Prometheus metrics

Anbox Cloud gathers various performance metrics that you can access through API endpoints to create a monitoring solution. See How to monitor Anbox Cloud for detailed information.

The reference implementation for Prometheus (see Example: Collect metrics) provides some basic dashboards. You can, however, update them to fit your needs.

The following sections list all metrics returned to Prometheus by the Anbox Cloud services that support Prometheus metrics.

AMS

The Anbox Management Service provides metrics about the Anbox Cloud cluster (or the Anbox Cloud Application server) and the AMS API access.

AMS metrics endpoint

You can access the AMS metrics from any machine that is on the same network as your Anbox Cloud installation. Use the following endpoint:

http://<AMS_server>:<AMS_port>/internal/1.0/metrics

Replace <AMS_server> with the IP address of your AMS server, which you can determine by running one of the following commands:

  • For a full Anbox Cloud deployment: juju run --unit ams/0 -- unit-get private-address
  • For the Anbox Cloud Appliance: juju run -m appliance:anbox-cloud --unit ams/0 -- unit-get private-address

Replace <AMS_port> with the port for the API endpoint, which you can determine by running one of the following commands:

  • For a full Anbox Cloud deployment: juju config ams prometheus_target_port
  • For the Anbox Cloud Appliance: juju config -m appliance:anbox-cloud ams prometheus_target_port

You can then access the endpoint with curl, for example:

curl http://192.0.2.55:20002/internal/1.0/metrics

AMS cluster metrics

Metrics prefixed with ams_cluster_ keep you informed about the status of your Anbox Cloud cluster.
These metrics are available since Anbox Cloud 1.0.0.

Name Description
ams_cluster_nodes_total Number of nodes in the cluster
ams_cluster_applications_total Number of applications
ams_cluster_containers_total Number of containers currently in the cluster
ams_cluster_container_boot_time_seconds_count Number of container boot time measured
ams_cluster_container_boot_time_seconds_sum Sum of all container boot times (can be used to compute the average boot time)
ams_cluster_container_boot_time_seconds_bucket Container boot times bucket (can be used for alerting when above a threshold; see the Prometheus documentation for more information)
ams_cluster_containers_per_application_total Number of containers per application
ams_cluster_containers_per_status_total Number of containers per container status
ams_cluster_available_cpu_total Total CPUs available in each worker node
ams_cluster_used_cpu_total Used CPUs in each worker node
ams_cluster_available_memory_total Total memory available in each worker node
ams_cluster_used_memory_total Used memory in each worker node

AMS API metrics

Metrics prefixed with ams_http_ allow to track access to the API.
These metrics are available since Anbox Cloud 1.10.0.

Name Description
ams_http_in_flight_requests Number of HTTP requests being processed at the moment
ams_http_request_duration_seconds_bucket The HTTP request latency in seconds
ams_http_request_size_bytes_bucket The HTTP request size in bytes
ams_http_requests_total Total number of HTTP requests made
ams_http_response_size_bytes_bucket The HTTP response sizes in bytes

API handlers

To give a more granular approach to monitoring, the AMS API metrics contain handlers that identify the kind of API access.

The AMS API can be accessed through HTTP/HTTPS or a Unix domain socket. Therefore, the API metrics distinguish between http, https and unix. For example:

ams_http_request_duration_seconds_bucket{handler="http_applications_GET",host="juju-2db13b-1",method="get",le="0.5"} 1
ams_http_request_duration_seconds_bucket{handler="https_applications_GET",host="juju-2db13b-1",method="get",le="0.5"} 1
ams_http_request_duration_seconds_bucket{handler="unix_applications_GET",host="juju-2db13b-1",method="get",le="0.5"} 1

All handler labels adopt the convention <transport method>_<object>_<http method>, for example, unix_containers_POST.

The following table contains all routes and their corresponding labels (ignoring the communication method prefix).

Method Route Label
GET /1.0 service_GET
GET /1.0/addons addons_GET
POST /1.0/addons addons_POST
GET /1.0/addons/<id> addon_GET
PATCH /1.0/addons/<id> addon_PATCH
DELETE /1.0/addons/<id> addon_DELETE
DELETE /1.0/addons/<id>/<version> addon_version_DELETE
GET /1.0/applications applications_GET
POST /1.0/applications applications_POST
GET /1.0/applications/<id> application_GET
PATCH /1.0/applications/<id> application_PATCH
DELETE /1.0/applications/<id> application_DELETE
GET /1.0/applications/<id>/<version> application_version_GET
PATCH /1.0/applications/<id>/<version> application_version_PATCH
DELETE /1.0/applications/<id>/<version> application_version_DELETE
GET /1.0/applications/<id>/manifest application_manifest_GET
GET /1.0/applications/<id>/<version>/manifest application_version_manifest_GET
GET /1.0/containers containers_GET
POST /1.0/containers containers_POST
GET /1.0/containers/<id> container_GET
DELETE /1.0/containers/<id> container_DELETE
POST /1.0/containers/<id>/exec container_exec_POST
GET /1.0/containers/<id>/logs container_logs_GET
GET /1.0/containers/<id>/logs/<name> container_log_GET
GET /1.0/version version_GET
GET /1.0/nodes nodes_GET
POST /1.0/nodes nodes_POST
GET /1.0/nodes/<id> node_GET
PATCH /1.0/nodes/<id> node_PATCH
DELETE /1.0/nodes/<id> node_DELETE
GET /1.0/images images_GET
POST /1.0/images images_POST
GET /1.0/images/<id> image_GET
PATCH /1.0/images/<id> image_PATCH
DELETE /1.0/images/<id> image_DELETE
DELETE /1.0/images/<id>/<version> image_version_DELETE
GET /1.0/config config_GET
PATCH /1.0/config config_PATCH
GET /1.0/tasks tasks_GET
GET /1.0/registry/applications registry_applications_GET
DELETE /1.0/registry/applications/<id> registry_application_DELETE
POST /1.0/registry/applications/<id>/push registry_application_push_POST
POST /1.0/registry/applications/<id>/pull registry_application_pull_POST
GET /1.0/artifacts/<id> internal_artifacts_GET
PATCH /1.0/containers/<id> internal_containers_PATCH

Anbox Stream Gateway

The Anbox Stream Gateway provides metrics about the streaming activities of your cluster or server and the Anbox Stream Gateway API access.

Anbox Stream Gateway endpoint

You can access the Anbox Stream Gateway metrics from any machine that is on the same network as your Anbox Cloud installation. Use the following endpoint:

https://<gateway_server>:<gateway_port>/internal/1.0/metrics

Replace <gateway_server> with the IP address of your Anbox Stream Gateway server, which you can determine by running one of the following commands:

  • For a full Anbox Cloud deployment: juju run --unit anbox-stream-gateway/0 -- unit-get private-address
  • For the Anbox Cloud Appliance: juju run -m appliance:anbox-cloud --unit anbox-stream-gateway/0 -- unit-get private-address

Replace <gateway_port> with the port for the API endpoint, which you can determine by running one of the following commands:

  • For a full Anbox Cloud deployment: juju config anbox-stream-gateway prometheus_port
  • For the Anbox Cloud Appliance: juju config -m appliance:anbox-cloud anbox-stream-gateway prometheus_port

The Anbox Stream Gateway endpoint is on HTTPS, and therefore you must authenticate to access it. You can retrieve the credentials from the /var/snap/anbox-stream-gateway/common/service/config.yaml file on the gateway server:

  • For a full Anbox Cloud deployment:

    juju ssh anbox-stream-gateway/0
    sudo cat /var/snap/anbox-stream-gateway/common/service/config.yaml
    
  • For the Anbox Cloud Appliance:

    juju ssh -m appliance:anbox-cloud anbox-stream-gateway/0
    sudo cat /var/snap/anbox-stream-gateway/common/service/config.yaml
    

You can then access the endpoint with curl, for example:

curl -k -u prometheusadmin:thepassword https://192.0.2.55:9105/internal/1.0/metrics

Anbox Stream Gateway metrics

Metrics prefixed with anbox_stream_gateway_ give information about your cluster related to streaming, for example, the number of sessions and agents.
These metrics are available since Anbox Cloud 1.7.2.

Name Description
anbox_stream_gateway_sessions_total Total number of sessions, categorised by status
anbox_stream_gateway_accounts_total Total number of accounts
anbox_stream_gateway_agents_total Number of active and unresponsive agents

Anbox Stream Gateway API metrics

Metrics prefixed with anbox_stream_gateway_http_ allow to track access to the streaming API.
These metrics are available since Anbox Cloud 1.9.0.

Name Description
anbox_stream_gateway_http_in_flight_requests Number of HTTP requests being processed at the moment
anbox_stream_gateway_http_request_duration_seconds_bucket The HTTP request latency in seconds
anbox_stream_gateway_http_request_size_bytes_bucket The HTTP request size in bytes
anbox_stream_gateway_http_requests_total Total number of HTTP requests made
anbox_stream_gateway_http_response_size_bytes_bucket The HTTP response sizes in bytes

API handlers

To give a more granular approach to monitoring, the Anbox Stream Gateway API metrics contain handlers that identify the route that is accessed. The routes are indicated in the handler labels. For example:

anbox_stream_gateway_http_request_duration_seconds_bucket{handler="get_sessions",host="juju-2db13b-1",method="get",le="0.5"} 1

In this case, the label for the route is get_sessions.

The following table contains all routes and their corresponding labels.

Method Route Label
GET /1.0/status get_service_status
POST /1.0/sessions create_new_session
GET /1.0/sessions get_sessions
GET /1.0/sessions/<id> get_session
DELETE /1.0/sessions/<id> delete_session
DELETE /1.0/sessions delete_sessions
POST /1.0/sessions/<id>/join join_session
GET /1.0/sessions/<id>/sockets/master streaming_websocket_master
GET /1.0/sessions/<id>/sockets/slave streaming_websocket_slave
GET /1.0/applications get_applications
GET /1.0/regions get_regions

LXD

LXD provides metrics about the LXD cluster that Anbox Cloud uses.

LXD metrics have been available since LXD version 4.19. You cannot access LXD metrics if you are running an older version of LXD.

LXD metrics endpoint

You can access the LXD metrics through the following endpoint:

 https://<LXD_server>:8443/1.0/metrics

Replace <LXD_server> with the IP address of your LXD server, which you can determine by running one of the following commands:

  • For a full Anbox Cloud deployment: juju run --unit lxd/0 -- unit-get private-address
  • For the Anbox Cloud Appliance: juju run -m appliance:anbox-cloud --unit lxd/0 -- unit-get private-address

The LXD metrics endpoint is on HTTPS, and therefore you must authenticate to access it. See Create metrics certificate in the LXD documentation for instructions on how to create a certificate.

Alternatively, if you are using the Anbox Cloud Appliance, you can also access the LXD metrics through the local Unix socket. In this case, you don’t need authentication. To use this method, enter the following command:

curl --unix-socket /var/snap/lxd/common/lxd/unix.socket s/1.0/metrics

LXD metrics

You can find the list of metrics that LXD provides in the LXD documentation.

WebRTC

The WebRTC metrics are collected by Telegraf on every cluster member.

WebRTC metrics endpoint

You can access the WebRTC metrics as part of the metrics that Telegraf collects and provides to Prometheus. You can access them from any machine that is on the same network as your Anbox Cloud installation. To find the endpoint, check the /var/snap/anbox-cloud-appliance/common/telegraf/main.conf file on each cluster member and look for the listen address and the path in the [[outputs.prometheus_client]] section.

You can then access the endpoint with curl, for example:

curl http://192.0.2.1:9001/metrics

WebRTC metrics

Metrics prefixed with webrtc_ give you detailed insight about the WebRTC protocol for every streaming instance. See the official W3C reference for more information.
These metrics are available since Anbox Cloud 1.8.0.

Name Description
webrtc_frames_encoded Total number of frames successfully encoded
webrtc_key_frames_encoded Total number of key frames, such as key frames in VP8 or IDR-frames in H.264 (webrtc_key_frames_encoded - webrtc_frames_encoded gives the number of delta frames)
webrtc_total_encode_time Total number of seconds that has been spent encoding the webrtc_frames_encoded frames (the average encode time can be calculated by dividing this value with webrtc_frames_encoded)
webrtc_target_bitrate The current encoder target in bits per second
webrtc_bytes_sent Total number of bytes sent for a specific SSRC (a SSRC represents one resource - video, audio or binary data - sent over a WebRTC track)
webrtc_retransmitted_bytes_sent Total number of bytes that were re-transmitted for a specific SSRC, only including payload bytes
webrtc_retransmitted_packets_sent Total number of packets that were re-transmitted for a specific SSRC
webrtc_total_packet_send_delay Total number of seconds that packets have spent buffered locally before being transmitted onto the network
webrtc_packets_sent Total number of RTP packets sent for this SSRC (includes re-transmissions)
webrtc_nack_count Total number of Negative Acknowledgement (NACK) packets received by this sender
webrtc_fir_count Total number of Full Intra Request (FIR) packets received by this sender (video only)
webrtc_pli_count Total number of Picture Loss Indication (PLI) packets received by this sender (video only)
webrtc_sli_count Total number of Slice Loss Indication (SLI) packets received by this sender (video only)

Last updated 2 months ago.