Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Add SQS InterruptionEvent.InstanceType #1104

Merged
merged 4 commits into from
Dec 31, 2024
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
1 change: 1 addition & 0 deletions pkg/monitor/sqsevent/asg-lifecycle-event.go
Original file line number Diff line number Diff line change
Expand Up @@ -91,6 +91,7 @@ func (m SQSMonitor) asgTerminationToInterruptionEvent(event *EventBridgeEvent, m
IsManaged: nodeInfo.IsManaged,
InstanceID: lifecycleDetail.EC2InstanceID,
ProviderID: nodeInfo.ProviderID,
InstanceType: nodeInfo.InstanceType,
Description: fmt.Sprintf("ASG Lifecycle Termination event received. Instance will be interrupted at %s \n", event.getTime()),
}

Expand Down
1 change: 1 addition & 0 deletions pkg/monitor/sqsevent/ec2-state-change-event.go
Original file line number Diff line number Diff line change
Expand Up @@ -75,6 +75,7 @@ func (m SQSMonitor) ec2StateChangeToInterruptionEvent(event *EventBridgeEvent, m
AutoScalingGroupName: nodeInfo.AsgName,
InstanceID: ec2StateChangeDetail.InstanceID,
ProviderID: nodeInfo.ProviderID,
InstanceType: nodeInfo.InstanceType,
Description: fmt.Sprintf("EC2 State Change event received. Instance %s went into %s at %s \n", ec2StateChangeDetail.InstanceID, ec2StateChangeDetail.State, event.getTime()),
}

Expand Down
1 change: 1 addition & 0 deletions pkg/monitor/sqsevent/rebalance-recommendation-event.go
Original file line number Diff line number Diff line change
Expand Up @@ -67,6 +67,7 @@ func (m SQSMonitor) rebalanceRecommendationToInterruptionEvent(event *EventBridg
IsManaged: nodeInfo.IsManaged,
InstanceID: nodeInfo.InstanceID,
ProviderID: nodeInfo.ProviderID,
InstanceType: nodeInfo.InstanceType,
Description: fmt.Sprintf("Rebalance recommendation event received. Instance %s will be cordoned at %s \n", rebalanceRecDetail.InstanceID, event.getTime()),
}
interruptionEvent.PostDrainTask = func(interruptionEvent monitor.InterruptionEvent, n node.Node) error {
Expand Down
1 change: 1 addition & 0 deletions pkg/monitor/sqsevent/scheduled-change-event.go
Original file line number Diff line number Diff line change
Expand Up @@ -102,6 +102,7 @@ func (m SQSMonitor) scheduledEventToInterruptionEvents(event *EventBridgeEvent,
NodeName: nodeInfo.Name,
InstanceID: nodeInfo.InstanceID,
ProviderID: nodeInfo.ProviderID,
InstanceType: nodeInfo.InstanceType,
IsManaged: nodeInfo.IsManaged,
Description: fmt.Sprintf("AWS Health scheduled change event received. Instance %s will be interrupted at %s \n", nodeInfo.InstanceID, event.getTime()),
}
Expand Down
1 change: 1 addition & 0 deletions pkg/monitor/sqsevent/spot-itn-event.go
Original file line number Diff line number Diff line change
Expand Up @@ -69,6 +69,7 @@ func (m SQSMonitor) spotITNTerminationToInterruptionEvent(event *EventBridgeEven
IsManaged: nodeInfo.IsManaged,
InstanceID: spotInterruptionDetail.InstanceID,
ProviderID: nodeInfo.ProviderID,
InstanceType: nodeInfo.InstanceType,
Description: fmt.Sprintf("Spot Interruption notice for instance %s was sent at %s \n", spotInterruptionDetail.InstanceID, event.getTime()),
}
interruptionEvent.PostDrainTask = func(interruptionEvent monitor.InterruptionEvent, n node.Node) error {
Expand Down
24 changes: 13 additions & 11 deletions pkg/monitor/sqsevent/sqs-monitor.go
Original file line number Diff line number Diff line change
Expand Up @@ -340,12 +340,13 @@ func (m SQSMonitor) completeLifecycleAction(input *autoscaling.CompleteLifecycle

// NodeInfo is relevant information about a single node
type NodeInfo struct {
AsgName string
InstanceID string
ProviderID string
IsManaged bool
Name string
Tags map[string]string
AsgName string
InstanceID string
ProviderID string
InstanceType string
IsManaged bool
Name string
Tags map[string]string
}

// getNodeInfo returns the NodeInfo record for the given instanceID.
Expand Down Expand Up @@ -411,11 +412,12 @@ func (m SQSMonitor) getNodeInfo(instanceID string) (*NodeInfo, error) {
}

nodeInfo := &NodeInfo{
Name: *instance.PrivateDnsName,
InstanceID: instanceID,
ProviderID: providerID,
Tags: make(map[string]string),
IsManaged: true,
Name: *instance.PrivateDnsName,
InstanceID: instanceID,
ProviderID: providerID,
InstanceType: *instance.InstanceType,
Tags: make(map[string]string),
IsManaged: true,
}
for _, t := range (*instance).Tags {
nodeInfo.Tags[*t.Key] = *t.Value
Expand Down
1 change: 1 addition & 0 deletions pkg/monitor/sqsevent/sqs-monitor_internal_test.go
Original file line number Diff line number Diff line change
Expand Up @@ -199,6 +199,7 @@ func getDescribeInstancesResp(instanceID string, privateDNSName string, tags map
GroupName: aws.String(""),
Tenancy: aws.String("default"),
},
InstanceType: aws.String("t3.medium"),
PrivateDnsName: aws.String(privateDNSName),
Tags: awsTags,
},
Expand Down
1 change: 1 addition & 0 deletions pkg/monitor/sqsevent/sqs-monitor_test.go
Original file line number Diff line number Diff line change
Expand Up @@ -929,6 +929,7 @@ func getDescribeInstancesResp(privateDNSName string, withASGTag bool, withManage
GroupName: aws.String(""),
Tenancy: aws.String("default"),
},
InstanceType: aws.String("t3.medium"),
PrivateDnsName: &privateDNSName,
Tags: tags,
},
Expand Down
1 change: 1 addition & 0 deletions pkg/monitor/types.go
Original file line number Diff line number Diff line change
Expand Up @@ -53,6 +53,7 @@ type InterruptionEvent struct {
Pods []string
InstanceID string
ProviderID string
InstanceType string
IsManaged bool
StartTime time.Time
EndTime time.Time
Expand Down
12 changes: 9 additions & 3 deletions pkg/webhook/webhook.go
Original file line number Diff line number Diff line change
Expand Up @@ -33,7 +33,8 @@ import (
type combinedDrainData struct {
ec2metadata.NodeMetadata
monitor.InterruptionEvent
InstanceID string
InstanceID string
InstanceType string
}

// Post makes a http post to send drain event data to webhook url
Expand All @@ -60,12 +61,17 @@ func Post(additionalInfo ec2metadata.NodeMetadata, event *monitor.InterruptionEv
return
}

// Need to merge the two data sources manually since both have an InstanceID field
// Need to merge the two data sources manually since both have
// InstanceID and InstanceType fields
instanceID := additionalInfo.InstanceID
if event.InstanceID != "" {
instanceID = event.InstanceID
}
var combined = combinedDrainData{NodeMetadata: additionalInfo, InterruptionEvent: *event, InstanceID: instanceID}
instanceType := additionalInfo.InstanceType
if event.InstanceType != "" {
instanceType = event.InstanceType
}
var combined = combinedDrainData{NodeMetadata: additionalInfo, InterruptionEvent: *event, InstanceID: instanceID, InstanceType: instanceType}

var byteBuffer bytes.Buffer
err = webhookTemplate.Execute(&byteBuffer, combined)
Expand Down
Loading