot != "" {
cache, err = localdisk.New(conf.CacheRoot)
if err != nil {
logger.Fatalf("Could not create localdisk cache: %v", err)
}
thumbsCacheDir := filepath.Join(os.TempDir(), "camli-publisher-cache")
if err := os.MkdirAll(thumbsCacheDir, 0700); err != nil {
logger.Fatalf("Could not create cache dir %s for %v publisher: %v", thumbsCacheDir, conf.RootName, err)
}
kv, err := sorted.NewKeyValue(map[string]interface{}{
"type": "kv",
"file": filepath.Join(thumbsCacheDir, conf.RootName+"-thumbnails.kv"),
})
if err != nil {
logger.Fatalf("Could not create kv for %v's thumbs cache: %v", conf.RootName, err)
}
thumbMeta = server.NewThumbMeta(kv)
}
return &publishHandler{
rootName: conf.RootName,
cl: cl,
resizeSem: syncutil.NewSem(maxResizeBytes),
staticFiles: Files,
goTemplate: goTemplate,
CSSFiles: CSSFiles,
JSDeps: JSDeps,
describedCache: make(map[string]*search.DescribedBlob),
cache: cache,
thumbMeta: thumbMeta,
}
}
func main() {
client.AddFlags()
flag.Parse()
if *flagGraph && flag.NArg() != 1 {
log.Fatalf("The --graph option requires exactly one parameter.")
}
var cl *client.Client
var items []*blobref.BlobRef
if *flagShared != "" {
if client.ExplicitServer() != "" {
log.Fatal("Can't use --shared with an explicit blobserver; blobserver is implicit from the --shared URL.")
}
if flag.NArg() != 0 {
log.Fatal("No arguments permitted when using --shared")
}
cl1, target, err := client.NewFromShareRoot(*flagShared)
if err != nil {
log.Fatal(err)
}
cl = cl1
items = append(items, target)
} else {
cl = client.NewOrFail()
for n := 0; n < flag.NArg(); n++ {
arg := flag.Arg(n)
br := blobref.Parse(arg)
if br == nil {
log.Fatalf("Failed to parse argument %q as a blobref.", arg)
}
items = append(items, br)
}
}
httpStats := &httputil.StatsTransport{
VerboseLog: *flagHTTP,
}
if *flagHTTP {
httpStats.Transport = &http.Transport{
Dial: func(net_, addr string) (net.Conn, error) {
log.Printf("Dialing %s", addr)
return net.Dial(net_, addr)
},
}
}
cl.SetHTTPClient(&http.Client{Transport: httpStats})
// Put a local disk cache in front of the HTTP client.
// TODO: this could be better about proactively cleaning things.
// Fetching 2 TB shouldn't write 2 TB to /tmp before it's done.
// Maybe the cache needs an LRU/size cap.
cacheDir, err := ioutil.TempDir("", "camlicache")
if err != nil {
log.Fatalf("Error creating temp cache directory: %v\n", err)
}
defer os.RemoveAll(cacheDir)
diskcache, err := localdisk.New(cacheDir)
if err != nil {
log.Fatalf("Error setting up local disk cache: %v", err)
}
if *flagVerbose {
log.Printf("Using temp blob cache directory %s", cacheDir)
}
fetcher := cacher.NewCachingFetcher(diskcache, cl)
for _, br := range items {
if *flagGraph {
printGraph(fetcher, br)
return
}
if *flagCheck {
// TODO: do HEAD requests checking if the blobs exists.
log.Fatal("not implemented")
return
}
if *flagOutput == "-" {
var rc io.ReadCloser
var err error
if *flagContents {
seekFetcher := blobref.SeekerFromStreamingFetcher(fetcher)
rc, err = schema.NewFileReader(seekFetcher, br)
if err == nil {
rc.(*schema.FileReader).LoadAllChunks()
}
} else {
rc, err = fetch(fetcher, br)
}
if err != nil {
log.Fatal(err)
}
defer rc.Close()
if _, err := io.Copy(os.Stdout, rc); err != nil {
log.Fatalf("Failed reading %q: %v", br, err)
}
} else {
if err := smartFetch(fetcher, *flagOutput, br); err != nil {
log.Fatal(err)
}
}
}
if *flagVerbose {
log.Printf("HTTP requests: %d\n", httpStats.Requests())
}
}
