Introduce system artifact manager cleanup job (#16879)

Signed-off-by: prahaladdarkin <prahaladd@vmware.com>
This commit is contained in:
prahaladdarkin 2022-06-09 11:46:50 +05:30 committed by GitHub
parent b8a71ac348
commit 4d062c33d1
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
10 changed files with 582 additions and 3 deletions

View File

@ -0,0 +1,32 @@
package systemartifact
import (
"context"
"github.com/goharbor/harbor/src/jobservice/logger"
"github.com/goharbor/harbor/src/lib/log"
"github.com/goharbor/harbor/src/pkg/scheduler"
"github.com/goharbor/harbor/src/pkg/task"
)
const (
SystemArtifactCleanupCallback = "SYSTEM_ARTIFACT_CLEANUP"
)
var (
cleanupController = Ctl
)
func init() {
if err := scheduler.RegisterCallbackFunc(SystemArtifactCleanupCallback, cleanupCallBack); err != nil {
log.Fatalf("failed to register the callback for the system artifact cleanup schedule, error %v", err)
}
}
func cleanupCallBack(ctx context.Context, param string) error {
err := cleanupController.Start(ctx, true, task.ExecutionTriggerSchedule)
if err != nil {
logger.Errorf("System artifact cleanup job encountered errors: %v", err)
}
return err
}

View File

@ -0,0 +1,56 @@
package systemartifact
import (
"context"
"errors"
"github.com/goharbor/harbor/src/pkg/task"
"github.com/goharbor/harbor/src/testing/controller/systemartifact"
"github.com/goharbor/harbor/src/testing/mock"
testifymock "github.com/stretchr/testify/mock"
"github.com/stretchr/testify/suite"
"testing"
)
type CallbackTestSuite struct {
suite.Suite
cleanupController *systemartifact.Controller
}
func (suite *CallbackTestSuite) SetupSuite() {
suite.cleanupController = &systemartifact.Controller{}
cleanupController = suite.cleanupController
}
func (suite *CallbackTestSuite) TestCleanupCallbackSuccess() {
{
ctx := context.TODO()
suite.cleanupController.On("Start", mock.Anything, mock.Anything, mock.Anything).Return(nil)
triggerScheduleMatcher := testifymock.MatchedBy(func(trigger string) bool {
return trigger == task.ExecutionTriggerSchedule
})
err := cleanupCallBack(ctx, "")
suite.NoErrorf(err, "Unexpected error : %v", err)
suite.cleanupController.AssertCalled(suite.T(), "Start", mock.Anything, true, triggerScheduleMatcher)
}
{
suite.cleanupController = nil
suite.cleanupController = &systemartifact.Controller{}
cleanupController = suite.cleanupController
}
{
ctx := context.TODO()
suite.cleanupController.On("Start", mock.Anything, mock.Anything, mock.Anything).Return(errors.New("test error"))
triggerScheduleMatcher := testifymock.MatchedBy(func(trigger string) bool {
return trigger == task.ExecutionTriggerSchedule
})
err := cleanupCallBack(ctx, "")
suite.Error(err)
suite.cleanupController.AssertCalled(suite.T(), "Start", mock.Anything, true, triggerScheduleMatcher)
}
}
func TestCallbackTestSuite(t *testing.T) {
suite.Run(t, &CallbackTestSuite{})
}

View File

@ -0,0 +1,156 @@
package systemartifact
import (
"context"
"github.com/goharbor/harbor/src/jobservice/job"
"github.com/goharbor/harbor/src/jobservice/logger"
"github.com/goharbor/harbor/src/lib/log"
"github.com/goharbor/harbor/src/lib/orm"
"github.com/goharbor/harbor/src/lib/q"
"github.com/goharbor/harbor/src/lib/retry"
"github.com/goharbor/harbor/src/pkg/scheduler"
"github.com/goharbor/harbor/src/pkg/systemartifact"
"github.com/goharbor/harbor/src/pkg/task"
"time"
)
const (
VendorTypeSystemArtifactCleanup = "SYSTEM_ARTIFACT_CLEANUP"
cronTypeDaily = "Daily"
cronSpec = "0 0 0 * * *"
)
var (
sched = scheduler.Sched
)
func init() {
task.SetExecutionSweeperCount(VendorTypeSystemArtifactCleanup, 50)
}
var Ctl = NewController()
type Controller interface {
Start(ctx context.Context, async bool, trigger string) error
}
func NewController() Controller {
return &controller{
execMgr: task.ExecMgr,
taskMgr: task.Mgr,
systemArtifactMgr: systemartifact.Mgr,
makeCtx: orm.Context,
}
}
type controller struct {
execMgr task.ExecutionManager
taskMgr task.Manager
systemArtifactMgr systemartifact.Manager
makeCtx func() context.Context
}
func (c *controller) Start(ctx context.Context, async bool, trigger string) error {
execId, err := c.execMgr.Create(ctx, VendorTypeSystemArtifactCleanup, 0, trigger)
if err != nil {
return err
}
// cleanup job would always be scheduled in async mode in production
// allowing for sync mode execution only for test mode purposes
// if there are any trigger settings then pass them to the cleanup manager first
jobParams := job.Parameters{}
if !async {
err := c.createCleanupTask(ctx, jobParams, execId)
if err != nil {
log.Errorf("failed to create system artifact clean-up task: %v", err)
return err
}
logger.Info("Created job for scan data export successfully")
return nil
}
go func(ctx context.Context) {
err := retry.Retry(func() error {
_, err := c.execMgr.Get(ctx, execId)
return err
})
if err != nil {
log.Errorf("failed to get the execution %d for the export data cleanup job", execId)
return
}
err = c.createCleanupTask(ctx, jobParams, execId)
if err != nil {
logger.Errorf("Encountered error in scan data artifact cleanup : %v", err)
return
}
}(c.makeCtx())
return nil
}
func (c *controller) createCleanupTask(ctx context.Context, jobParams job.Parameters, execId int64) error {
j := &task.Job{
Name: job.SystemArtifactCleanup,
Metadata: &job.Metadata{
JobKind: job.KindGeneric,
},
Parameters: jobParams,
}
_, err := c.taskMgr.Create(ctx, execId, j)
if err != nil {
logger.Errorf("Unable to create a scan data export job in clean-up mode : %v", err)
c.markError(ctx, execId, err)
return err
}
return nil
}
func (c *controller) markError(ctx context.Context, executionID int64, err error) {
// try to stop the execution first in case that some tasks are already created
if err := c.execMgr.StopAndWait(ctx, executionID, 10*time.Second); err != nil {
logger.Errorf("failed to stop the execution %d: %v", executionID, err)
}
if err := c.execMgr.MarkError(ctx, executionID, err.Error()); err != nil {
logger.Errorf("failed to mark error for the execution %d: %v", executionID, err)
}
}
// ScheduleCleanupTask schedules a system artifact cleanup task
func ScheduleCleanupTask(ctx context.Context) {
scheduleSystemArtifactCleanJob(ctx)
}
func scheduleSystemArtifactCleanJob(ctx context.Context) {
schedule, err := getSystemArtifactCleanupSchedule(ctx)
if err != nil {
return
}
if schedule != nil {
logger.Debugf(" Export data cleanup job already scheduled with ID : %v.", schedule.ID)
return
}
scheduleId, err := sched.Schedule(ctx, VendorTypeSystemArtifactCleanup, 0, cronTypeDaily, cronSpec, SystemArtifactCleanupCallback, nil, nil)
if err != nil {
log.Errorf("Encountered error when scheduling scan data export cleanup job : %v", err)
return
}
log.Infof("Scheduled scan data export cleanup job with ID : %v", scheduleId)
}
func getSystemArtifactCleanupSchedule(ctx context.Context) (*scheduler.Schedule, error) {
query := q.New(map[string]interface{}{"vendor_type": VendorTypeSystemArtifactCleanup})
schedules, err := sched.ListSchedules(ctx, query)
if err != nil {
logger.Errorf("Unable to check if export data cleanup job is already scheduled : %v", err)
return nil, err
}
if len(schedules) > 0 {
logger.Infof("Found export data cleanup job with schedule id : %v", schedules[0].ID)
return schedules[0], nil
}
return nil, nil
}

View File

@ -0,0 +1,217 @@
package systemartifact
import (
"context"
"github.com/goharbor/harbor/src/lib/orm"
scheduler2 "github.com/goharbor/harbor/src/pkg/scheduler"
"github.com/goharbor/harbor/src/pkg/task"
ormtesting "github.com/goharbor/harbor/src/testing/lib/orm"
"github.com/goharbor/harbor/src/testing/mock"
"github.com/goharbor/harbor/src/testing/pkg/scheduler"
"github.com/goharbor/harbor/src/testing/pkg/systemartifact"
testingTask "github.com/goharbor/harbor/src/testing/pkg/task"
"github.com/pkg/errors"
testifymock "github.com/stretchr/testify/mock"
"github.com/stretchr/testify/suite"
"testing"
)
type SystemArtifactCleanupTestSuite struct {
suite.Suite
execMgr *testingTask.ExecutionManager
taskMgr *testingTask.Manager
cleanupMgr *systemartifact.Manager
ctl *controller
sched *scheduler.Scheduler
}
func (suite *SystemArtifactCleanupTestSuite) SetupSuite() {
}
func (suite *SystemArtifactCleanupTestSuite) TestStartCleanup() {
suite.taskMgr = &testingTask.Manager{}
suite.execMgr = &testingTask.ExecutionManager{}
suite.cleanupMgr = &systemartifact.Manager{}
suite.ctl = &controller{
execMgr: suite.execMgr,
taskMgr: suite.taskMgr,
systemArtifactMgr: suite.cleanupMgr,
makeCtx: func() context.Context { return orm.NewContext(nil, &ormtesting.FakeOrmer{}) },
}
{
ctx := context.TODO()
executionID := int64(1)
taskId := int64(1)
suite.execMgr.On("Create", ctx, "SYSTEM_ARTIFACT_CLEANUP", int64(0), "SCHEDULE").Return(executionID, nil).Once()
suite.taskMgr.On("Create", ctx, executionID, mock.Anything).Return(taskId, nil).Once()
suite.execMgr.On("MarkDone", ctx, executionID, mock.Anything).Return(nil).Once()
err := suite.ctl.Start(ctx, false, "SCHEDULE")
suite.NoError(err)
jobMatcher := testifymock.MatchedBy(func(j *task.Job) bool {
return "SYSTEM_ARTIFACT_CLEANUP" == j.Name
})
suite.taskMgr.AssertCalled(suite.T(), "Create", ctx, executionID, jobMatcher)
}
}
func (suite *SystemArtifactCleanupTestSuite) TestStartCleanupErrorDuringCreate() {
suite.taskMgr = &testingTask.Manager{}
suite.execMgr = &testingTask.ExecutionManager{}
suite.cleanupMgr = &systemartifact.Manager{}
suite.ctl = &controller{
execMgr: suite.execMgr,
taskMgr: suite.taskMgr,
systemArtifactMgr: suite.cleanupMgr,
makeCtx: func() context.Context { return orm.NewContext(nil, &ormtesting.FakeOrmer{}) },
}
{
ctx := context.TODO()
executionID := int64(1)
suite.execMgr.On(
"Create", ctx, "SYSTEM_ARTIFACT_CLEANUP", int64(0), "SCHEDULE",
).Return(int64(0), errors.New("test error")).Once()
suite.execMgr.On("MarkDone", ctx, executionID, mock.Anything).Return(nil).Once()
err := suite.ctl.Start(ctx, false, "SCHEDULE")
suite.Error(err)
}
}
func (suite *SystemArtifactCleanupTestSuite) TestStartCleanupErrorDuringTaskCreate() {
suite.taskMgr = &testingTask.Manager{}
suite.execMgr = &testingTask.ExecutionManager{}
suite.cleanupMgr = &systemartifact.Manager{}
suite.ctl = &controller{
execMgr: suite.execMgr,
taskMgr: suite.taskMgr,
systemArtifactMgr: suite.cleanupMgr,
makeCtx: func() context.Context { return orm.NewContext(nil, &ormtesting.FakeOrmer{}) },
}
{
ctx := context.TODO()
executionID := int64(1)
taskId := int64(0)
suite.execMgr.On(
"Create", ctx, "SYSTEM_ARTIFACT_CLEANUP", int64(0), "SCHEDULE",
).Return(executionID, nil).Once()
suite.taskMgr.On("Create", ctx, executionID, mock.Anything).Return(taskId, errors.New("test error")).Once()
suite.execMgr.On("MarkError", ctx, executionID, mock.Anything).Return(nil).Once()
suite.execMgr.On("StopAndWait", ctx, executionID, mock.Anything).Return(nil).Once()
err := suite.ctl.Start(ctx, false, "SCHEDULE")
suite.Error(err)
}
}
func (suite *SystemArtifactCleanupTestSuite) TestScheduleCleanupJobNoPreviousSchedule() {
suite.taskMgr = &testingTask.Manager{}
suite.execMgr = &testingTask.ExecutionManager{}
suite.cleanupMgr = &systemartifact.Manager{}
suite.sched = &scheduler.Scheduler{}
suite.ctl = &controller{
execMgr: suite.execMgr,
taskMgr: suite.taskMgr,
systemArtifactMgr: suite.cleanupMgr,
makeCtx: func() context.Context { return orm.NewContext(nil, &ormtesting.FakeOrmer{}) },
}
var extraAttrs map[string]interface{}
suite.sched.On("Schedule", mock.Anything,
VendorTypeSystemArtifactCleanup, int64(0), cronTypeDaily, cronSpec, SystemArtifactCleanupCallback, nil, extraAttrs).Return(int64(1), nil)
suite.sched.On("ListSchedules", mock.Anything, mock.Anything).Return(make([]*scheduler2.Schedule, 0), nil)
sched = suite.sched
ctx := context.TODO()
ScheduleCleanupTask(ctx)
suite.sched.AssertCalled(suite.T(), "Schedule", mock.Anything,
VendorTypeSystemArtifactCleanup, int64(0), cronTypeDaily, cronSpec, SystemArtifactCleanupCallback, nil, extraAttrs)
}
func (suite *SystemArtifactCleanupTestSuite) TestScheduleCleanupJobPreviousSchedule() {
suite.taskMgr = &testingTask.Manager{}
suite.execMgr = &testingTask.ExecutionManager{}
suite.cleanupMgr = &systemartifact.Manager{}
suite.sched = &scheduler.Scheduler{}
suite.ctl = &controller{
execMgr: suite.execMgr,
taskMgr: suite.taskMgr,
systemArtifactMgr: suite.cleanupMgr,
makeCtx: func() context.Context { return orm.NewContext(nil, &ormtesting.FakeOrmer{}) },
}
var extraAttrs map[string]interface{}
suite.sched.On("Schedule", mock.Anything,
VendorTypeSystemArtifactCleanup, int64(0), cronTypeDaily, cronSpec, SystemArtifactCleanupCallback, nil, extraAttrs).Return(int64(1), nil)
existingSchedule := scheduler2.Schedule{ID: int64(10)}
suite.sched.On("ListSchedules", mock.Anything, mock.Anything).Return([]*scheduler2.Schedule{&existingSchedule}, nil)
sched = suite.sched
ctx := context.TODO()
ScheduleCleanupTask(ctx)
suite.sched.AssertNotCalled(suite.T(), "Schedule", mock.Anything,
VendorTypeSystemArtifactCleanup, int64(0), cronTypeDaily, cronSpec, SystemArtifactCleanupCallback, nil, extraAttrs)
}
func (suite *SystemArtifactCleanupTestSuite) TestScheduleCleanupJobPreviousScheduleError() {
suite.taskMgr = &testingTask.Manager{}
suite.execMgr = &testingTask.ExecutionManager{}
suite.cleanupMgr = &systemartifact.Manager{}
suite.sched = &scheduler.Scheduler{}
suite.ctl = &controller{
execMgr: suite.execMgr,
taskMgr: suite.taskMgr,
systemArtifactMgr: suite.cleanupMgr,
makeCtx: func() context.Context { return orm.NewContext(nil, &ormtesting.FakeOrmer{}) },
}
suite.sched.On("Schedule", mock.Anything,
VendorTypeSystemArtifactCleanup, int64(0), cronTypeDaily, cronSpec, SystemArtifactCleanupCallback, nil, mock.Anything).Return(int64(1), nil)
suite.sched.On("ListSchedules", mock.Anything, mock.Anything).Return(nil, errors.New("test error"))
sched = suite.sched
ctx := context.TODO()
ScheduleCleanupTask(ctx)
extraAttributesMatcher := testifymock.MatchedBy(func(attrs map[string]interface{}) bool {
return len(attrs) == 0
})
suite.sched.AssertNotCalled(suite.T(), "Schedule", mock.Anything,
VendorTypeSystemArtifactCleanup, int64(0), cronTypeDaily, cronSpec, SystemArtifactCleanupCallback, nil, extraAttributesMatcher)
}
func (suite *SystemArtifactCleanupTestSuite) TearDownSuite() {
suite.execMgr = nil
suite.taskMgr = nil
suite.cleanupMgr = nil
suite.ctl = nil
suite.sched = nil
}
func TestScanDataExportExecutionTestSuite(t *testing.T) {
suite.Run(t, &SystemArtifactCleanupTestSuite{})
}

View File

@ -25,6 +25,8 @@ import (
"syscall"
"time"
"github.com/goharbor/harbor/src/controller/systemartifact"
"github.com/beego/beego"
"github.com/goharbor/harbor/src/core/session"
@ -227,6 +229,7 @@ func main() {
if err != nil {
log.Warningf("oidc.FixEmptySubIss() errors out, error: %v", err)
}
systemartifact.ScheduleCleanupTask(ctx)
beego.RunWithMiddleWares("", middlewares.MiddleWares()...)
}

View File

@ -0,0 +1,45 @@
package systemartifact
import (
"github.com/goharbor/harbor/src/jobservice/job"
"github.com/goharbor/harbor/src/pkg/systemartifact"
)
type Cleanup struct {
sysArtifactManager systemartifact.Manager
}
func (c *Cleanup) MaxFails() uint {
return 1
}
func (c *Cleanup) MaxCurrency() uint {
return 1
}
func (c *Cleanup) ShouldRetry() bool {
return true
}
func (c *Cleanup) Validate(params job.Parameters) error {
return nil
}
func (c *Cleanup) Run(ctx job.Context, params job.Parameters) error {
logger := ctx.GetLogger()
logger.Infof("Running system data artifact cleanup job...")
c.init()
numRecordsDeleted, totalSizeReclaimed, err := c.sysArtifactManager.Cleanup(ctx.SystemContext())
if err != nil {
logger.Errorf("Error when executing system artifact cleanup job: %v", err)
return err
}
logger.Infof("Num System artifacts cleaned up: %d, Total space reclaimed: %d.", numRecordsDeleted, totalSizeReclaimed)
return nil
}
func (c *Cleanup) init() {
if c.sysArtifactManager == nil {
c.sysArtifactManager = systemartifact.NewManager()
}
}

View File

@ -0,0 +1,64 @@
package systemartifact
import (
"errors"
"github.com/goharbor/harbor/src/jobservice/job"
mockjobservice "github.com/goharbor/harbor/src/testing/jobservice"
"github.com/goharbor/harbor/src/testing/mock"
"github.com/goharbor/harbor/src/testing/pkg/systemartifact"
"github.com/stretchr/testify/suite"
"testing"
)
type SystemArtifactCleanupSuite struct {
suite.Suite
sysArtifactMgr *systemartifact.Manager
job *Cleanup
}
func (suite *SystemArtifactCleanupSuite) SetupTest() {
suite.sysArtifactMgr = &systemartifact.Manager{}
suite.job = &Cleanup{sysArtifactManager: suite.sysArtifactMgr}
}
func (suite *SystemArtifactCleanupSuite) TestRun() {
mock.OnAnything(suite.sysArtifactMgr, "Cleanup").Return(int64(100), int64(100), nil)
params := job.Parameters{}
ctx := &mockjobservice.MockJobContext{}
err := suite.job.Run(ctx, params)
suite.NoError(err)
// assert that job manager is invoked in this mode
suite.sysArtifactMgr.AssertCalled(suite.T(), "Cleanup", mock.Anything)
}
func (suite *SystemArtifactCleanupSuite) TestRunFailure() {
mock.OnAnything(suite.sysArtifactMgr, "Cleanup").Return(int64(0), int64(0), errors.New("test error"))
params := job.Parameters{}
ctx := &mockjobservice.MockJobContext{}
err := suite.job.Run(ctx, params)
suite.Error(err)
// assert that job manager is invoked in this mode
suite.sysArtifactMgr.AssertCalled(suite.T(), "Cleanup", mock.Anything)
}
func (suite *SystemArtifactCleanupSuite) TestMaxFails() {
suite.Equal(uint(1), suite.job.MaxFails())
}
func (suite *SystemArtifactCleanupSuite) TestMaxConcurrency() {
suite.Equal(uint(1), suite.job.MaxCurrency())
}
func (suite *SystemArtifactCleanupSuite) TestShouldRetry() {
suite.Equal(true, suite.job.ShouldRetry())
}
func (suite *SystemArtifactCleanupSuite) TestValidate() {
suite.NoError(suite.job.Validate(job.Parameters{}))
}
func TestSystemArtifactCleanupSuite(t *testing.T) {
suite.Run(t, &SystemArtifactCleanupSuite{})
}

View File

@ -36,4 +36,6 @@ const (
P2PPreheat = "P2P_PREHEAT"
// PurgeAudit : the name of purge audit job
PurgeAudit = "PURGE_AUDIT"
// SystemArtifactCleanup : the name of the SystemArtifact cleanup job
SystemArtifactCleanup = "SYSTEM_ARTIFACT_CLEANUP"
)

View File

@ -24,6 +24,8 @@ import (
"syscall"
"time"
"github.com/goharbor/harbor/src/jobservice/job/impl/systemartifact"
"github.com/goharbor/harbor/src/jobservice/api"
"github.com/goharbor/harbor/src/jobservice/common/utils"
"github.com/goharbor/harbor/src/jobservice/config"
@ -319,9 +321,10 @@ func (bs *Bootstrap) loadAndRunRedisWorkerPool(
// In v2.2 we migrate the scheduled replication, garbage collection and scan all to
// the scheduler mechanism, the following three jobs are kept for the legacy jobs
// and they can be removed after several releases
"IMAGE_REPLICATE": (*legacy.ReplicationScheduler)(nil),
"IMAGE_GC": (*legacy.GarbageCollectionScheduler)(nil),
"IMAGE_SCAN_ALL": (*legacy.ScanAllScheduler)(nil),
"IMAGE_REPLICATE": (*legacy.ReplicationScheduler)(nil),
"IMAGE_GC": (*legacy.GarbageCollectionScheduler)(nil),
"IMAGE_SCAN_ALL": (*legacy.ScanAllScheduler)(nil),
job.SystemArtifactCleanup: (*systemartifact.Cleanup)(nil),
}); err != nil {
// exit
return nil, err

View File

@ -30,3 +30,4 @@ package controller
//go:generate mockery --case snake --dir ../../controller/repository --name Controller --output ./repository --outpkg repository
//go:generate mockery --case snake --dir ../../controller/purge --name Controller --output ./purge --outpkg purge
//go:generate mockery --case snake --dir ../../controller/jobservice --name SchedulerController --output ./jobservice --outpkg jobservice
//go:generate mockery --case snake --dir ../../controller/systemartifact --name Controller --output ./systemartifact --outpkg systemartifact