Golang pfsutil.StartCommit函数代码示例

func TestFromCommit(t *testing.T) {
	t.Parallel()
	repo := uniqueString("TestFromCommit")
	pachClient := getPachClient(t)
	seed := time.Now().UnixNano()
	rand := rand.New(rand.NewSource(seed))
	err := pfsutil.CreateRepo(pachClient, repo)
	require.NoError(t, err)
	commit1, err := pfsutil.StartCommit(pachClient, repo, "")
	require.NoError(t, err)
	_, err = pfsutil.PutFile(pachClient, repo, commit1.Id, "file", 0, workload.NewReader(rand, KB))
	require.NoError(t, err)
	err = pfsutil.FinishCommit(pachClient, repo, commit1.Id)
	require.NoError(t, err)
	commit2, err := pfsutil.StartCommit(pachClient, repo, commit1.Id)
	require.NoError(t, err)
	_, err = pfsutil.PutFile(pachClient, repo, commit2.Id, "file", 0, workload.NewReader(rand, KB))
	require.NoError(t, err)
	err = pfsutil.FinishCommit(pachClient, repo, commit2.Id)
	require.NoError(t, err)
	var buffer bytes.Buffer
	require.NoError(t, pfsutil.GetFile(pachClient, repo, commit2.Id, "file", 0, 0, commit1.Id, nil, &buffer))
	require.Equal(t, buffer.Len(), KB)
	buffer = bytes.Buffer{}
	require.NoError(t, pfsutil.GetFile(pachClient, repo, commit2.Id, "file", 0, 0, "", nil, &buffer))
	require.Equal(t, buffer.Len(), 2*KB)
}

func testFailures(t *testing.T, apiClient pfs.APIClient, cluster Cluster) {
	repoName := "testFailuresRepo"

	err := pfsutil.CreateRepo(apiClient, repoName)
	require.NoError(t, err)

	commit, err := pfsutil.StartCommit(apiClient, repoName, "")
	require.NoError(t, err)
	require.NotNil(t, commit)
	newCommitID := commit.Id

	err = pfsutil.MakeDirectory(apiClient, repoName, newCommitID, "a/b")
	require.NoError(t, err)
	err = pfsutil.MakeDirectory(apiClient, repoName, newCommitID, "a/c")
	require.NoError(t, err)

	doWrites(t, apiClient, repoName, newCommitID)

	err = pfsutil.FinishCommit(apiClient, repoName, newCommitID)
	require.NoError(t, err)

	checkWrites(t, apiClient, repoName, newCommitID)

	cluster.KillRoleAssigner()
	for server := 0; server < testNumReplicas; server++ {
		cluster.Kill(server)
	}
	cluster.RestartRoleAssigner()
	cluster.WaitForAvailability()

	checkWrites(t, apiClient, repoName, newCommitID)
}

func testMountBig(t *testing.T, apiClient pfs.APIClient, cluster Cluster) {
	repoName := "testMountBigRepo"

	err := pfsutil.CreateRepo(apiClient, repoName)
	require.NoError(t, err)

	directory := "/compile/testMount"
	mounter := fuse.NewMounter("localhost", apiClient)
	ready := make(chan bool)
	go func() {
		err = mounter.Mount(directory, 0, 1, ready)
		require.NoError(t, err)
	}()
	<-ready

	_, err = os.Stat(filepath.Join(directory, repoName))
	require.NoError(t, err)

	_, err = os.Stat(filepath.Join(directory, repoName, "scratch"))
	require.NoError(t, err)

	commit, err := pfsutil.StartCommit(apiClient, repoName, "scratch")
	require.NoError(t, err)
	require.NotNil(t, commit)
	newCommitID := commit.Id

	bigValue := make([]byte, 1024*1024*300)
	for i := 0; i < 1024*1024*300; i++ {
		bigValue[i] = 'a'
	}

	wg := sync.WaitGroup{}
	for j := 0; j < 5; j++ {
		wg.Add(1)
		go func(j int) {
			defer wg.Done()
			err := ioutil.WriteFile(filepath.Join(directory, repoName, newCommitID, fmt.Sprintf("big%d", j)), bigValue, 0666)
			require.NoError(t, err)
		}(j)
	}
	wg.Wait()

	err = pfsutil.FinishCommit(apiClient, repoName, newCommitID)
	require.NoError(t, err)

	wg = sync.WaitGroup{}
	for j := 0; j < 5; j++ {
		wg.Add(1)
		go func(j int) {
			defer wg.Done()
			data, err := ioutil.ReadFile(filepath.Join(directory, repoName, newCommitID, fmt.Sprintf("big%d", j)))
			require.NoError(t, err)
			require.Equal(t, bigValue, data)
		}(j)
	}
	wg.Wait()

	err = mounter.Unmount(directory)
	require.NoError(t, err)
}

func TestMountBig(t *testing.T) {
	t.Skip()
	t.Parallel()
	apiClient := getPfsClient(t)
	repoName := uniqueString("testMountBigRepo")

	err := pfsutil.CreateRepo(apiClient, repoName)
	require.NoError(t, err)

	directory := "/compile/testMount"
	mounter := fuse.NewMounter("localhost", apiClient)
	ready := make(chan bool)
	go func() {
		err = mounter.Mount(directory, &pfs.Shard{Number: 0, Modulus: 1}, nil, ready)
		require.NoError(t, err)
	}()
	<-ready

	_, err = os.Stat(filepath.Join(directory, repoName))
	require.NoError(t, err)

	commit, err := pfsutil.StartCommit(apiClient, repoName, "")
	require.NoError(t, err)
	require.NotNil(t, commit)
	newCommitID := commit.Id

	bigValue := make([]byte, 1024*1024*300)
	for i := 0; i < 1024*1024*300; i++ {
		bigValue[i] = 'a'
	}

	wg := sync.WaitGroup{}
	for j := 0; j < 5; j++ {
		wg.Add(1)
		go func(j int) {
			defer wg.Done()
			err := ioutil.WriteFile(filepath.Join(directory, repoName, newCommitID, fmt.Sprintf("big%d", j)), bigValue, 0666)
			require.NoError(t, err)
		}(j)
	}
	wg.Wait()

	err = pfsutil.FinishCommit(apiClient, repoName, newCommitID)
	require.NoError(t, err)

	wg = sync.WaitGroup{}
	for j := 0; j < 5; j++ {
		wg.Add(1)
		go func(j int) {
			defer wg.Done()
			data, err := ioutil.ReadFile(filepath.Join(directory, repoName, newCommitID, fmt.Sprintf("big%d", j)))
			require.NoError(t, err)
			require.Equal(t, bigValue, data)
		}(j)
	}
	wg.Wait()

	err = mounter.Unmount(directory)
	require.NoError(t, err)
}

func TestJob(t *testing.T) {
	t.Parallel()
	dataRepo := uniqueString("TestJob.data")
	pachClient := getPachClient(t)
	require.NoError(t, pfsutil.CreateRepo(pachClient, dataRepo))
	commit, err := pfsutil.StartCommit(pachClient, dataRepo, "")
	require.NoError(t, err)
	_, err = pfsutil.PutFile(pachClient, dataRepo, commit.Id, "file", 0, strings.NewReader("foo\n"))
	require.NoError(t, err)
	require.NoError(t, pfsutil.FinishCommit(pachClient, dataRepo, commit.Id))
	job, err := ppsutil.CreateJob(
		pachClient,
		"",
		[]string{"cp", path.Join("/pfs", dataRepo, "file"), "/pfs/out/file"},
		nil,
		1,
		[]*pps.JobInput{{Commit: commit}},
		"",
	)
	require.NoError(t, err)
	inspectJobRequest := &pps.InspectJobRequest{
		Job:         job,
		BlockOutput: true,
		BlockState:  true,
	}
	jobInfo, err := pachClient.InspectJob(context.Background(), inspectJobRequest)
	require.NoError(t, err)
	require.Equal(t, pps.JobState_JOB_STATE_SUCCESS.String(), jobInfo.State.String())
	commitInfo, err := pfsutil.InspectCommit(pachClient, jobInfo.OutputCommit.Repo.Name, jobInfo.OutputCommit.Id)
	require.NoError(t, err)
	require.Equal(t, pfs.CommitType_COMMIT_TYPE_READ, commitInfo.CommitType)
	var buffer bytes.Buffer
	require.NoError(t, pfsutil.GetFile(pachClient, jobInfo.OutputCommit.Repo.Name, jobInfo.OutputCommit.Id, "file", 0, 0, "", nil, &buffer))
	require.Equal(t, "foo\n", buffer.String())
}

func TestJob(t *testing.T) {
	dataRepo := uniqueString("TestJob.data")
	pfsClient := getPfsClient(t)
	require.NoError(t, pfsutil.CreateRepo(pfsClient, dataRepo))
	commit, err := pfsutil.StartCommit(pfsClient, dataRepo, "")
	require.NoError(t, err)
	_, err = pfsutil.PutFile(pfsClient, dataRepo, commit.Id, "file", 0, strings.NewReader("foo"))
	require.NoError(t, err)
	require.NoError(t, pfsutil.FinishCommit(pfsClient, dataRepo, commit.Id))
	ppsClient := getPpsClient(t)
	job, err := ppsutil.CreateJob(
		ppsClient,
		"",
		[]string{"cp", path.Join("/pfs", dataRepo, "file"), "/pfs/out/file"},
		"",
		1,
		[]*pfs.Commit{commit},
		"",
	)
	require.NoError(t, err)
	inspectJobRequest := &pps.InspectJobRequest{
		Job:         job,
		BlockOutput: true,
		BlockState:  true,
	}
	jobInfo, err := ppsClient.InspectJob(context.Background(), inspectJobRequest)
	require.NoError(t, err)
	var buffer bytes.Buffer
	require.NoError(t, pfsutil.GetFile(pfsClient, jobInfo.OutputCommit.Repo.Name, jobInfo.OutputCommit.Id, "file", 0, 0, nil, &buffer))
	require.Equal(t, "foo", buffer.String())
}

func BenchmarkFuse(b *testing.B) {
	apiClient := getPfsClient(b)
	repoName := uniqueString("benchMountRepo")

	if err := pfsutil.CreateRepo(apiClient, repoName); err != nil {
		b.Error(err)
	}

	directory := "/compile/benchMount"
	mounter := fuse.NewMounter("localhost", apiClient)
	ready := make(chan bool)
	go func() {
		err := mounter.Mount(directory, &pfs.Shard{Number: 0, Modulus: 1}, nil, ready)
		require.NoError(b, err)
	}()
	<-ready

	defer func() {
		if err := mounter.Unmount(directory); err != nil {
			b.Error(err)
		}
	}()

	bigValue := make([]byte, 1024*1024)
	for i := 0; i < 1024*1024; i++ {
		bigValue[i] = 'a'
	}

	b.ResetTimer()
	for i := 0; i < b.N; i++ {
		commit, err := pfsutil.StartCommit(apiClient, repoName, "")
		if err != nil {
			b.Error(err)
		}
		if commit == nil {
			b.Error("nil branch")
		}
		newCommitID := commit.Id
		var wg sync.WaitGroup
		for j := 0; j < 1024; j++ {
			wg.Add(1)
			go func(j int) {
				defer wg.Done()
				if err = ioutil.WriteFile(filepath.Join(directory, repoName, newCommitID, fmt.Sprintf("big%d", j)), bigValue, 0666); err != nil {
					b.Error(err)
				}
			}(j)
		}
		wg.Wait()
		if err := pfsutil.FinishCommit(apiClient, repoName, newCommitID); err != nil {
			b.Error(err)
		}
	}
}

func benchMount(b *testing.B, apiClient pfs.ApiClient) {
	repositoryName := "benchMountRepo"

	if err := pfsutil.CreateRepo(apiClient, repositoryName); err != nil {
		b.Error(err)
	}

	directory := "/compile/benchMount"
	mounter := fuse.NewMounter(apiClient)
	if err := mounter.Mount(repositoryName, "", directory, 0, 1); err != nil {
		b.Error(err)
	}

	defer func() {
		if err := mounter.Unmount(directory); err != nil {
			b.Error(err)
		}
		if err := mounter.Wait(directory); err != nil {
			b.Error(err)
		}
	}()

	bigValue := make([]byte, 1024*1024)
	for i := 0; i < 1024*1024; i++ {
		bigValue[i] = 'a'
	}

	b.ResetTimer()
	for i := 0; i < b.N; i++ {
		commit, err := pfsutil.StartCommit(apiClient, repositoryName, "scratch")
		if err != nil {
			b.Error(err)
		}
		if commit == nil {
			b.Error("nil branch")
		}
		newCommitID := commit.Id
		var wg sync.WaitGroup
		for j := 0; j < 1024; j++ {
			wg.Add(1)
			go func(j int) {
				defer wg.Done()
				if err = ioutil.WriteFile(filepath.Join(directory, newCommitID, fmt.Sprintf("big%d", j)), bigValue, 0666); err != nil {
					b.Error(err)
				}
			}(j)
		}
		wg.Wait()
		if err := pfsutil.FinishCommit(apiClient, repositoryName, newCommitID); err != nil {
			b.Error(err)
		}
	}
}

func TestGrep(t *testing.T) {
	t.Skip()
	t.Parallel()
	dataRepo := uniqueString("TestGrep.data")
	pfsClient := getPfsClient(t)
	require.NoError(t, pfsutil.CreateRepo(pfsClient, dataRepo))
	commit, err := pfsutil.StartCommit(pfsClient, dataRepo, "")
	require.NoError(t, err)
	for i := 0; i < 100; i++ {
		_, err = pfsutil.PutFile(pfsClient, dataRepo, commit.Id, fmt.Sprintf("file%d", i), 0, strings.NewReader("foo\nbar\nfizz\nbuzz\n"))
		require.NoError(t, err)
	}
	require.NoError(t, pfsutil.FinishCommit(pfsClient, dataRepo, commit.Id))
	ppsClient := getPpsClient(t)
	job1, err := ppsutil.CreateJob(
		ppsClient,
		"",
		[]string{"bash"},
		fmt.Sprintf("grep foo /pfs/%s/* >/pfs/out/foo", dataRepo),
		1,
		[]*pps.JobInput{{Commit: commit}},
		"",
	)
	require.NoError(t, err)
	job2, err := ppsutil.CreateJob(
		ppsClient,
		"",
		[]string{"bash"},
		fmt.Sprintf("grep foo /pfs/%s/* >/pfs/out/foo", dataRepo),
		4,
		[]*pps.JobInput{{Commit: commit}},
		"",
	)
	require.NoError(t, err)
	inspectJobRequest := &pps.InspectJobRequest{
		Job:         job1,
		BlockOutput: true,
		BlockState:  true,
	}
	job1Info, err := ppsClient.InspectJob(context.Background(), inspectJobRequest)
	require.NoError(t, err)
	inspectJobRequest.Job = job2
	job2Info, err := ppsClient.InspectJob(context.Background(), inspectJobRequest)
	require.NoError(t, err)
	repo1Info, err := pfsutil.InspectRepo(pfsClient, job1Info.OutputCommit.Repo.Name)
	require.NoError(t, err)
	repo2Info, err := pfsutil.InspectRepo(pfsClient, job2Info.OutputCommit.Repo.Name)
	require.NoError(t, err)
	require.Equal(t, repo1Info.SizeBytes, repo2Info.SizeBytes)
}

func TestSharding(t *testing.T) {
	t.Parallel()
	repo := uniqueString("TestSharding")
	pachClient := getPachClient(t)
	err := pfsutil.CreateRepo(pachClient, repo)
	require.NoError(t, err)
	commit, err := pfsutil.StartCommit(pachClient, repo, "")
	require.NoError(t, err)
	var wg sync.WaitGroup
	for i := 0; i < NUMFILES; i++ {
		i := i
		wg.Add(1)
		go func() {
			defer wg.Done()
			rand := rand.New(rand.NewSource(int64(i)))
			_, err = pfsutil.PutFile(pachClient, repo, commit.Id, fmt.Sprintf("file%d", i), 0, workload.NewReader(rand, KB))
			require.NoError(t, err)
		}()
	}
	wg.Wait()
	err = pfsutil.FinishCommit(pachClient, repo, commit.Id)
	require.NoError(t, err)
	wg = sync.WaitGroup{}
	for i := 0; i < NUMFILES; i++ {
		i := i
		wg.Add(1)
		go func() {
			defer wg.Done()
			var buffer1Shard bytes.Buffer
			var buffer4Shard bytes.Buffer
			shard := &pfs.Shard{FileModulus: 1, BlockModulus: 1}
			err := pfsutil.GetFile(pachClient, repo, commit.Id,
				fmt.Sprintf("file%d", i), 0, 0, "", shard, &buffer1Shard)
			require.NoError(t, err)
			shard.BlockModulus = 4
			for blockNumber := uint64(0); blockNumber < 4; blockNumber++ {
				shard.BlockNumber = blockNumber
				err := pfsutil.GetFile(pachClient, repo, commit.Id,
					fmt.Sprintf("file%d", i), 0, 0, "", shard, &buffer4Shard)
				require.NoError(t, err)
			}
			require.Equal(t, buffer1Shard.Len(), buffer4Shard.Len())
		}()
	}
	wg.Wait()
}

func TestBigWrite(t *testing.T) {
	t.Parallel()
	protolog.SetLevel(protolog.Level_LEVEL_DEBUG)
	repo := uniqueString("TestBigWrite")
	pfsClient := getPfsClient(t)
	err := pfsutil.CreateRepo(pfsClient, repo)
	require.NoError(t, err)
	commit, err := pfsutil.StartCommit(pfsClient, repo, "")
	require.NoError(t, err)
	rand := rand.New(rand.NewSource(5))
	_, err = pfsutil.PutFile(pfsClient, repo, commit.Id, "file", 0, workload.NewReader(rand, 10000))
	require.NoError(t, err)
	err = pfsutil.FinishCommit(pfsClient, repo, commit.Id)
	require.NoError(t, err)
	var buffer bytes.Buffer
	err = pfsutil.GetFile(pfsClient, repo, commit.Id, "file", 0, 0, nil, &buffer)
	require.NoError(t, err)
}

func (b *brancher) branch(
	repositoryName string,
	commitID string,
) (string, error) {
	commit, err := pfsutil.StartCommit(
		b.pfsAPIClient,
		repositoryName,
		commitID,
	)
	if err != nil {
		return "", err
	}
	b.lock.Lock()
	defer b.lock.Unlock()
	if existingCommitID, ok := b.outputRepositoryToBranchID[repositoryName]; ok {
		// TODO(pedge) delete new branch
		return existingCommitID, nil
	}
	b.outputRepositoryToBranchID[repositoryName] = commit.Id
	return commit.Id, nil
}

func TestSharding(t *testing.T) {
	t.Parallel()
	repo := uniqueString("TestSharding")
	pfsClient := getPfsClient(t)
	err := pfsutil.CreateRepo(pfsClient, repo)
	require.NoError(t, err)
	commit, err := pfsutil.StartCommit(pfsClient, repo, "")
	require.NoError(t, err)
	var wg sync.WaitGroup
	for i := 0; i < NUMFILES; i++ {
		i := i
		wg.Add(1)
		go func() {
			defer wg.Done()
			rand := rand.New(rand.NewSource(int64(i)))
			_, err = pfsutil.PutFile(pfsClient, repo, commit.Id, fmt.Sprintf("file%d", i), 0, workload.NewReader(rand, 128*1024*1024))
			require.NoError(t, err)
		}()
	}
	wg.Wait()
	err = pfsutil.FinishCommit(pfsClient, repo, commit.Id)
	require.NoError(t, err)
}

func TestGrep(t *testing.T) {
	t.Parallel()
	dataRepo := uniqueString("pachyderm.TestGrep.data")
	pfsClient := getPfsClient(t)
	require.NoError(t, pfsutil.CreateRepo(pfsClient, dataRepo))
	commit, err := pfsutil.StartCommit(pfsClient, dataRepo, "")
	require.NoError(t, err)
	for i := 0; i < 100; i++ {
		_, err = pfsutil.PutFile(pfsClient, dataRepo, commit.Id, fmt.Sprintf("file%d", i), 0, strings.NewReader("foo\nbar\nfizz\nbuzz\n"))
		require.NoError(t, err)
	}
	require.NoError(t, pfsutil.FinishCommit(pfsClient, dataRepo, commit.Id))
	ppsClient := getPpsClient(t)
	_, err = ppsutil.CreateJob(
		ppsClient,
		"",
		[]string{"bash"},
		fmt.Sprintf("grep foo /pfs/%s/* >/pfs/out/foo", dataRepo),
		1,
		[]*pps.JobInput{{Commit: commit}},
		"",
	)
	require.NoError(t, err)
}

func TestBlockListCommits(t *testing.T) {
	t.Parallel()
	apiClient := getPfsClient(t)
	repoName := uniqueString("testBlockListCommitsRepo")

	err := pfsutil.CreateRepo(apiClient, repoName)
	require.NoError(t, err)

	baseCommit, err := pfsutil.StartCommit(apiClient, repoName, "")
	require.NoError(t, err)
	err = pfsutil.FinishCommit(apiClient, repoName, baseCommit.Id)
	require.NoError(t, err)

	repo := &pfs.Repo{
		Name: repoName,
	}
	listCommitRequest := &pfs.ListCommitRequest{
		Repo:       []*pfs.Repo{repo},
		FromCommit: []*pfs.Commit{baseCommit},
	}
	commitInfos, err := apiClient.ListCommit(
		context.Background(),
		listCommitRequest,
	)
	require.NoError(t, err)
	require.Equal(t, len(commitInfos.CommitInfo), 0)

	var newCommit *pfs.Commit
	var wg sync.WaitGroup
	wg.Add(1)
	go func() {
		defer wg.Done()
		time.Sleep(1)
		commit, err := pfsutil.StartCommit(apiClient, repoName, baseCommit.Id)
		require.NoError(t, err)
		require.NotNil(t, commit)
		newCommit = commit
	}()
	listCommitRequest.Block = true
	listCommitRequest.CommitType = pfs.CommitType_COMMIT_TYPE_WRITE
	commitInfos, err = apiClient.ListCommit(
		context.Background(),
		listCommitRequest,
	)
	wg.Wait()
	require.NoError(t, err)
	require.Equal(t, len(commitInfos.CommitInfo), 1)
	require.Equal(t, newCommit, commitInfos.CommitInfo[0].Commit)

	wg = sync.WaitGroup{}
	wg.Add(1)
	go func() {
		defer wg.Done()
		time.Sleep(1)
		err := pfsutil.FinishCommit(apiClient, repoName, newCommit.Id)
		require.NoError(t, err)
	}()
	listCommitRequest.Block = true
	listCommitRequest.CommitType = pfs.CommitType_COMMIT_TYPE_READ
	commitInfos, err = apiClient.ListCommit(
		context.Background(),
		listCommitRequest,
	)
	wg.Wait()
	require.NoError(t, err)
	require.Equal(t, len(commitInfos.CommitInfo), 1)
	require.Equal(t, newCommit, commitInfos.CommitInfo[0].Commit)
}