ridv/gorealis
1
0
Fork 0
Code Issues Pull requests Projects Releases 23 Packages Wiki Activity Actions

Upgrading gorealis v1 to Thrift 0.12.0 code generation. End to end tests cleanup (#96)

Browse source 
* Ported all code from Thrift 0.9.3 to Thrift 0.12.0 while backporting some fixes from gorealis v2

* Removing git.apache.org dependency from Vendor folder as this dependency has migrated to github.

* Adding github.com thrift dependency back but now it points to github.com

* Removing unnecessary files from Thrift Vendor folder and adding them to .gitignore.

* Updating dep dependencies to include Thrift 0.12.0 from github.com

* Adding changelog.

* End to end tests: Adding coverage for killinstances.

*  End to end tests: Deleting instances after partition policy recovers them.

*  End to end tests: Adding more coverage to the realis API.

*  End to end tests: Allowing arguments to be passed to runTestMac so that '-run <test name>' can be passed in.

*  End to end tests: Reducing the resources used by CreateJob test.

*  End to end tests: Adding coverage for Pause and Resume update.

*   End to end tests: Removed checks for Aurora_OK response as that should always be handled by the error returned by the API. Changed names to be less verbose and repetitive.

*  End to end tests: Reducing watch time for instance running when creating service for reducing time it takes to run end to end test.
This commit is contained in:
Renan DelValle 2019-02-20 11:11:46 -08:00 • committed by GitHub
parent 2b7eb3a852
commit 79fa7ba16d
No known key found for this signature in database
GPG key ID: 4AEE18F83AFDEB23
2150 changed files with 32523 additions and 412691 deletions
Download patch file Download diff file Expand all files Collapse all files

250
realis_e2e_test.go
View file

@ -22,7 +22,7 @@ import (
"testing"
"time"
"git.apache.org/thrift.git/lib/go/thrift"
"github.com/apache/thrift/lib/go/thrift"
realis "github.com/paypal/gorealis"
"github.com/paypal/gorealis/gen-go/apache/aurora"
"github.com/paypal/gorealis/response"
@ -188,48 +188,85 @@ func TestRealisClient_CreateJob_Thermos(t *testing.T) {
Name("create_thermos_job_test").
ExecutorName(aurora.AURORA_EXECUTOR_NAME).
ExecutorData(string(thermosPayload)).
CPU(.5).
RAM(64).
Disk(100).
CPU(.1).
RAM(16).
Disk(50).
IsService(true).
InstanceCount(2).
AddPorts(1)
resp, err := r.CreateJob(job)
_, err := r.CreateJob(job)
assert.NoError(t, err)
assert.Equal(t, aurora.ResponseCode_OK, resp.ResponseCode)
// Test Instances Monitor
success, err := monitor.Instances(job.JobKey(), job.GetInstanceCount(), 1, 50)
assert.True(t, success)
assert.NoError(t, err)
//Fetch all Jobs
// Fetch all Jobs
_, result, err := r.GetJobs(role)
fmt.Printf("GetJobs length: %+v \n", len(result.Configs))
assert.Len(t, result.Configs, 1)
assert.NoError(t, err)
// Test asking the scheduler to perform a Snpshot
t.Run("TestRealisClient_Snapshot", func(t *testing.T) {
t.Run("Snapshot", func(t *testing.T) {
err := r.Snapshot()
assert.NoError(t, err)
})
// Test asking the scheduler to backup a Snapshot
t.Run("TestRealisClient_PerformBackup", func(t *testing.T) {
t.Run("PerformBackup", func(t *testing.T) {
err := r.PerformBackup()
assert.NoError(t, err)
})
// Tasks must exist for it to, be killed
t.Run("TestRealisClient_KillJob_Thermos", func(t *testing.T) {
resp, err := r.KillJob(job.JobKey())
t.Run("GetTaskStatus", func(t *testing.T) {
status, err := r.GetTaskStatus(&aurora.TaskQuery{
JobKeys: []*aurora.JobKey{job.JobKey()},
Statuses: []aurora.ScheduleStatus{aurora.ScheduleStatus_RUNNING}})
assert.NoError(t, err)
assert.NotNil(t, status)
assert.Len(t, status, 2)
assert.Equal(t, aurora.ResponseCode_OK, resp.ResponseCode)
// TODO: Assert that assigned task matches the configuration of the task scheduled
})
t.Run("AddInstances", func(t *testing.T) {
_, err := r.AddInstances(aurora.InstanceKey{JobKey: job.JobKey(), InstanceId: 0}, 2)
assert.NoError(t, err)
success, err := monitor.Instances(job.JobKey(), 4, 1, 50)
assert.True(t, success)
assert.NoError(t, err)
})
t.Run("KillInstances", func(t *testing.T) {
_, err := r.KillInstances(job.JobKey(), 2, 3)
assert.NoError(t, err)
success, err := monitor.Instances(job.JobKey(), 2, 1, 50)
assert.True(t, success)
assert.NoError(t, err)
})
t.Run("RestartInstances", func(t *testing.T) {
_, err := r.RestartInstances(job.JobKey(), 0)
assert.NoError(t, err)
success, err := monitor.Instances(job.JobKey(), 2, 1, 50)
assert.True(t, success)
assert.NoError(t, err)
})
t.Run("RestartJobs", func(t *testing.T) {
_, err := r.RestartJob(job.JobKey())
assert.NoError(t, err)
success, err := monitor.Instances(job.JobKey(), 2, 1, 50)
assert.True(t, success)
assert.NoError(t, err)
})
// Tasks must exist for it to, be killed
t.Run("KillJob", func(t *testing.T) {
_, err := r.KillJob(job.JobKey())
assert.NoError(t, err)
success, err := monitor.Instances(job.JobKey(), 0, 1, 50)
assert.True(t, success)
assert.NoError(t, err)
@ -312,15 +349,14 @@ func TestRealisClient_CreateService_WithPulse_Thermos(t *testing.T) {
AddLabel("currentTime", time.Now().String())
settings := realis.NewUpdateSettings()
settings.BlockIfNoPulsesAfterMs = thrift.Int32Ptr(30)
settings.BlockIfNoPulsesAfterMs = thrift.Int32Ptr(500)
settings.UpdateGroupSize = 1
settings.MinWaitInInstanceRunningMs = 5000
settings.WaitForBatchCompletion = true
job.InstanceCount(2)
resp, result, err := r.CreateService(job, settings)
fmt.Println(result.String())
_, result, err := r.CreateService(job, settings)
assert.NoError(t, err)
assert.Equal(t, aurora.ResponseCode_OK, resp.ResponseCode)
updateQ := aurora.JobUpdateQuery{
Key: result.GetKey(),
@ -329,7 +365,7 @@ func TestRealisClient_CreateService_WithPulse_Thermos(t *testing.T) {
var updateDetails []*aurora.JobUpdateDetails
ticker := time.NewTicker(time.Second * 3)
ticker := time.NewTicker(time.Second * 1)
timer := time.NewTimer(time.Minute * 6)
defer ticker.Stop()
defer timer.Stop()
@ -339,35 +375,28 @@ pulseLoop:
select {
case <-ticker.C:
fmt.Println("sending PulseJobUpdate....")
resp, err = r.PulseJobUpdate(result.GetKey())
require.NotNil(t, resp, "received a nil response from Aurora")
assert.Nil(t, err)
_, err = r.PulseJobUpdate(result.GetKey())
assert.Nil(t, err, "unable to pulse job update")
respDetail, err := r.JobUpdateDetails(updateQ)
assert.Nil(t, err)
updateDetails = response.JobUpdateDetails(respDetail)
if len(updateDetails) == 0 {
fmt.Println("No update found")
assert.NotEqual(t, len(updateDetails), 0)
}
status := updateDetails[0].Update.Summary.State.Status
assert.Len(t, updateDetails, 1, "No update found")
if _, ok := aurora.ACTIVE_JOB_UPDATE_STATES[status]; !ok {
status := updateDetails[0].Update.Summary.State.Status
if _, ok := realis.ActiveJobUpdateStates[status]; !ok {
// Rolled forward is the only state in which an update has been successfully updated
// if we encounter an inactive state and it is not at rolled forward, update failed
if status == aurora.JobUpdateStatus_ROLLED_FORWARD {
fmt.Println("Update succeded")
fmt.Println("update succeed")
break pulseLoop
} else {
fmt.Println("Update failed")
fmt.Println("update failed")
break pulseLoop
}
}
fmt.Println("Polling, update still active...")
case <-timer.C:
_, err := r.AbortJobUpdate(*updateDetails[0].GetUpdate().GetSummary().GetKey(), "")
assert.NoError(t, err)
@ -376,9 +405,8 @@ pulseLoop:
}
}
resp, err = r.KillJob(job.JobKey())
_, err = r.KillJob(job.JobKey())
assert.NoError(t, err)
assert.Equal(t, aurora.ResponseCode_OK, resp.ResponseCode)
}
// Test configuring an executor that doesn't exist for CreateJob API
@ -399,12 +427,23 @@ func TestRealisClient_CreateService(t *testing.T) {
settings := realis.NewUpdateSettings()
settings.UpdateGroupSize = 2
settings.MinWaitInInstanceRunningMs = 5000
job.InstanceCount(3)
resp, result, err := r.CreateService(job, settings)
_, result, err := r.CreateService(job, settings)
assert.NoError(t, err)
assert.NotNil(t, result)
assert.Equal(t, aurora.ResponseCode_OK, resp.GetResponseCode())
// Test asking the scheduler to backup a Snapshot
t.Run("PauseJobUpdate", func(t *testing.T) {
_, err = r.PauseJobUpdate(result.GetKey(), "")
assert.NoError(t, err)
})
t.Run("ResumeJobUpdate", func(t *testing.T) {
_, err = r.ResumeJobUpdate(result.GetKey(), "")
assert.NoError(t, err)
})
var ok bool
var mErr error
@ -412,9 +451,7 @@ func TestRealisClient_CreateService(t *testing.T) {
if ok, mErr = monitor.JobUpdate(*result.GetKey(), 5, 240); !ok || mErr != nil {
// Update may already be in a terminal state so don't check for error
_, err := r.AbortJobUpdate(*result.GetKey(), "Monitor timed out.")
_, err = r.KillJob(job.JobKey())
assert.NoError(t, err)
}
@ -423,7 +460,6 @@ func TestRealisClient_CreateService(t *testing.T) {
// Kill task test task after confirming it came up fine
_, err = r.KillJob(job.JobKey())
assert.NoError(t, err)
}
@ -444,8 +480,8 @@ func TestRealisClient_CreateService_ExecutorDoesNotExist(t *testing.T) {
settings := realis.NewUpdateSettings()
job.InstanceCount(3)
resp, result, err := r.CreateService(job, settings)
resp, result, err := r.CreateService(job, settings)
assert.Error(t, err)
assert.Nil(t, result)
assert.Equal(t, aurora.ResponseCode_INVALID_REQUEST, resp.GetResponseCode())
@ -471,41 +507,24 @@ func TestRealisClient_ScheduleCronJob_Thermos(t *testing.T) {
CronSchedule("* * * * *").
IsService(false)
resp, err := r.ScheduleCronJob(job)
if err != nil {
fmt.Println(err)
os.Exit(1)
}
assert.Equal(t, aurora.ResponseCode_OK, resp.ResponseCode)
t.Run("TestRealisClient_StartCronJob_Thermos", func(t *testing.T) {
start := time.Now()
resp, err := r.StartCronJob(job.JobKey())
end := time.Now()
_, err = r.ScheduleCronJob(job)
assert.NoError(t, err)
t.Run("Start", func(t *testing.T) {
_, err := r.StartCronJob(job.JobKey())
assert.NoError(t, err)
assert.Equal(t, aurora.ResponseCode_OK, resp.ResponseCode)
fmt.Printf("Schedule cron call took %d ns\n", (end.UnixNano() - start.UnixNano()))
})
t.Run("TestRealisClient_DeschedulerCronJob_Thermos", func(t *testing.T) {
start := time.Now()
resp, err := r.DescheduleCronJob(job.JobKey())
end := time.Now()
t.Run("Deschedule", func(t *testing.T) {
_, err := r.DescheduleCronJob(job.JobKey())
assert.NoError(t, err)
assert.Equal(t, aurora.ResponseCode_OK, resp.ResponseCode)
fmt.Printf("Deschedule cron call took %d ns\n", (end.UnixNano() - start.UnixNano()))
})
}
func TestRealisClient_StartMaintenance(t *testing.T) {
hosts := []string{"localhost"}
_, _, err := r.StartMaintenance(hosts...)
if err != nil {
fmt.Printf("error: %+v\n", err.Error())
os.Exit(1)
}
assert.NoError(t, err, "unable to start maintenance")
// Monitor change to DRAINING and DRAINED mode
hostResults, err := monitor.HostMaintenance(
@ -530,22 +549,24 @@ func TestRealisClient_StartMaintenance(t *testing.T) {
func TestRealisClient_DrainHosts(t *testing.T) {
hosts := []string{"localhost"}
_, _, err := r.DrainHosts(hosts...)
if err != nil {
fmt.Printf("error: %+v\n", err.Error())
os.Exit(1)
}
// Monitor change to DRAINING and DRAINED mode
hostResults, err := monitor.HostMaintenance(
hosts,
[]aurora.MaintenanceMode{aurora.MaintenanceMode_DRAINED, aurora.MaintenanceMode_DRAINING},
1,
50)
assert.Equal(t, map[string]bool{"localhost": true}, hostResults)
assert.NoError(t, err)
t.Run("DrainHosts", func(t *testing.T) {
_, _, err := r.DrainHosts(hosts...)
assert.NoError(t, err, "unable to drain host")
})
t.Run("TestRealisClient_MonitorNontransitioned", func(t *testing.T) {
t.Run("MonitorTransitionToDrained", func(t *testing.T) {
// Monitor change to DRAINING and DRAINED mode
hostResults, err := monitor.HostMaintenance(
hosts,
[]aurora.MaintenanceMode{aurora.MaintenanceMode_DRAINED, aurora.MaintenanceMode_DRAINING},
1,
50)
assert.Equal(t, map[string]bool{"localhost": true}, hostResults)
assert.NoError(t, err)
})
t.Run("MonitorNonExistentHost", func(t *testing.T) {
// Monitor change to DRAINING and DRAINED mode
hostResults, err := monitor.HostMaintenance(
append(hosts, "IMAGINARY_HOST"),
@ -558,7 +579,7 @@ func TestRealisClient_DrainHosts(t *testing.T) {
assert.Equal(t, map[string]bool{"localhost": true, "IMAGINARY_HOST": false}, hostResults)
})
t.Run("TestRealisClient_EndMaintenance", func(t *testing.T) {
t.Run("EndMaintenance", func(t *testing.T) {
_, _, err := r.EndMaintenance(hosts...)
assert.NoError(t, err)
@ -578,10 +599,7 @@ func TestRealisClient_SLADrainHosts(t *testing.T) {
policy := aurora.SlaPolicy{PercentageSlaPolicy: &aurora.PercentageSlaPolicy{Percentage: 50.0}}
_, err := r.SLADrainHosts(&policy, 30, hosts...)
if err != nil {
fmt.Printf("error: %+v\n", err.Error())
os.Exit(1)
}
assert.NoError(t, err, "unable to drain host with SLA policy")
// Monitor change to DRAINING and DRAINED mode
hostResults, err := monitor.HostMaintenance(
@ -619,30 +637,25 @@ func TestRealisClient_SessionThreadSafety(t *testing.T) {
Disk(10).
InstanceCount(1000) // Impossible amount to go live in any sane machine
resp, err := r.CreateJob(job)
_, err := r.CreateJob(job)
assert.NoError(t, err)
assert.Equal(t, aurora.ResponseCode_OK, resp.ResponseCode)
wg := sync.WaitGroup{}
for i := 0; i < 20; i++ {
wg.Add(1)
threadCount := 20
wg.Add(threadCount)
for i := 0; i < threadCount; i++ {
// Launch multiple monitors that will poll every second
go func() {
defer wg.Done()
// Test Schedule status monitor for terminal state and timing out after 30 seconds
success, err := monitor.ScheduleStatus(job.JobKey(), job.GetInstanceCount(), aurora.LIVE_STATES, 1, 30)
success, err := monitor.ScheduleStatus(job.JobKey(), job.GetInstanceCount(), realis.LiveStates, 1, 30)
assert.False(t, success)
assert.Error(t, err)
resp, err := r.KillJob(job.JobKey())
_, err = r.KillJob(job.JobKey())
assert.NoError(t, err)
assert.Equal(t, aurora.ResponseCode_OK, resp.ResponseCode)
}()
}
@ -650,14 +663,20 @@ func TestRealisClient_SessionThreadSafety(t *testing.T) {
}
// Test setting and getting the quota
func TestRealisClient_SetQuota(t *testing.T) {
var cpu = 3.5
var ram int64 = 20480
var disk int64 = 10240
resp, err := r.SetQuota("vagrant", &cpu, &ram, &disk)
assert.NoError(t, err)
assert.Equal(t, aurora.ResponseCode_OK, resp.ResponseCode)
t.Run("TestRealisClient_GetQuota", func(t *testing.T) {
func TestRealisClient_Quota(t *testing.T) {
var resp *aurora.Response
var err error
cpu := 3.5
ram := int64(20480)
disk := int64(10240)
t.Run("Set", func(t *testing.T) {
resp, err = r.SetQuota("vagrant", &cpu, &ram, &disk)
assert.NoError(t, err)
})
t.Run("Get", func(t *testing.T) {
// Test GetQuota based on previously set values
var result *aurora.GetQuotaResult_
resp, err = r.GetQuota("vagrant")
@ -665,8 +684,8 @@ func TestRealisClient_SetQuota(t *testing.T) {
result = resp.GetResult_().GetQuotaResult_
}
assert.NoError(t, err)
assert.Equal(t, aurora.ResponseCode_OK, resp.ResponseCode)
for res := range result.Quota.GetResources() {
for _, res := range result.Quota.GetResources() {
switch true {
case res.DiskMb != nil:
assert.Equal(t, disk, *res.DiskMb)
@ -679,7 +698,6 @@ func TestRealisClient_SetQuota(t *testing.T) {
break
}
}
fmt.Print("GetQuota Result", result.String())
})
}
@ -694,15 +712,13 @@ func TestRealisClient_ForceExplicitTaskReconciliation(t *testing.T) {
assert.NoError(t, err)
// Custom batch value
var batchSize int32 = 32
err = r.ForceExplicitTaskReconciliation(&batchSize)
err = r.ForceExplicitTaskReconciliation(thrift.Int32Ptr(32))
assert.NoError(t, err)
}
func TestRealisClient_PartitionPolicy(t *testing.T) {
role := "vagrant"
var partitionDelay int64 = 30
job := realis.NewJob().
Environment("prod").
Role(role).
@ -714,14 +730,13 @@ func TestRealisClient_PartitionPolicy(t *testing.T) {
Disk(100).
IsService(true).
InstanceCount(2).
PartitionPolicy(&aurora.PartitionPolicy{Reschedule: true, DelaySecs: &partitionDelay})
PartitionPolicy(&aurora.PartitionPolicy{Reschedule: true, DelaySecs: thrift.Int64Ptr(30)})
settings := realis.NewUpdateSettings()
settings.UpdateGroupSize = 2
resp, result, err := r.CreateService(job, settings)
assert.NoError(t, err)
assert.Equal(t, aurora.ResponseCode_OK, resp.ResponseCode)
_, result, err := r.CreateService(job, settings)
assert.NoError(t, err)
var ok bool
var mErr error
@ -729,10 +744,9 @@ func TestRealisClient_PartitionPolicy(t *testing.T) {
if ok, mErr = monitor.JobUpdate(*result.GetKey(), 5, 180); !ok || mErr != nil {
// Update may already be in a terminal state so don't check for error
_, err := r.AbortJobUpdate(*result.GetKey(), "Monitor timed out.")
_, err = r.KillJob(job.JobKey())
assert.NoError(t, err)
}
// Clean up after finishing test
_, err = r.KillJob(job.JobKey())
}