Misc. fixes for tests (#16)

* Bumping up CI to go1.17 and enabling CI for PRs.

* Adding go.sum now that issues seem to have gone away.

* Bump up aurora to 0.25.0 and mesos to 1.9.0

* Fixing Mac tests. Adding extra time for killing thermos jobs.

* Reduce the thermos overhead for unit tests

Co-authored-by: lenhattan86 <lenhattan86@users.noreply.github.com>

.github/workflows/main.yml

@@ -1,6 +1,12 @@
 name: CI
 
-on: [push]
+on:
+  push:
+    branches:
+      - master
+  pull_request:
+    branches:
+      - master
 
 jobs:
   build:
@@ -12,7 +18,7 @@ jobs:
     - name: Setup Go for use with actions
       uses: actions/setup-go@v2
       with:
-        go-version: 1.15
+        go-version: 1.17
     - name: Install goimports
       run: go get golang.org/x/tools/cmd/goimports
     - name: Set env with list of directories in repo containin go code

.gitignore

@@ -41,6 +41,3 @@ _testmain.go
 # Example client build
 examples/client
 examples/jsonClient
-
-# Use checksum database
-go.sum

docker-compose.yml

@@ -14,7 +14,7 @@ services:
         ipv4_address: 192.168.33.2
 
   master:
-    image: rdelvalle/mesos-master:1.6.2
+    image: quay.io/aurorascheduler/mesos-master:1.9.0
     restart: on-failure
     ports:
     - "5050:5050"
@@ -32,7 +32,7 @@ services:
     - zk
 
   agent-one:
-    image: rdelvalle/mesos-agent:1.6.2
+    image: quay.io/aurorascheduler/mesos-agent:1.9.0
     pid: host
     restart: on-failure
     ports:
@@ -56,7 +56,7 @@ services:
     - zk
 
   aurora-one:
-    image: rdelvalle/aurora:0.22.0
+    image: quay.io/aurorascheduler/scheduler:0.25.0
     pid: host
     ports:
     - "8081:8081"
@@ -70,6 +70,7 @@ services:
         -shiro_realm_modules=INI_AUTHNZ
         -shiro_ini_path=/etc/aurora/security.ini
         -min_required_instances_for_sla_check=1
+        -thermos_executor_cpu=0.09
     volumes:
       - ./.aurora-config:/etc/aurora
     networks:

go.sum

@@ -0,0 +1,18 @@
+github.com/apache/thrift v0.14.0 h1:vqZ2DP42i8th2OsgCcYZkirtbzvpZEFx53LiWDJXIAs=
+github.com/apache/thrift v0.14.0/go.mod h1:cp2SuWMxlEZw2r+iP2GNCdIi4C1qmUzdZFSVb+bacwQ=
+github.com/davecgh/go-spew v1.1.0 h1:ZDRjVQ15GmhC3fiQ8ni8+OwkZQO4DARzQgrnXU1Liz8=
+github.com/davecgh/go-spew v1.1.0/go.mod h1:J7Y8YcW2NihsgmVo/mv3lAwl/skON4iLHjSsI+c5H38=
+github.com/pkg/errors v0.9.1 h1:FEBLx1zS214owpjy7qsBeixbURkuhQAwrK5UwLGTwt4=
+github.com/pkg/errors v0.9.1/go.mod h1:bwawxfHBFNV+L2hUp1rHADufV3IMtnDRdf1r5NINEl0=
+github.com/pmezard/go-difflib v1.0.0 h1:4DBwDE0NGyQoBHbLQYPwSUPoCMWR5BEzIk/f1lZbAQM=
+github.com/pmezard/go-difflib v1.0.0/go.mod h1:iKH77koFhYxTK1pcRnkKkqfTogsbg7gZNVY4sRDYZ/4=
+github.com/samuel/go-zookeeper v0.0.0-20171117190445-471cd4e61d7a h1:EYL2xz/Zdo0hyqdZMXR4lmT2O11jDLTPCEqIe/FR6W4=
+github.com/samuel/go-zookeeper v0.0.0-20171117190445-471cd4e61d7a/go.mod h1:gi+0XIa01GRL2eRQVjQkKGqKF3SF9vZR/HnPullcV2E=
+github.com/stretchr/objx v0.1.0 h1:4G4v2dO3VZwixGIRoQ5Lfboy6nUhCyYzaqnIAPPhYs4=
+github.com/stretchr/objx v0.1.0/go.mod h1:HFkY916IF+rwdDfMAkV7OtwuqBVzrE8GR6GFx+wExME=
+github.com/stretchr/testify v1.5.0 h1:DMOzIV76tmoDNE9pX6RSN0aDtCYeCg5VueieJaAo1uw=
+github.com/stretchr/testify v1.5.0/go.mod h1:5W2xD1RspED5o8YsWQXVCued0rvSQ+mT+I5cxcmMvtA=
+gopkg.in/check.v1 v0.0.0-20161208181325-20d25e280405 h1:yhCVgyC4o1eVCa2tZl7eS0r+SDo693bJlVdllGtEeKM=
+gopkg.in/check.v1 v0.0.0-20161208181325-20d25e280405/go.mod h1:Co6ibVJAznAaIkqp8huTwlJQCZ016jof/cbN4VW5Yz0=
+gopkg.in/yaml.v2 v2.2.2 h1:ZCJp+EgiOT7lHqUV2J862kp8Qj64Jo6az82+3Td9dZw=
+gopkg.in/yaml.v2 v2.2.2/go.mod h1:hI93XBmqTisBFMUTm0b8Fm+jr3Dg1NNxqwp+5A1VGuI=

jobUpdate.go

@@ -77,7 +77,7 @@ func (j *JobUpdate) BatchSize(size int32) *JobUpdate {
 
 // Minimum number of seconds a shard must remain in RUNNING state before considered a success.
 func (j *JobUpdate) WatchTime(timeout time.Duration) *JobUpdate {
-	j.request.Settings.MinWaitInInstanceRunningMs = int32(timeout.Seconds() * 1000)
+	j.request.Settings.MinWaitInInstanceRunningMs = int32(timeout.Milliseconds())
 	return j
 }
 

monitors.go

@@ -245,7 +245,7 @@ func (c *Client) MonitorHostMaintenance(hosts []string,
 	}
 }
 
-// AutoPaused monitor is a special monitor for auto pause enabled batch updates. This monitor ensures that the update
+// MonitorAutoPausedUpdate is a special monitor for auto pause enabled batch updates. This monitor ensures that the update
 // being monitored is capable of auto pausing and has auto pausing enabled. After verifying this information,
 // the monitor watches for the job to enter the ROLL_FORWARD_PAUSED state and calculates the current batch
 // the update is in using information from the update configuration.
@@ -294,8 +294,9 @@ func (c *Client) MonitorAutoPausedUpdate(key aurora.JobUpdateKey, interval, time
 		return -1, err
 	}
 
-	// Summary 0 is assumed to exist because MonitorJobUpdateQuery will return an error if there is Summaries
+	// Summary 0 is assumed to exist because MonitorJobUpdateQuery will return an error if there is no summaries
-	if summary[0].State.Status != aurora.JobUpdateStatus_ROLL_FORWARD_PAUSED {
+	if !(summary[0].State.Status == aurora.JobUpdateStatus_ROLL_FORWARD_PAUSED ||
+		summary[0].State.Status == aurora.JobUpdateStatus_ROLLED_FORWARD) {
 		return -1, errors.Errorf("update is in a terminal state %v", summary[0].State.Status)
 	}
 

realis_e2e_test.go

@@ -94,7 +94,7 @@ func TestBadCredentials(t *testing.T) {
 	job := realis.NewJob().
 		Environment("prod").
 		Role("vagrant").
-		Name("create_thermos_job_test").
+		Name("create_thermos_job_bad_creds_test").
 		ThermosExecutor(thermosExec).
 		CPU(.5).
 		RAM(64).
@@ -209,7 +209,6 @@ func TestValidAuroraURL(t *testing.T) {
 }
 
 func TestRealisClient_ReestablishConn(t *testing.T) {
-
 	// Test that we're able to tear down the old connection and create a new one.
 	err := r.ReestablishConn()
 
@@ -220,11 +219,9 @@ func TestGetCACerts(t *testing.T) {
 	certs, err := realis.GetCerts("./examples/certs")
 	assert.NoError(t, err)
 	assert.Equal(t, len(certs.Subjects()), 2)
-
 }
 
 func TestRealisClient_CreateJob_Thermos(t *testing.T) {
-
 	role := "vagrant"
 	job := realis.NewJob().
 		Environment("prod").
@@ -251,7 +248,7 @@ func TestRealisClient_CreateJob_Thermos(t *testing.T) {
 
 	// Fetch all Jobs
 	result, err := r.GetJobs(role)
-	fmt.Printf("GetJobs length: %+v \n", len(result.Configs))
+	fmt.Println("GetJobs length: ", len(result.Configs))
 	assert.Len(t, result.Configs, 1)
 	assert.NoError(t, err)
 
@@ -272,7 +269,7 @@ func TestRealisClient_CreateJob_Thermos(t *testing.T) {
 		err := r.KillJob(job.JobKey())
 		assert.NoError(t, err)
 
-		success, err := r.MonitorInstances(job.JobKey(), 0, 1*time.Second, 60*time.Second)
+		success, err := r.MonitorInstances(job.JobKey(), 0, 1*time.Second, 90*time.Second)
 		assert.True(t, success)
 		assert.NoError(t, err)
 	})
@@ -280,7 +277,6 @@ func TestRealisClient_CreateJob_Thermos(t *testing.T) {
 
 // Test configuring an executor that doesn't exist for CreateJob API
 func TestRealisClient_CreateJob_ExecutorDoesNotExist(t *testing.T) {
-
 	// Create a single job
 	job := realis.NewJob().
 		Environment("prod").
@@ -299,7 +295,6 @@ func TestRealisClient_CreateJob_ExecutorDoesNotExist(t *testing.T) {
 
 // Test configuring an executor that doesn't exist for CreateJob API
 func TestRealisClient_GetPendingReason(t *testing.T) {
-
 	env := "prod"
 	role := "vagrant"
 	name := "pending_reason_test"
@@ -333,7 +328,6 @@ func TestRealisClient_GetPendingReason(t *testing.T) {
 }
 
 func TestRealisClient_CreateService_WithPulse_Thermos(t *testing.T) {
-
 	fmt.Println("Creating service")
 	role := "vagrant"
 	job := realis.NewJobUpdate().
@@ -813,18 +807,15 @@ func TestRealisClient_BatchAwareAutoPause(t *testing.T) {
 	job := realis.NewJob().
 		Environment("prod").
 		Role("vagrant").
-		Name("BatchAwareAutoPauseTest").
+		Name("batch_aware_auto_pause_test").
 		ThermosExecutor(thermosExec).
 		CPU(.01).
 		RAM(4).
 		Disk(10).
 		InstanceCount(6).
-		IsService(true).
+		IsService(true)
-		Production(false).
-		Tier("preemptible").
-		Priority(0)
 
-	updateGroups := []int32{1, 2, 3}
+	updateGroups := []int32{1, 3}
 	strategy := realis.JobUpdateFromAuroraTask(job.AuroraTask()).
 		VariableBatchStrategy(true, updateGroups...).
 		InstanceCount(6).
@@ -837,22 +828,28 @@ func TestRealisClient_BatchAwareAutoPause(t *testing.T) {
 	key := *result.GetKey()
 
 	for i := range updateGroups {
-		curStep, mErr := r.MonitorAutoPausedUpdate(key, time.Second*5, time.Second*240)
+		curStep, mErr := r.MonitorAutoPausedUpdate(key, time.Second*5, time.Minute*5)
 		if mErr != nil {
+			fmt.Println(mErr)
 			// Update may already be in a terminal state so don't check for error
-			assert.NoError(t, r.AbortJobUpdate(key, "Monitor timed out."))
+			_ = r.AbortJobUpdate(key, "Monitor timed out.")
 		}
 
 		assert.Equal(t, i, curStep)
-		require.NoError(t, r.ResumeJobUpdate(key, "auto resuming test"))
+
+		if i != len(updateGroups)-1 {
+			require.NoError(t, err)
+			require.NoError(t, r.ResumeJobUpdate(key, "auto resuming test"))
+		}
 	}
+	assert.NoError(t, r.AbortJobUpdate(key, ""))
 	assert.NoError(t, r.KillJob(strategy.JobKey()))
 }
 
 func TestRealisClient_GetJobSummary(t *testing.T) {
 	role := "vagrant"
 	env := "prod"
-	name := "GetJobSummaryJob"
+	name := "test_get_job_summary"
 	// Create a single job
 	job := realis.NewJob().
 		Environment(env).
@@ -863,14 +860,10 @@ func TestRealisClient_GetJobSummary(t *testing.T) {
 		RAM(4).
 		Disk(10).
 		InstanceCount(3).
-		WatchTime(20 * time.Second).
 		IsService(true).
 		Production(false).
 		Tier("preemptible").
-		Priority(0).
+		Priority(0)
-		BatchSize(2)
-
-	result, err := r.CreateService(job)
 
 	err := r.CreateJob(job)
 	assert.NoError(t, err)

runTestsMac.sh

@@ -1,4 +1,4 @@
 #!/bin/bash
 
 # Since we run our docker compose setup in bridge mode to be able to run on MacOS, we have to launch a Docker container within the bridge network in order to avoid any routing issues.
-docker run --rm -t -v $(pwd):/go/src/github.com/aurora-scheduler/gorealis --network gorealis_aurora_cluster golang:1.13-stretch go test -v github.com/aurora-scheduler/gorealis $@
+docker run --rm -t -w /gorealis  -v $GOPATH/pkg:/go/pkg -v $(pwd):/gorealis --network gorealis_aurora_cluster golang:1.17-buster go test -v github.com/aurora-scheduler/gorealis/v2 $@