Internal TCP/UDP load balancer logging and monitoring

This document shows you how to configure and use Cloud Logging and Cloud Monitoring for internal TCP/UDP load balancers.

Logging

Logs provide useful information for troubleshooting and monitoring the pass-through Google Cloud load balancer. Logs are aggregated per connection and exported in near real time. Logs are generated for each load-balanced instance TCP and UDP flows for both ingress and egress traffic. For more information about the fields provided in the log entry, see Log fields.

There are no additional charges for using logs. Based on how you ingest logs, standard pricing for Cloud Logging, BigQuery, or Pub/Sub apply. Enabling logs has no effect on the performance of the load balancer.

Logging provides the following benefits:

Internal TCP/UDP load balancer traffic monitoring. Per-connection logging gives you insight into how each connection is routed to serving backends.
Network troubleshooting. You can utilize internal TCP/UDP load balancer logs for troubleshooting. For more information, see Troubleshoot internal TCP/UDP load balancers.

Sample log format for VM to VM flows

The following diagram shows the inbound and outbound traffic for an internal client (192.168.1.2), internal TCP/UDP load balancer (10.240.0.200), and backend instance (10.240.0.3).

Internal client to backend VM services flows. — Inbound and outbound flows for VM to VM flows.

Internal TCP/UDP load balancer logs for connections from the client to the backend instance are formatted as follows:

connection.clientIp: 192.168.1.2
connection.serverIp: 10.240.0.200
bytesSent: 1256
bytesReceived: 4521

Logs sampling and collection

Google Cloud samples the packets that leave and enter load balancer backend VMs. Those sampled packets are processed to generate logs.

Not every packet is sampled. Google Cloud samples a variable subset of packets depending on the amount of traffic on the physical host. The lowest possible sampling rate is one out of 1,024 packets. The sampling rate is dynamically controlled by Google Cloud. You cannot adjust the sampling rate.

The packet sampling interacts with firewall rules in the following ways:

Packets are sampled before egress firewall rules are applied.
Packets are sampled after ingress firewall rules are applied.

After packet sampling, Google Cloud processes the sampled packets according to the following procedure:

Aggregation: Sampled packets are aggregated over a five-second interval to produce a single flow entry.
Configurable (secondary) log sampling: This is a second sampling process, sampling the flows. You control the fraction of the flow entries that are emitted as log entries according to the logConfig.sampleRate parameter. When logConfig.sampleRate is 1.0 (100%), this means that all of the sampled packets are processed.
Write to logging: The log entries are written to Cloud Logging.

Optional fields

Log records contain required fields and optional fields. The Log fields section lists which fields are optional and which are required. All required fields are always included. You can customize which optional fields you keep.

If you select include all optional, all optional fields in the log record format are included in the flow logs. When new optional fields are added to the record format, the flow logs automatically include the new fields.
If you select exclude all optional, this omits all optional fields.
If you select custom, you can specify the optional fields that you want to include by the parent field, such as serverInstance, or by their full names, such as serverInstance.vm.

When new optional fields are added to the record format, the logs do not include these fields unless they are a new field within a parent field that you have specified to include.

If you specify a custom optional field by using parent fields, when new optional fields are added to the record format within that parent field, the logs automatically include the new fields. For example, if you choose to include optionalFieldA, then a new field added to the logs with the name optionalFieldA.subField1 is included automatically.

For instructions about customizing optional fields, see Enable logging on a new backend service.

Response packet source IP address requirements

Internal TCP/UDP load balancer logging samples the response packets from the backend VMs only if the source IP address for those packets matches the load balancer's forwarding rule IP address. For TCP connections, response packets must always have sources that match the request packet's destination; however, for other protocols, it's possible for response packets to use a different source IP address. For additional details, see IP addresses for request and return packets.

The packet sampling process used by internal TCP/UDP load balancers omits any response packets from backend VMs if those response packets have sources that do not match an IP address of a forwarding rule for an internal TCP/UDP load balancer.

Enable logging for a new backend service

Console

In the Google Cloud console, go to the Load balancing page.

Go to Load balancing
Click the name of your load balancer.
Click Edit, and then click Backend Configuration.
Select Create a backend service, and then complete the required backend service fields.
Click Enable logging.
Set a Sample rate fraction. You can set a rate to 0.0 through 1.0 (default).
Optional: To include all the optional fields in the logs, in the Optional fields section, click Include all optional fields.
Pro tip: To specify the CUSTOM option, use the gcloud CLI and the REST API.
To finish editing the backend service, click Update.
To finish editing the load balancer, click Update.

gcloud

Create a backend service and enable logging by using the gcloud compute backend-services create command.

gcloud compute backend-services create BACKEND_SERVICE \
  --region=REGION \
  --enable-logging \
  --logging-sample-rate=SAMPLE_RATE
  --logging-optional=LOGGING_OPTIONAL \
  --logging-optional-fields=OPTIONAL_FIELDS

Replace the following:

BACKEND_SERVICE: the name of the backend service.
REGION: the region of the backend service to create.
SAMPLE_RATE: this field can only be specified if logging is enabled for this backend service.

The value of the field must be from 0.0 to 1.0, where 0.0 means that no logs are reported and 1.0 means that all logged requests are reported. Enabling logging but setting the sampling rate to 0.0 is equivalent to disabling logging. The default value is 1.0.
LOGGING_OPTIONAL: the optional annotations that you want to include in the logs:
- INCLUDE_ALL_OPTIONAL to include all optional annotations.
- EXCLUDE_ALL_OPTIONAL (default) to exclude all optional annotations.
- CUSTOM to include a custom list of optional fields that you specify in OPTIONAL_FIELDS.
OPTIONAL_FIELDS: a comma-separated list of optional fields that you want to include in the logs.

For example, serverInstance.vm, serverGkeDetails. Can only be set if LOGGING_OPTIONAL is set to CUSTOM.

API

Make a POST request to the regionBackendServices.insert method.

 POST https://compute.googleapis.com/compute/v1/projects/PROJECT_ID/regions/REGION/backendServices

Exclude all optional fields

 {
 "name": "BACKEND_SERVICE",
  "loadBalancingScheme": "INTERNAL",
  "logConfig": {
   "enable": true,
   "sampleRate": SAMPLE_RATE,
   "optionalMode": "EXCLUDE_ALL_OPTIONAL"
  }
 }

Include all optional fields

{
 "name": "BACKEND_SERVICE",
 "loadBalancingScheme": "INTERNAL",
 "logConfig": {
   "enable": true,
   "sampleRate": SAMPLE_RATE,
   "optionalMode": "INCLUDE_ALL_OPTIONAL"
  }
}

Include a custom list of optional fields

{
 "name": "BACKEND_SERVICE",
 "loadBalancingScheme": "INTERNAL",
 "logConfig": {
   "enable": true,
   "sampleRate": SAMPLE_RATE,
   "optionalMode": "CUSTOM",
   "optionalFields": ["field1","field2",...]
  }
}

Enable logging on an existing backend service

Console

In the Google Cloud console, go to the Load balancing page.

Go to Load balancing
Click the name of your load balancer.
Click Edit, and then click Backend Configuration.
Next to your backend service, click Edit.
Click Enable logging.
Set a Sample rate fraction. You can set a rate to 0.0 through 1.0 (default).
Optional: To include all the optional fields in the logs, in the Optional fields section, click Include all optional fields.
Pro tip: To specify the CUSTOM option, use the gcloud CLI and the REST API.
To finish editing the backend service, click Update.
To finish editing the load balancer, click Update.

gcloud

Enable logging on an existing backend service with the gcloud compute backend-services update command.

gcloud compute backend-services update BACKEND_SERVICE \
--region=REGION \
--enable-logging \
--logging-sample-rate=SAMPLE_RATE \
--logging-optional=LOGGING_OPTIONAL \
--logging-optional-fields=OPTIONAL_FIELDS

Replace the following:

BACKEND_SERVICE: the name of the backend service.
REGION: the region of the backend service to create.
SAMPLE_RATE: this field can only be specified if logging is enabled for this backend service.

The value of the field must be from 0.0 to 1.0, where 0.0 means that no logs are reported and 1.0 means that all logged requests are reported. Enabling logging but setting the sampling rate to 0.0 is equivalent to disabling logging. The default value is 1.0.
LOGGING_OPTIONAL: the optional annotations that you want to include in the logs.
- INCLUDE_ALL_OPTIONAL to include all optional annotations.
- EXCLUDE_ALL_OPTIONAL (default) to exclude all optional annotations.
- CUSTOM to include a custom list of optional fields that you specify in OPTIONAL_FIELDS.
OPTIONAL_FIELDS: a comma-separated list of optional fields that you want to include in the logs.

For example, serverInstance.vm,serverGkeDetails. Can only be set if LOGGING_OPTIONAL is set to CUSTOM.

API

Make a PATCH request to the regionBackendServices/patch method.

 PATCH https://compute.googleapis.com/compute/v1/projects/PROJECT_ID/regions/REGION/backendServices/BACKEND_SERVICE

Exclude all optional fields

 "logConfig": {
   "enable": true,
   "sampleRate": SAMPLE_RATE,
   "optionalMode": "EXCLUDE_ALL_OPTIONAL"
  }

Include all optional fields

 "logConfig": {
   "enable": true,
   "sampleRate": SAMPLE_RATE,
   "optionalMode": "INCLUDE_ALL_OPTIONAL"
  }

Include a custom list of optional fields

 "logConfig": {
   "enable": true,
   "sampleRate": SAMPLE_RATE,
   "optionalMode": "CUSTOM",
   "optionalFields": ["field1","field2",...]
  }

Disable logging on an existing backend service

Console

In the Google Cloud console, go to the Load balancing page.

Go to Load balancing
Click the name of your load balancer.
Click Edit, and then click Backend Configuration.
To disable logging entirely, clear Enable logging.
If you leave logging enabled, you can set a different Sample rate fraction. You can set the rate to 0.0 through 1.0 (default). To cut the number of stored logs to 20%, set the value to 0.2.
To finish editing the backend service, click Update.
To finish editing the load balancer, click Update.

gcloud

Disable logging on a backend service with the gcloud compute backend-services update command.

gcloud compute backend-services update BACKEND_SERVICE \
  --region=REGION \
  --no-enable-logging

Replace the following:

BACKEND_SERVICE: the name of the backend service.
REGION: the region of the backend service to create.

API

Make a PATCH request to the regionBackendServices/patch method.

 PATCH https://compute.googleapis.com/compute/v1/projects/PROJECT_ID/regions/REGION/backendServices/BACKEND_SERVICE

 {
 "logConfig": {
   "enable": false
  }
 }

View logs

When logs are ingested into Cloud Logging and not excluded through a Log router sink, you can read logs by using the Cloud Logging API and the Google Cloud CLI.

To view all the internal TCP/UDP load balancer logs:

Console

In the Google Cloud console, go to the Logs Explorer page.

Go to Logs Explorer
Select the Google Cloud Internal Network Load Balancer Rule resource type.
Select the loadbalancing.googleapis.com/flows log name.

Console query

In the Google Cloud console, go to the Logs Explorer page.

Go to Logs Explorer
Click the Show query toggle.

Paste the following into the query field. Replace PROJECT_ID with your project ID.

resource.type="loadbalancing.googleapis.com/InternalNetworkLoadBalancerRule"
logName="projects/PROJECT_ID/logs/loadbalancing.googleapis.com%2Fflows"

Click Run query.

View logs for a backend service

To view the internal TCP/UDP load balancer logs for a specific backend service:

Console query

In the Google Cloud console, go to the Logs Explorer page.

Go to Logs Explorer
Click the Show query toggle.

Paste the following into the query field. Replace PROJECT_ID with your project ID and BACKEND_SERVICE_NAME with the name of your backend service.

resource.type="loadbalancing.googleapis.com/InternalNetworkLoadBalancerRule"
logName="projects/PROJECT_ID/logs/loadbalancing.googleapis.com%2Fflows"
resource.labels.backend_service_name="BACKEND_SERVICE_NAME"

Click Run query.

View logs for a backend instance group

To view the internal TCP/UDP load balancer logs for a specific backend instance group or network endpoint group (NEG) with GCE_VM_IP endpoints, follow these steps:

Console query

In the Google Cloud console, go to the Logs Explorer page.

Go to Logs Explorer
Click the Show query toggle.

Paste the following into the query field. Replace PROJECT_ID with your project ID and the BACKEND_GROUP_NAME with the name of the instance group or NEG.

resource.type="loadbalancing.googleapis.com/InternalNetworkLoadBalancerRule"
logName="projects/PROJECT_ID/logs/loadbalancing.googleapis.com%2Fflows"
resource.labels.backend_group_name="BACKEND_GROUP_NAME"

Click Run query.

Log fields

Log records contain required fields—the default fields of every log record—and optional fields that add additional information. Optional fields can be omitted to save storage costs.

Some log fields are in a multi-field format, with more than one piece of data in a given field. For example, the connection field is of the IpConnection format, which contains the source and destination IP address and port plus the protocol in a single field. These multi-field fields are described in the following record format table.

The monitored resource is loadbalancing.googleapis.com/InternalNetworkLoadBalancerRule.

Field	Field format	Field type: Required or Optional	Description
connection	IpConnection	Required	5-Tuple describing this connection.
startTime	string	Required	Timestamp (RFC 3339 date string format) of the first observed packet during the aggregated time interval.
endTime	string	Required	Timestamp (RFC 3339 date string format) of the last observed packet during the aggregated time interval.
bytesSent	string	Required	Number of bytes sent from the server to the client.
bytesReceived	int64	Required	Number of bytes received by the server from the client.
packetsSent	int64	Required	Number of packets sent from the server to the client.
packetsReceived	int64	Required	Number of packets received by the server from the client.
rtt	string	Required	Latency is measured only for TCP connections. Latency is the sum of estimated network round-trip time (RTT) plus time consumed processing the packet within the client's VM operating system. For sampled packets, the RTT is calculated from the perspective of a load-balanced backend by measuring time differences between the backend sending a TCP segment and the backend receiving a TCP acknowledgment for the sequence number of the sent segment. Latency is formatted as a string that begins with the number of seconds and ends with "s" to indicate seconds. Nanoseconds is expressed as fractional seconds. For example, the latency of 250 milliseconds is formatted as "0.250000000s".
clientInstance	InstanceDetails	Optional	VM instance details of the client. Only available if the client's VM instance is located in the same project.
serverInstance	InstanceDetails	Optional	VM instance details of the backend VM.
clientVpc	VpcDetails	Optional	VPC network details of the client. Only available if the client is an endpoint inside a VPC network located in the same project.
clientGkeDetails	GkeDetails	Optional	GKE metadata for the client. Only available if the client is a GKE endpoint in the same project.
serverGkeDetails	GkeDetails	Optional	GKE metadata for the server backend. Only available if the backend is a GKE endpoint.

IpConnection field format

Field	Type	Description
clientIp	string	Client IP address
clientPort	int32	Client port. Set for TCP and UDP connections only.
serverIp	string	Server IP address (forwarding rule IP)
serverPort	int32	Server port. Set for TCP and UDP connections only.
protocol	int32	IANA protocol number

InstanceDetails field format

Field	Type	Description
projectId	string	ID of the project containing the VM
vm	string	Instance name of the VM
region	string	Region of the VM
zone	string	Zone of the VM
vmIp	string	Primary internal IP of the NIC that served the connection

VpcDetails field format

Field	Type	Description
projectId	string	ID of the project containing the VPC network
vpc	string	VPC network on which the VM is operating
subnetwork	string	Subnetwork on which the VM is operating

GkeDetails field format

Field	Type	Description
cluster	ClusterDetail	GKE cluster metadata
pod	PodDetails	GKE Pod metadata, populated when the source or destination of the traffic is a Pod
service	ServiceDetails	GKE Service metadata, populated in service endpoints only. The record contains up to two services. If there are more than two relevant services, this field contains a single service with a special MANY_SERVICES marker.

ClusterDetails field format

Field	Type	Description
cluster	string	GKE cluster name
clusterLocation	string	Location of the cluster. The cluster location can be a zone or a region.

PodDetails field format

Field	Type	Description
pod	string	Name of the Pod
podNamespace	string	Namespace of the Pod

ServiceDetails field format

Field	Type	Description
service	string	Name of the service. If there are more than two relevant services, the field is set to a special MANY_SERVICES marker.
serviceNamespace	string	Namespace of the service

Monitoring

An internal TCP/UDP load balancer exports monitoring data to Cloud Monitoring.

Monitoring metrics can be used for the following purposes:

Evaluate an internal TCP/UDP load balancer's configuration, usage, and performance
Troubleshoot problems
Improve resource utilization and user experience

In addition to the predefined dashboards in Monitoring, you can create custom dashboards, set up alerts, and query the metrics through the Monitoring API.

View Monitoring dashboards

Console

In the Google Cloud console, go to the Monitoring page.

Go to Monitoring
If Resources appears in the navigation pane, select Resources, and then select Google Cloud Load Balancers. Otherwise, select Dashboards, and then select the dashboard named Google Cloud Load Balancers.
Click the name of your load balancer.

In the left pane, you can see various details for the selected load balancer. In the right pane, you can see time series graphs. Click the Breakdowns link to see specific breakdowns. The left pane presents currently configured data, while the right pane can present data served by historical configurations not currently reflected in the left pane.

Define Monitoring custom dashboards

You can create custom Monitoring dashboards for internal TCP/UDP load balancer metrics.

Console

In the Google Cloud console, go to the Monitoring page.

Go to Monitoring
Select Dashboards > Create Dashboard.
Click Add Chart.
Give the chart a title.
Select metrics and filters. For metrics, the resource type is Google Cloud TCP Load Balancer (Internal) Rule (internal_tcp_lb_rule) or Google Cloud UDP Load Balancer (Internal) Rule (internal_udp_lb_rule).
Click Save.

Define alerting policies

Console

You can create alerting policies to monitor the values of metrics and to notify you when those metrics violate a condition.

In the Google Cloud console, go to the Monitoring page.
Go to Monitoring
In the Monitoring navigation pane, select Alerting.
If you haven't created your notification channels and if you want to be notified, then click Edit Notification Channels and add your notification channels. Return to the Alerting page after you add your channels.
From the Alerting page, select Create policy.
To select the metric, expand the Select a metric menu and then do the following:
1. To limit the menu to relevant entries, enter Google Cloud TCP Load Balancer or Google Cloud UDP Load Balancer into the filter bar. If there are no results after you filter the menu, then disable the Show only active resources & metrics toggle.
2. For the Resource type, select Google Cloud TCP Load Balancer or Google Cloud UDP Load Balancer.
3. Select a Metric category and a Metric, and then select Apply.
Click Next.
The settings in the Configure alert trigger page determine when the alert is triggered. Select a condition type and, if necessary, specify a threshold. For more information, see Create metric-threshold alerting policies.
Click Next.
Optional: To add notifications to your alerting policy, click Notification channels. In the dialog, select one or more notification channels from the menu, and then click OK.
Optional: Update the Incident autoclose duration. This field determines when Monitoring closes incidents in the absence of metric data.
Optional: Click Documentation, and then add any information that you want included in a notification message.
Click Alert name and enter a name for the alerting policy.
Click Create Policy.

For more information, see Alerting policies.

Metrics for internal TCP/UDP load balancers

The following metrics for internal TCP/UDP load balancers are reported into Monitoring.

Metric	Description
Inbound Throughput	The number of bytes sent towards internal TCP/UDP load balancer forwarding rules, as received by the backends.
Inbound Packets	The number of packets sent towards internal TCP/UDP load balancer forwarding rules, as received by the backends.
Outbound Throughput	The number of bytes sent by internal load-balanced backends on connections bound to forwarding rule IP addresses.
Outbound Packets	The number of packets sent by internal load-balanced backends on connections bound to forwarding rule IP addresses.
Latency(*)	The distribution of the RTT measured for bundles of packets over each internal load-balanced connection. Typically reduced to 95th-percentile in Monitoring views.

(*) Available only for TCP traffic.

Filter dimensions for internal TCP/UDP load balancer metrics

Metrics are aggregated for each internal TCP/UDP load balancer. Metrics can be further broken down by the following dimensions:

Property	Description
BACKEND NAME	The name of the instance group or network endpoint group (NEG) with `GCE_VM_IP` endpoints.
BACKEND SCOPE	The scope (region or zone) of the backend that received the connection.
BACKEND ZONE	For zonal instance groups and network endpoint groups, zone of the backend that served the connection.
CLIENT NETWORK	The network of the client instance that connected to the internal TCP/UDP load balancer.
CLIENT SUBNETWORK	The subnetwork of the client instance that connected to the internal TCP/UDP load balancer.
CLIENT ZONE	The Google Cloud zone of the instance that connected to the forwarding rule.
FORWARDING RULE	The name of the internal TCP/UDP load balancer's forwarding rule.

Metric reporting frequency and retention

Metrics for the internal TCP/UDP load balancers are exported to Monitoring in one-minute granularity batches. Monitoring data is retained for six weeks. Metrics are based on sampled traffic (sampling rate is dynamic and cannot be adjusted). The dashboard provides data analysis in default intervals of one hour (1H), six hours (6H), one day (1D), one week (1W), and six weeks (6W). You can manually request analysis in any interval from six weeks to one minute.

What's next

For information about fundamentals, see Internal TCP/UDP load balancer overview.
For information about failover, see Failover for internal TCP/UDP load balancers.
For available DNS name options for your load balancer, see Internal load balancing and DNS names.
For an example internal TCP/UDP load balancer configuration, see Set up an internal TCP/UDP load balancer with VM instance group backends.
For information about how to configure and test an internal TCP/UDP load balancer that uses failover configuration, see Configure failover for internal TCP/UDP load balancers.
For information about accessing internal load balancers from peer networks connected to your VPC network, see Internal TCP/UDP load balancers and connected networks.
For information about how to troubleshoot, see Troubleshoot internal TCP/UDP load balancers.
For information about how to troubleshoot issues with logs, see Troubleshoot logging issues.