lib/crunchrun/copier_test.go

   1 // Copyright (C) The Arvados Authors. All rights reserved.
   2 //
   3 // SPDX-License-Identifier: AGPL-3.0
   4
   5 package crunchrun
   6
   7 import (
   8         "bytes"
   9         "encoding/json"
  10         "fmt"
  11         "io"
  12         "io/fs"
  13         "os"
  14         "path"
  15         "runtime"
  16         "sort"
  17         "syscall"
  18
  19         "git.arvados.org/arvados.git/sdk/go/arvados"
  20         "git.arvados.org/arvados.git/sdk/go/arvadosclient"
  21         "git.arvados.org/arvados.git/sdk/go/arvadostest"
  22         "git.arvados.org/arvados.git/sdk/go/keepclient"
  23         "github.com/sirupsen/logrus"
  24         check "gopkg.in/check.v1"
  25 )
  26
  27 var _ = check.Suite(&copierSuite{})
  28
  29 type copierSuite struct {
  30         cp  copier
  31         log bytes.Buffer
  32 }
  33
  34 func (s *copierSuite) SetUpTest(c *check.C) {
  35         tmpdir := c.MkDir()
  36         s.log = bytes.Buffer{}
  37
  38         cl, err := arvadosclient.MakeArvadosClient()
  39         c.Assert(err, check.IsNil)
  40         kc, err := keepclient.MakeKeepClient(cl)
  41         c.Assert(err, check.IsNil)
  42         collfs, err := (&arvados.Collection{}).FileSystem(arvados.NewClientFromEnv(), kc)
  43         c.Assert(err, check.IsNil)
  44
  45         s.cp = copier{
  46                 client:        arvados.NewClientFromEnv(),
  47                 keepClient:    kc,
  48                 hostOutputDir: tmpdir,
  49                 ctrOutputDir:  "/ctr/outdir",
  50                 mounts: map[string]arvados.Mount{
  51                         "/ctr/outdir": {Kind: "tmp"},
  52                 },
  53                 secretMounts: map[string]arvados.Mount{
  54                         "/secret_text": {Kind: "text", Content: "xyzzy"},
  55                 },
  56                 logger: &logrus.Logger{Out: &s.log, Formatter: &logrus.TextFormatter{}, Level: logrus.InfoLevel},
  57                 staged: collfs,
  58         }
  59 }
  60
  61 func (s *copierSuite) TestEmptyOutput(c *check.C) {
  62         err := s.cp.walkMount("", s.cp.ctrOutputDir, 10, true)
  63         c.Check(err, check.IsNil)
  64         c.Check(s.cp.dirs, check.DeepEquals, []string(nil))
  65         c.Check(len(s.cp.files), check.Equals, 0)
  66 }
  67
  68 func (s *copierSuite) TestEmptyWritableMount(c *check.C) {
  69         s.writeFileInOutputDir(c, ".arvados#collection", `{"manifest_text":""}`)
  70         s.cp.mounts[s.cp.ctrOutputDir] = arvados.Mount{
  71                 Kind:     "collection",
  72                 Writable: true,
  73         }
  74
  75         err := s.cp.walkMount("", s.cp.ctrOutputDir, 10, true)
  76         c.Assert(err, check.IsNil)
  77         c.Check(s.cp.dirs, check.DeepEquals, []string(nil))
  78         c.Check(len(s.cp.files), check.Equals, 0)
  79         rootdir, err := s.cp.staged.Open(".")
  80         c.Assert(err, check.IsNil)
  81         defer rootdir.Close()
  82         fis, err := rootdir.Readdir(-1)
  83         c.Assert(err, check.IsNil)
  84         c.Check(fis, check.HasLen, 0)
  85 }
  86
  87 func (s *copierSuite) TestOutputCollectionWithOnlySubmounts(c *check.C) {
  88         s.writeFileInOutputDir(c, ".arvados#collection", `{"manifest_text":""}`)
  89         s.cp.mounts[s.cp.ctrOutputDir] = arvados.Mount{
  90                 Kind:     "collection",
  91                 Writable: true,
  92         }
  93         s.cp.mounts[path.Join(s.cp.ctrOutputDir, "foo")] = arvados.Mount{
  94                 Kind:             "collection",
  95                 Path:             "foo",
  96                 PortableDataHash: arvadostest.FooCollectionPDH,
  97         }
  98
  99         err := s.cp.walkMount("", s.cp.ctrOutputDir, 10, true)
 100         c.Assert(err, check.IsNil)
 101
 102         // s.cp.dirs and s.cp.files are empty, because nothing needs
 103         // to be copied from disk.
 104         c.Check(s.cp.dirs, check.DeepEquals, []string(nil))
 105         c.Check(len(s.cp.files), check.Equals, 0)
 106
 107         // The "foo" file has already been copied from FooCollection
 108         // to s.cp.staged via Snapshot+Splice.
 109         rootdir, err := s.cp.staged.Open(".")
 110         c.Assert(err, check.IsNil)
 111         defer rootdir.Close()
 112         fis, err := rootdir.Readdir(-1)
 113         c.Assert(err, check.IsNil)
 114         c.Assert(fis, check.HasLen, 1)
 115         c.Check(fis[0].Size(), check.Equals, int64(3))
 116 }
 117
 118 func (s *copierSuite) TestRepetitiveMountsInOutputDir(c *check.C) {
 119         var memstats0 runtime.MemStats
 120         runtime.ReadMemStats(&memstats0)
 121
 122         s.writeFileInOutputDir(c, ".arvados#collection", `{"manifest_text":""}`)
 123         s.cp.mounts[s.cp.ctrOutputDir] = arvados.Mount{
 124                 Kind:     "collection",
 125                 Writable: true,
 126         }
 127         nmounts := 200
 128         ncollections := 1
 129         pdh := make([]string, ncollections)
 130         s.cp.manifestCache = make(map[string]string)
 131         for i := 0; i < ncollections; i++ {
 132                 mtxt := arvadostest.FakeManifest(1, nmounts, 2, 4<<20)
 133                 pdh[i] = arvados.PortableDataHash(mtxt)
 134                 s.cp.manifestCache[pdh[i]] = mtxt
 135         }
 136         for i := 0; i < nmounts; i++ {
 137                 filename := fmt.Sprintf("file%d", i)
 138                 s.cp.mounts[path.Join(s.cp.ctrOutputDir, filename)] = arvados.Mount{
 139                         Kind:             "collection",
 140                         Path:             fmt.Sprintf("dir0/dir%d/file%d", i, i),
 141                         PortableDataHash: pdh[i%ncollections],
 142                 }
 143         }
 144         err := s.cp.walkMount("", s.cp.ctrOutputDir, 10, true)
 145         c.Assert(err, check.IsNil)
 146
 147         // Files mounted under output dir have been copied from the
 148         // fake collections to s.cp.staged via Snapshot+Splice.
 149         rootdir, err := s.cp.staged.Open(".")
 150         c.Assert(err, check.IsNil)
 151         defer rootdir.Close()
 152         fis, err := rootdir.Readdir(-1)
 153         c.Assert(err, check.IsNil)
 154         c.Assert(fis, check.HasLen, nmounts)
 155
 156         var memstats runtime.MemStats
 157         runtime.ReadMemStats(&memstats)
 158         c.Logf("%s Alloc=%d Sys=%d", time.Now(), memstats.Alloc, memstats.Sys)
 159 }
 160
 161 func (s *copierSuite) TestRegularFilesAndDirs(c *check.C) {
 162         err := os.MkdirAll(s.cp.hostOutputDir+"/dir1/dir2/dir3", 0755)
 163         c.Assert(err, check.IsNil)
 164         f, err := os.OpenFile(s.cp.hostOutputDir+"/dir1/foo", os.O_CREATE|os.O_WRONLY, 0644)
 165         c.Assert(err, check.IsNil)
 166         _, err = io.WriteString(f, "foo")
 167         c.Assert(err, check.IsNil)
 168         c.Assert(f.Close(), check.IsNil)
 169         err = syscall.Mkfifo(s.cp.hostOutputDir+"/dir1/fifo", 0644)
 170         c.Assert(err, check.IsNil)
 171
 172         err = s.cp.walkMount("", s.cp.ctrOutputDir, 10, true)
 173         c.Check(err, check.IsNil)
 174         c.Check(s.cp.dirs, check.DeepEquals, []string{"/dir1", "/dir1/dir2", "/dir1/dir2/dir3"})
 175         c.Check(s.cp.files, check.DeepEquals, []filetodo{
 176                 {src: os.DevNull, dst: "/dir1/dir2/dir3/.keep"},
 177                 {src: s.cp.hostOutputDir + "/dir1/foo", dst: "/dir1/foo", size: 3},
 178         })
 179         c.Check(s.log.String(), check.Matches, `.* msg="Skipping unsupported file type \(mode 200000644\) in output dir: \\"/ctr/outdir/dir1/fifo\\""\n`)
 180 }
 181
 182 func (s *copierSuite) TestSymlinkCycle(c *check.C) {
 183         c.Assert(os.Mkdir(s.cp.hostOutputDir+"/dir1", 0755), check.IsNil)
 184         c.Assert(os.Mkdir(s.cp.hostOutputDir+"/dir2", 0755), check.IsNil)
 185         c.Assert(os.Symlink("../dir2", s.cp.hostOutputDir+"/dir1/l_dir2"), check.IsNil)
 186         c.Assert(os.Symlink("../dir1", s.cp.hostOutputDir+"/dir2/l_dir1"), check.IsNil)
 187         err := s.cp.walkMount("", s.cp.ctrOutputDir, 10, true)
 188         c.Check(err, check.ErrorMatches, `.*cycle.*`)
 189 }
 190
 191 func (s *copierSuite) TestSymlinkTargetMissing(c *check.C) {
 192         c.Assert(os.Symlink("./missing", s.cp.hostOutputDir+"/symlink"), check.IsNil)
 193         err := s.cp.walkMount("", s.cp.ctrOutputDir, 10, true)
 194         c.Check(err, check.ErrorMatches, `.*/ctr/outdir/missing.*`)
 195 }
 196
 197 func (s *copierSuite) TestSymlinkTargetNotMounted(c *check.C) {
 198         c.Assert(os.Symlink("../boop", s.cp.hostOutputDir+"/symlink"), check.IsNil)
 199         err := s.cp.walkMount("", s.cp.ctrOutputDir, 10, true)
 200         c.Check(err, check.ErrorMatches, `.*/ctr/boop.*`)
 201 }
 202
 203 func (s *copierSuite) TestSymlinkToSecret(c *check.C) {
 204         c.Assert(os.Symlink("/secret_text", s.cp.hostOutputDir+"/symlink"), check.IsNil)
 205         err := s.cp.walkMount("", s.cp.ctrOutputDir, 10, true)
 206         c.Check(err, check.IsNil)
 207         c.Check(len(s.cp.dirs), check.Equals, 0)
 208         c.Check(len(s.cp.files), check.Equals, 0)
 209 }
 210
 211 func (s *copierSuite) TestSecretInOutputDir(c *check.C) {
 212         s.cp.secretMounts["/ctr/outdir/secret_text"] = s.cp.secretMounts["/secret_text"]
 213         s.writeFileInOutputDir(c, "secret_text", "xyzzy")
 214         err := s.cp.walkMount("", s.cp.ctrOutputDir, 10, true)
 215         c.Check(err, check.IsNil)
 216         c.Check(len(s.cp.dirs), check.Equals, 0)
 217         c.Check(len(s.cp.files), check.Equals, 0)
 218 }
 219
 220 func (s *copierSuite) TestSymlinkToMountedCollection(c *check.C) {
 221         // simulate mounted read-only collection
 222         s.cp.mounts["/mnt"] = arvados.Mount{
 223                 Kind:             "collection",
 224                 PortableDataHash: arvadostest.FooCollectionPDH,
 225         }
 226
 227         // simulate mounted writable collection
 228         bindtmp := c.MkDir()
 229         f, err := os.OpenFile(bindtmp+"/.arvados#collection", os.O_CREATE|os.O_WRONLY, 0644)
 230         c.Assert(err, check.IsNil)
 231         _, err = io.WriteString(f, `{"manifest_text":". 37b51d194a7513e45b56f6524f2d51f2+3 0:3:bar\n"}`)
 232         c.Assert(err, check.IsNil)
 233         c.Assert(f.Close(), check.IsNil)
 234         s.cp.mounts["/mnt-w"] = arvados.Mount{
 235                 Kind:             "collection",
 236                 PortableDataHash: arvadostest.FooCollectionPDH,
 237                 Writable:         true,
 238         }
 239         s.cp.bindmounts = map[string]bindmount{
 240                 "/mnt-w": bindmount{HostPath: bindtmp, ReadOnly: false},
 241         }
 242
 243         c.Assert(os.Symlink("../../mnt", s.cp.hostOutputDir+"/l_dir"), check.IsNil)
 244         c.Assert(os.Symlink("/mnt/foo", s.cp.hostOutputDir+"/l_file"), check.IsNil)
 245         c.Assert(os.Symlink("/mnt-w/bar", s.cp.hostOutputDir+"/l_file_w"), check.IsNil)
 246
 247         err = s.cp.walkMount("", s.cp.ctrOutputDir, 10, true)
 248         c.Check(err, check.IsNil)
 249         s.checkStagedFile(c, "l_dir/foo", 3)
 250         s.checkStagedFile(c, "l_file", 3)
 251         s.checkStagedFile(c, "l_file_w", 3)
 252 }
 253
 254 func (s *copierSuite) checkStagedFile(c *check.C, path string, size int64) {
 255         fi, err := s.cp.staged.Stat(path)
 256         if c.Check(err, check.IsNil) {
 257                 c.Check(fi.Size(), check.Equals, size)
 258         }
 259 }
 260
 261 func (s *copierSuite) TestSymlink(c *check.C) {
 262         hostfile := s.cp.hostOutputDir + "/dir1/file"
 263
 264         err := os.MkdirAll(s.cp.hostOutputDir+"/dir1/dir2/dir3", 0755)
 265         c.Assert(err, check.IsNil)
 266         s.writeFileInOutputDir(c, "dir1/file", "file")
 267         for _, err := range []error{
 268                 os.Symlink(s.cp.ctrOutputDir+"/dir1/file", s.cp.hostOutputDir+"/l_abs_file"),
 269                 os.Symlink(s.cp.ctrOutputDir+"/dir1/dir2", s.cp.hostOutputDir+"/l_abs_dir2"),
 270                 os.Symlink("../../dir1/file", s.cp.hostOutputDir+"/dir1/dir2/l_rel_file"),
 271                 os.Symlink("dir1/file", s.cp.hostOutputDir+"/l_rel_file"),
 272                 os.MkdirAll(s.cp.hostOutputDir+"/morelinks", 0755),
 273                 os.Symlink("../dir1/dir2", s.cp.hostOutputDir+"/morelinks/l_rel_dir2"),
 274                 os.Symlink("dir1/dir2/dir3", s.cp.hostOutputDir+"/l_rel_dir3"),
 275                 // rel. symlink -> rel. symlink -> regular file
 276                 os.Symlink("../dir1/dir2/l_rel_file", s.cp.hostOutputDir+"/morelinks/l_rel_l_rel_file"),
 277         } {
 278                 c.Assert(err, check.IsNil)
 279         }
 280
 281         err = s.cp.walkMount("", s.cp.ctrOutputDir, 10, true)
 282         c.Check(err, check.IsNil)
 283         c.Check(s.cp.dirs, check.DeepEquals, []string{
 284                 "/dir1", "/dir1/dir2", "/dir1/dir2/dir3",
 285                 "/l_abs_dir2", "/l_abs_dir2/dir3",
 286                 "/l_rel_dir3",
 287                 "/morelinks", "/morelinks/l_rel_dir2", "/morelinks/l_rel_dir2/dir3",
 288         })
 289         c.Check(s.cp.files, check.DeepEquals, []filetodo{
 290                 {dst: "/dir1/dir2/dir3/.keep", src: os.DevNull},
 291                 {dst: "/dir1/dir2/l_rel_file", src: hostfile, size: 4},
 292                 {dst: "/dir1/file", src: hostfile, size: 4},
 293                 {dst: "/l_abs_dir2/dir3/.keep", src: os.DevNull},
 294                 {dst: "/l_abs_dir2/l_rel_file", src: hostfile, size: 4},
 295                 {dst: "/l_abs_file", src: hostfile, size: 4},
 296                 {dst: "/l_rel_dir3/.keep", src: os.DevNull},
 297                 {dst: "/l_rel_file", src: hostfile, size: 4},
 298                 {dst: "/morelinks/l_rel_dir2/dir3/.keep", src: os.DevNull},
 299                 {dst: "/morelinks/l_rel_dir2/l_rel_file", src: hostfile, size: 4},
 300                 {dst: "/morelinks/l_rel_l_rel_file", src: hostfile, size: 4},
 301         })
 302 }
 303
 304 func (s *copierSuite) TestUnsupportedOutputMount(c *check.C) {
 305         s.cp.mounts["/ctr/outdir"] = arvados.Mount{Kind: "waz"}
 306         err := s.cp.walkMount("", s.cp.ctrOutputDir, 10, true)
 307         c.Check(err, check.NotNil)
 308 }
 309
 310 func (s *copierSuite) TestUnsupportedMountKindBelow(c *check.C) {
 311         s.cp.mounts["/ctr/outdir/dirk"] = arvados.Mount{Kind: "waz"}
 312         err := s.cp.walkMount("", s.cp.ctrOutputDir, 10, true)
 313         c.Check(err, check.NotNil)
 314 }
 315
 316 func (s *copierSuite) TestWritableMountBelow(c *check.C) {
 317         s.cp.mounts["/ctr/outdir/mount"] = arvados.Mount{
 318                 Kind:             "collection",
 319                 PortableDataHash: arvadostest.FooCollectionPDH,
 320                 Writable:         true,
 321         }
 322         c.Assert(os.MkdirAll(s.cp.hostOutputDir+"/mount", 0755), check.IsNil)
 323         s.writeFileInOutputDir(c, "file", "file")
 324         s.writeFileInOutputDir(c, "mount/foo", "foo")
 325
 326         err := s.cp.walkMount("", s.cp.ctrOutputDir, 10, true)
 327         c.Check(err, check.IsNil)
 328         c.Check(s.cp.dirs, check.DeepEquals, []string{"/mount"})
 329         c.Check(s.cp.files, check.DeepEquals, []filetodo{
 330                 {src: s.cp.hostOutputDir + "/file", dst: "/file", size: 4},
 331                 {src: s.cp.hostOutputDir + "/mount/foo", dst: "/mount/foo", size: 3},
 332         })
 333 }
 334
 335 // Check some glob-matching edge cases. In particular, check that
 336 // patterns like "foo/**" do not match regular files named "foo"
 337 // (unless of course they are inside a directory named "foo").
 338 func (s *copierSuite) TestMatchGlobs(c *check.C) {
 339         s.cp.globs = []string{"foo*/**"}
 340         c.Check(s.cp.matchGlobs("foo", true), check.Equals, true)
 341         c.Check(s.cp.matchGlobs("food", true), check.Equals, true)
 342         c.Check(s.cp.matchGlobs("foo", false), check.Equals, false)
 343         c.Check(s.cp.matchGlobs("food", false), check.Equals, false)
 344         c.Check(s.cp.matchGlobs("foo/bar", false), check.Equals, true)
 345         c.Check(s.cp.matchGlobs("food/bar", false), check.Equals, true)
 346         c.Check(s.cp.matchGlobs("foo/bar", true), check.Equals, true)
 347         c.Check(s.cp.matchGlobs("food/bar", true), check.Equals, true)
 348
 349         s.cp.globs = []string{"ba[!/]/foo*/**"}
 350         c.Check(s.cp.matchGlobs("bar/foo", true), check.Equals, true)
 351         c.Check(s.cp.matchGlobs("bar/food", true), check.Equals, true)
 352         c.Check(s.cp.matchGlobs("bar/foo", false), check.Equals, false)
 353         c.Check(s.cp.matchGlobs("bar/food", false), check.Equals, false)
 354         c.Check(s.cp.matchGlobs("bar/foo/z\\[", true), check.Equals, true)
 355         c.Check(s.cp.matchGlobs("bar/food/z\\[", true), check.Equals, true)
 356         c.Check(s.cp.matchGlobs("bar/foo/z\\[", false), check.Equals, true)
 357         c.Check(s.cp.matchGlobs("bar/food/z\\[", false), check.Equals, true)
 358
 359         s.cp.globs = []string{"waz/**/foo*/**"}
 360         c.Check(s.cp.matchGlobs("waz/quux/foo", true), check.Equals, true)
 361         c.Check(s.cp.matchGlobs("waz/quux/food", true), check.Equals, true)
 362         c.Check(s.cp.matchGlobs("waz/quux/foo", false), check.Equals, false)
 363         c.Check(s.cp.matchGlobs("waz/quux/food", false), check.Equals, false)
 364         c.Check(s.cp.matchGlobs("waz/quux/foo/foo", true), check.Equals, true)
 365         c.Check(s.cp.matchGlobs("waz/quux/food/foo", true), check.Equals, true)
 366         c.Check(s.cp.matchGlobs("waz/quux/foo/foo", false), check.Equals, true)
 367         c.Check(s.cp.matchGlobs("waz/quux/food/foo", false), check.Equals, true)
 368
 369         s.cp.globs = []string{"foo/**/*"}
 370         c.Check(s.cp.matchGlobs("foo", false), check.Equals, false)
 371         c.Check(s.cp.matchGlobs("foo/bar", false), check.Equals, true)
 372         c.Check(s.cp.matchGlobs("foo/bar/baz", false), check.Equals, true)
 373         c.Check(s.cp.matchGlobs("foo/bar/baz/waz", false), check.Equals, true)
 374 }
 375
 376 func (s *copierSuite) TestSubtreeCouldMatch(c *check.C) {
 377         for _, trial := range []struct {
 378                 mount string // relative to output dir
 379                 glob  string
 380                 could bool
 381         }{
 382                 {mount: "abc", glob: "*"},
 383                 {mount: "abc", glob: "abc/*", could: true},
 384                 {mount: "abc", glob: "a*/**", could: true},
 385                 {mount: "abc", glob: "**", could: true},
 386                 {mount: "abc", glob: "*/*", could: true},
 387                 {mount: "abc", glob: "**/*.txt", could: true},
 388                 {mount: "abc/def", glob: "*"},
 389                 {mount: "abc/def", glob: "*/*"},
 390                 {mount: "abc/def", glob: "*/*.txt"},
 391                 {mount: "abc/def", glob: "*/*/*", could: true},
 392                 {mount: "abc/def", glob: "**", could: true},
 393                 {mount: "abc/def", glob: "**/bar", could: true},
 394                 {mount: "abc/def", glob: "abc/**", could: true},
 395                 {mount: "abc/def/ghi", glob: "*c/**/bar", could: true},
 396                 {mount: "abc/def/ghi", glob: "*c/*f/bar"},
 397                 {mount: "abc/def/ghi", glob: "abc/d[^/]f/ghi/*", could: true},
 398         } {
 399                 c.Logf("=== %+v", trial)
 400                 got := (&copier{
 401                         globs: []string{trial.glob},
 402                 }).subtreeCouldMatch(trial.mount)
 403                 c.Check(got, check.Equals, trial.could)
 404         }
 405 }
 406
 407 func (s *copierSuite) TestCopyFromLargeCollection_Readonly(c *check.C) {
 408         s.testCopyFromLargeCollection(c, false)
 409 }
 410
 411 func (s *copierSuite) TestCopyFromLargeCollection_Writable(c *check.C) {
 412         s.testCopyFromLargeCollection(c, true)
 413 }
 414
 415 func (s *copierSuite) testCopyFromLargeCollection(c *check.C, writable bool) {
 416         bindtmp := c.MkDir()
 417         mtxt := arvadostest.FakeManifest(100, 100, 2, 4<<20)
 418         pdh := arvados.PortableDataHash(mtxt)
 419         json, err := json.Marshal(arvados.Collection{ManifestText: mtxt, PortableDataHash: pdh})
 420         c.Assert(err, check.IsNil)
 421         err = os.WriteFile(bindtmp+"/.arvados#collection", json, 0644)
 422         // This symlink tricks walkHostFS into calling walkMount on
 423         // the fakecollection dir. If we did the obvious thing instead
 424         // (i.e., mount a collection under the output dir) walkMount
 425         // would see that our fakecollection dir is actually a regular
 426         // directory, conclude that the mount has been deleted and
 427         // replaced by a regular directory tree, and process the tree
 428         // as regular files, bypassing the manifest-copying code path
 429         // we're trying to test.
 430         err = os.Symlink("/fakecollection", s.cp.hostOutputDir+"/fakecollection")
 431         c.Assert(err, check.IsNil)
 432         s.cp.mounts["/fakecollection"] = arvados.Mount{
 433                 Kind:             "collection",
 434                 PortableDataHash: pdh,
 435                 Writable:         writable,
 436         }
 437         s.cp.bindmounts = map[string]bindmount{
 438                 "/fakecollection": bindmount{HostPath: bindtmp, ReadOnly: !writable},
 439         }
 440         s.cp.manifestCache = map[string]string{pdh: mtxt}
 441         err = s.cp.walkMount("", s.cp.ctrOutputDir, 10, true)
 442         c.Check(err, check.IsNil)
 443         c.Log(s.log.String())
 444
 445         // Check some files to ensure they were copied properly.
 446         // Specifically, arbitrarily check every 17th file in every
 447         // 13th dir.  (This is better than checking all of the files
 448         // only in that it's less likely to show up as a distracting
 449         // signal in CPU profiling.)
 450         for i := 0; i < 100; i += 13 {
 451                 for j := 0; j < 100; j += 17 {
 452                         fnm := fmt.Sprintf("/fakecollection/dir%d/dir%d/file%d", i, j, j)
 453                         _, err := s.cp.staged.Stat(fnm)
 454                         c.Assert(err, check.IsNil, check.Commentf("%s", fnm))
 455                 }
 456         }
 457 }
 458
 459 func (s *copierSuite) TestMountBelowExcludedByGlob(c *check.C) {
 460         bindtmp := c.MkDir()
 461         s.cp.mounts["/ctr/outdir/include/includer"] = arvados.Mount{
 462                 Kind:             "collection",
 463                 PortableDataHash: arvadostest.FooCollectionPDH,
 464         }
 465         s.cp.mounts["/ctr/outdir/include/includew"] = arvados.Mount{
 466                 Kind:             "collection",
 467                 PortableDataHash: arvadostest.FooCollectionPDH,
 468                 Writable:         true,
 469         }
 470         s.cp.mounts["/ctr/outdir/exclude/excluder"] = arvados.Mount{
 471                 Kind:             "collection",
 472                 PortableDataHash: arvadostest.FooCollectionPDH,
 473         }
 474         s.cp.mounts["/ctr/outdir/exclude/excludew"] = arvados.Mount{
 475                 Kind:             "collection",
 476                 PortableDataHash: arvadostest.FooCollectionPDH,
 477                 Writable:         true,
 478         }
 479         s.cp.mounts["/ctr/outdir/nonexistent/collection"] = arvados.Mount{
 480                 // As extra assurance, plant a collection that will
 481                 // fail if copier attempts to load its manifest.  (For
 482                 // performance reasons it's important that copier
 483                 // doesn't try to load the manifest before deciding
 484                 // not to copy the contents.)
 485                 Kind:             "collection",
 486                 PortableDataHash: "aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa+1234",
 487         }
 488         s.cp.globs = []string{
 489                 "?ncl*/*r/*",
 490                 "*/?ncl*/**",
 491         }
 492         c.Assert(os.MkdirAll(s.cp.hostOutputDir+"/include/includer", 0755), check.IsNil)
 493         c.Assert(os.MkdirAll(s.cp.hostOutputDir+"/include/includew", 0755), check.IsNil)
 494         c.Assert(os.MkdirAll(s.cp.hostOutputDir+"/exclude/excluder", 0755), check.IsNil)
 495         c.Assert(os.MkdirAll(s.cp.hostOutputDir+"/exclude/excludew", 0755), check.IsNil)
 496         s.writeFileInOutputDir(c, "include/includew/foo", "foo")
 497         s.writeFileInOutputDir(c, "exclude/excludew/foo", "foo")
 498         s.cp.bindmounts = map[string]bindmount{
 499                 "/ctr/outdir/include/includew": bindmount{HostPath: bindtmp, ReadOnly: false},
 500         }
 501         s.cp.bindmounts = map[string]bindmount{
 502                 "/ctr/outdir/include/excludew": bindmount{HostPath: bindtmp, ReadOnly: false},
 503         }
 504
 505         err := s.cp.walkMount("", s.cp.ctrOutputDir, 10, true)
 506         c.Check(err, check.IsNil)
 507         c.Log(s.log.String())
 508
 509         // Note it's OK that "/exclude" is not excluded by walkMount:
 510         // it is just a local filesystem directory, not a mount point
 511         // that's expensive to walk.  In real-life usage, it will be
 512         // removed from cp.dirs before any copying happens.
 513         c.Check(s.cp.dirs, check.DeepEquals, []string{"/exclude", "/include", "/include/includew"})
 514         c.Check(s.cp.files, check.DeepEquals, []filetodo{
 515                 {src: s.cp.hostOutputDir + "/include/includew/foo", dst: "/include/includew/foo", size: 3},
 516         })
 517         manifest, err := s.cp.staged.MarshalManifest(".")
 518         c.Assert(err, check.IsNil)
 519         c.Check(manifest, check.Matches, `(?ms).*\./include/includer .*`)
 520         c.Check(manifest, check.Not(check.Matches), `(?ms).*exclude.*`)
 521         c.Check(s.log.String(), check.Matches, `(?ms).*not copying \\"exclude/excluder\\".*`)
 522         c.Check(s.log.String(), check.Matches, `(?ms).*not copying \\"nonexistent/collection\\".*`)
 523 }
 524
 525 func (s *copierSuite) writeFileInOutputDir(c *check.C, path, data string) {
 526         f, err := os.OpenFile(s.cp.hostOutputDir+"/"+path, os.O_CREATE|os.O_WRONLY, 0644)
 527         c.Assert(err, check.IsNil)
 528         _, err = io.WriteString(f, data)
 529         c.Assert(err, check.IsNil)
 530         c.Assert(f.Close(), check.IsNil)
 531 }
 532
 533 // applyGlobsToFilesAndDirs uses the same glob-matching code as
 534 // applyGlobsToStaged, so we don't need to test all of the same
 535 // glob-matching behavior covered in TestApplyGlobsToCollectionFS.  We
 536 // do need to check that (a) the glob is actually being used to filter
 537 // out files, and (b) non-matching dirs still included if and only if
 538 // they are ancestors of matching files.
 539 func (s *copierSuite) TestApplyGlobsToFilesAndDirs(c *check.C) {
 540         dirs := []string{"dir1", "dir1/dir11", "dir1/dir12", "dir2"}
 541         files := []string{"dir1/file11", "dir1/dir11/file111", "dir2/file2"}
 542         for _, trial := range []struct {
 543                 globs []string
 544                 dirs  []string
 545                 files []string
 546         }{
 547                 {
 548                         globs: []string{},
 549                         dirs:  append([]string{}, dirs...),
 550                         files: append([]string{}, files...),
 551                 },
 552                 {
 553                         globs: []string{"**"},
 554                         dirs:  append([]string{}, dirs...),
 555                         files: append([]string{}, files...),
 556                 },
 557                 {
 558                         globs: []string{"**/file111"},
 559                         dirs:  []string{"dir1", "dir1/dir11"},
 560                         files: []string{"dir1/dir11/file111"},
 561                 },
 562                 {
 563                         globs: []string{"nothing"},
 564                         dirs:  nil,
 565                         files: nil,
 566                 },
 567                 {
 568                         globs: []string{"**/dir12"},
 569                         dirs:  []string{"dir1", "dir1/dir12"},
 570                         files: nil,
 571                 },
 572                 {
 573                         globs: []string{"**/file*"},
 574                         dirs:  []string{"dir1", "dir1/dir11", "dir2"},
 575                         files: append([]string{}, files...),
 576                 },
 577                 {
 578                         globs: []string{"**/dir1[12]"},
 579                         dirs:  []string{"dir1", "dir1/dir11", "dir1/dir12"},
 580                         files: nil,
 581                 },
 582                 {
 583                         globs: []string{"**/dir1[^2]"},
 584                         dirs:  []string{"dir1", "dir1/dir11"},
 585                         files: nil,
 586                 },
 587                 {
 588                         globs: []string{"dir1/**"},
 589                         dirs:  []string{"dir1", "dir1/dir11", "dir1/dir12"},
 590                         files: []string{"dir1/file11", "dir1/dir11/file111"},
 591                 },
 592         } {
 593                 c.Logf("=== globs: %q", trial.globs)
 594                 cp := copier{
 595                         globs: trial.globs,
 596                         dirs:  dirs,
 597                 }
 598                 for _, path := range files {
 599                         cp.files = append(cp.files, filetodo{dst: path})
 600                 }
 601                 cp.applyGlobsToFilesAndDirs()
 602                 var gotFiles []string
 603                 for _, file := range cp.files {
 604                         gotFiles = append(gotFiles, file.dst)
 605                 }
 606                 c.Check(cp.dirs, check.DeepEquals, trial.dirs)
 607                 c.Check(gotFiles, check.DeepEquals, trial.files)
 608         }
 609 }
 610
 611 func (s *copierSuite) TestApplyGlobsToCollectionFS(c *check.C) {
 612         for _, trial := range []struct {
 613                 globs  []string
 614                 expect []string
 615         }{
 616                 {
 617                         globs:  nil,
 618                         expect: []string{"foo", "bar", "baz/quux", "baz/parent1/item1"},
 619                 },
 620                 {
 621                         globs:  []string{"foo"},
 622                         expect: []string{"foo"},
 623                 },
 624                 {
 625                         globs:  []string{"baz/parent1/item1"},
 626                         expect: []string{"baz/parent1/item1"},
 627                 },
 628                 {
 629                         globs:  []string{"**"},
 630                         expect: []string{"foo", "bar", "baz/quux", "baz/parent1/item1"},
 631                 },
 632                 {
 633                         globs:  []string{"**/*"},
 634                         expect: []string{"foo", "bar", "baz/quux", "baz/parent1/item1"},
 635                 },
 636                 {
 637                         globs:  []string{"*"},
 638                         expect: []string{"foo", "bar"},
 639                 },
 640                 {
 641                         globs:  []string{"baz"},
 642                         expect: nil,
 643                 },
 644                 {
 645                         globs:  []string{"b*/**"},
 646                         expect: []string{"baz/quux", "baz/parent1/item1"},
 647                 },
 648                 {
 649                         globs:  []string{"baz"},
 650                         expect: nil,
 651                 },
 652                 {
 653                         globs:  []string{"baz/**"},
 654                         expect: []string{"baz/quux", "baz/parent1/item1"},
 655                 },
 656                 {
 657                         globs:  []string{"baz/*"},
 658                         expect: []string{"baz/quux"},
 659                 },
 660                 {
 661                         globs:  []string{"baz/**/*uu?"},
 662                         expect: []string{"baz/quux"},
 663                 },
 664                 {
 665                         globs:  []string{"**/*m1"},
 666                         expect: []string{"baz/parent1/item1"},
 667                 },
 668                 {
 669                         globs:  []string{"*/*/*/**/*1"},
 670                         expect: nil,
 671                 },
 672                 {
 673                         globs:  []string{"f*", "**/q*"},
 674                         expect: []string{"foo", "baz/quux"},
 675                 },
 676                 {
 677                         globs:  []string{"\\"}, // invalid pattern matches nothing
 678                         expect: nil,
 679                 },
 680                 {
 681                         globs:  []string{"\\", "foo"},
 682                         expect: []string{"foo"},
 683                 },
 684                 {
 685                         globs:  []string{"foo/**"},
 686                         expect: nil,
 687                 },
 688                 {
 689                         globs:  []string{"foo*/**"},
 690                         expect: nil,
 691                 },
 692         } {
 693                 c.Logf("=== globs: %q", trial.globs)
 694                 collfs, err := (&arvados.Collection{ManifestText: ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:foo 0:0:bar 0:0:baz/quux 0:0:baz/parent1/item1\n"}).FileSystem(nil, nil)
 695                 c.Assert(err, check.IsNil)
 696                 cp := copier{globs: trial.globs, staged: collfs}
 697                 err = cp.applyGlobsToStaged()
 698                 if !c.Check(err, check.IsNil) {
 699                         continue
 700                 }
 701                 var got []string
 702                 fs.WalkDir(arvados.FS(collfs), "", func(path string, ent fs.DirEntry, err error) error {
 703                         if !ent.IsDir() {
 704                                 got = append(got, path)
 705                         }
 706                         return nil
 707                 })
 708                 sort.Strings(got)
 709                 sort.Strings(trial.expect)
 710                 c.Check(got, check.DeepEquals, trial.expect)
 711         }
 712 }