Skip to content

Commit

Permalink
Network metrics: drop container_name label (#95)
Browse files Browse the repository at this point in the history 
I made a mistake in #81, incorrectly determining that network stats were
scoped to containers as well as network interfaces. They are not, in
spite of being laid out that way in the API response. And my tests
consisted only of ECS tasks with a single container.

(The stats for a single network interface _can_ vary slightly among
different containers within a single task stats API response, but I
believe that's only due to the ECS agent querying the stats from docker
for each container at slightly different times.)

So, there is a fully redundant label dimension right now. Remove it.

Signed-off-by: Ian Kerins <[email protected]>
  • Loading branch information
@isker
isker authored Jan 28, 2025
1 parent 057171b commit 61f6562
Show file tree
Hide file tree
Showing 2 changed files with 64 additions and 57 deletions.
58 changes: 29 additions & 29 deletions README.md
View file
Original file line number Diff line number Diff line change
Expand Up @@ -53,7 +53,7 @@ None.
```
# HELP ecs_container_cpu_usage_seconds_total Cumulative total container CPU usage in seconds.
# TYPE ecs_container_cpu_usage_seconds_total counter
ecs_container_cpu_usage_seconds_total{container_name="ecs-exporter"} 0.027095748000000003
ecs_container_cpu_usage_seconds_total{container_name="ecs-exporter"} 0.028057878
# HELP ecs_container_memory_limit_bytes Configured container memory limit in bytes, set from the container-level limit in the task definition if any, otherwise the task-level limit.
# TYPE ecs_container_memory_limit_bytes gauge
ecs_container_memory_limit_bytes{container_name="ecs-exporter"} 5.36870912e+08
Expand All @@ -62,34 +62,34 @@ ecs_container_memory_limit_bytes{container_name="ecs-exporter"} 5.36870912e+08
ecs_container_memory_page_cache_size_bytes{container_name="ecs-exporter"} 0
# HELP ecs_container_memory_usage_bytes Current container memory usage in bytes.
# TYPE ecs_container_memory_usage_bytes gauge
ecs_container_memory_usage_bytes{container_name="ecs-exporter"} 4.452352e+06
# HELP ecs_container_network_receive_bytes_total Cumulative total size of container network packets received in bytes.
# TYPE ecs_container_network_receive_bytes_total counter
ecs_container_network_receive_bytes_total{container_name="ecs-exporter",interface="eth1"} 1.1112267e+07
# HELP ecs_container_network_receive_errors_total Cumulative total count of container network errors in receiving.
# TYPE ecs_container_network_receive_errors_total counter
ecs_container_network_receive_errors_total{container_name="ecs-exporter",interface="eth1"} 0
# HELP ecs_container_network_receive_packets_dropped_total Cumulative total count of container network packets dropped in receiving.
# TYPE ecs_container_network_receive_packets_dropped_total counter
ecs_container_network_receive_packets_dropped_total{container_name="ecs-exporter",interface="eth1"} 0
# HELP ecs_container_network_receive_packets_total Cumulative total count of container network packets received.
# TYPE ecs_container_network_receive_packets_total counter
ecs_container_network_receive_packets_total{container_name="ecs-exporter",interface="eth1"} 8039
# HELP ecs_container_network_transmit_bytes_total Cumulative total size of container network packets transmitted in bytes.
# TYPE ecs_container_network_transmit_bytes_total counter
ecs_container_network_transmit_bytes_total{container_name="ecs-exporter",interface="eth1"} 165338
# HELP ecs_container_network_transmit_dropped_total Cumulative total count of container network packets dropped in transmit.
# TYPE ecs_container_network_transmit_dropped_total counter
ecs_container_network_transmit_dropped_total{container_name="ecs-exporter",interface="eth1"} 0
# HELP ecs_container_network_transmit_errors_total Cumulative total count of container network errors in transmit.
# TYPE ecs_container_network_transmit_errors_total counter
ecs_container_network_transmit_errors_total{container_name="ecs-exporter",interface="eth1"} 0
# HELP ecs_container_network_transmit_packets_total Cumulative total count of container network packets transmitted.
# TYPE ecs_container_network_transmit_packets_total counter
ecs_container_network_transmit_packets_total{container_name="ecs-exporter",interface="eth1"} 713
ecs_container_memory_usage_bytes{container_name="ecs-exporter"} 4.243456e+06
# HELP ecs_exporter_build_info A metric with a constant '1' value labeled by version, revision, branch, goversion from which ecs_exporter was built, and the goos and goarch for the build.
# TYPE ecs_exporter_build_info gauge
ecs_exporter_build_info{branch="",goarch="arm64",goos="linux",goversion="go1.23.2",revision="unknown",tags="unknown",version=""} 1
# HELP ecs_network_receive_bytes_total Cumulative total size of network packets received in bytes.
# TYPE ecs_network_receive_bytes_total counter
ecs_network_receive_bytes_total{interface="eth1"} 1.1172419e+07
# HELP ecs_network_receive_errors_total Cumulative total count of network errors in receiving.
# TYPE ecs_network_receive_errors_total counter
ecs_network_receive_errors_total{interface="eth1"} 0
# HELP ecs_network_receive_packets_dropped_total Cumulative total count of network packets dropped in receiving.
# TYPE ecs_network_receive_packets_dropped_total counter
ecs_network_receive_packets_dropped_total{interface="eth1"} 0
# HELP ecs_network_receive_packets_total Cumulative total count of network packets received.
# TYPE ecs_network_receive_packets_total counter
ecs_network_receive_packets_total{interface="eth1"} 8084
# HELP ecs_network_transmit_bytes_total Cumulative total size of network packets transmitted in bytes.
# TYPE ecs_network_transmit_bytes_total counter
ecs_network_transmit_bytes_total{interface="eth1"} 178817
# HELP ecs_network_transmit_dropped_total Cumulative total count of network packets dropped in transmit.
# TYPE ecs_network_transmit_dropped_total counter
ecs_network_transmit_dropped_total{interface="eth1"} 0
# HELP ecs_network_transmit_errors_total Cumulative total count of network errors in transmit.
# TYPE ecs_network_transmit_errors_total counter
ecs_network_transmit_errors_total{interface="eth1"} 0
# HELP ecs_network_transmit_packets_total Cumulative total count of network packets transmitted.
# TYPE ecs_network_transmit_packets_total counter
ecs_network_transmit_packets_total{interface="eth1"} 897
# HELP ecs_task_cpu_limit_vcpus Configured task CPU limit in vCPUs (1 vCPU = 1024 CPU units). This is optional when running on EC2; if no limit is set, this metric has no value.
# TYPE ecs_task_cpu_limit_vcpus gauge
ecs_task_cpu_limit_vcpus 0.25
Expand All @@ -101,16 +101,16 @@ ecs_task_ephemeral_storage_allocated_bytes 2.1491613696e+10
ecs_task_ephemeral_storage_used_bytes 3.7748736e+07
# HELP ecs_task_image_pull_start_timestamp_seconds The time at which the task started pulling docker images for its containers.
# TYPE ecs_task_image_pull_start_timestamp_seconds gauge
ecs_task_image_pull_start_timestamp_seconds 1.7291179014941156e+09
ecs_task_image_pull_start_timestamp_seconds 1.737156015124145e+09
# HELP ecs_task_image_pull_stop_timestamp_seconds The time at which the task stopped (i.e. completed) pulling docker images for its containers.
# TYPE ecs_task_image_pull_stop_timestamp_seconds gauge
ecs_task_image_pull_stop_timestamp_seconds 1.7291179144469e+09
ecs_task_image_pull_stop_timestamp_seconds 1.7371560172684324e+09
# HELP ecs_task_memory_limit_bytes Configured task memory limit in bytes. This is optional when running on EC2; if no limit is set, this metric has no value.
# TYPE ecs_task_memory_limit_bytes gauge
ecs_task_memory_limit_bytes 5.36870912e+08
# HELP ecs_task_metadata_info ECS task metadata, sourced from the task metadata endpoint version 4.
# TYPE ecs_task_metadata_info gauge
ecs_task_metadata_info{availability_zone="us-east-1a",cluster="arn:aws:ecs:us-east-1:829490980523:cluster/prom-ecs-exporter-sandbox",desired_status="RUNNING",family="prom-ecs-exporter-sandbox-isker-fargate",known_status="RUNNING",launch_type="FARGATE",revision="11",task_arn="arn:aws:ecs:us-east-1:829490980523:task/prom-ecs-exporter-sandbox/0c7f6b0414dc47d0a15019a099cd919b"} 1
ecs_task_metadata_info{availability_zone="us-east-1a",cluster="arn:aws:ecs:us-east-1:829490980523:cluster/prom-ecs-exporter-sandbox",desired_status="RUNNING",family="prom-ecs-exporter-sandbox-isker-fix-network-metrics-fargate",known_status="RUNNING",launch_type="FARGATE",revision="1",task_arn="arn:aws:ecs:us-east-1:829490980523:task/prom-ecs-exporter-sandbox/c8387acdc4884a0fa13dae78e68a989f"} 1
```

## Example task definition
Expand Down
63 changes: 35 additions & 28 deletions ecscollector/collector.go
View file
Original file line number Diff line number Diff line change
Expand Up @@ -93,44 +93,44 @@ var (
containerLabels, nil)

networkRxBytesDesc = prometheus.NewDesc(
"ecs_container_network_receive_bytes_total",
"Cumulative total size of container network packets received in bytes.",
containerNetworkLabels, nil)
"ecs_network_receive_bytes_total",
"Cumulative total size of network packets received in bytes.",
networkLabels, nil)

networkRxPacketsDesc = prometheus.NewDesc(
"ecs_container_network_receive_packets_total",
"Cumulative total count of container network packets received.",
containerNetworkLabels, nil)
"ecs_network_receive_packets_total",
"Cumulative total count of network packets received.",
networkLabels, nil)

networkRxDroppedDesc = prometheus.NewDesc(
"ecs_container_network_receive_packets_dropped_total",
"Cumulative total count of container network packets dropped in receiving.",
containerNetworkLabels, nil)
"ecs_network_receive_packets_dropped_total",
"Cumulative total count of network packets dropped in receiving.",
networkLabels, nil)

networkRxErrorsDesc = prometheus.NewDesc(
"ecs_container_network_receive_errors_total",
"Cumulative total count of container network errors in receiving.",
containerNetworkLabels, nil)
"ecs_network_receive_errors_total",
"Cumulative total count of network errors in receiving.",
networkLabels, nil)

networkTxBytesDesc = prometheus.NewDesc(
"ecs_container_network_transmit_bytes_total",
"Cumulative total size of container network packets transmitted in bytes.",
containerNetworkLabels, nil)
"ecs_network_transmit_bytes_total",
"Cumulative total size of network packets transmitted in bytes.",
networkLabels, nil)

networkTxPacketsDesc = prometheus.NewDesc(
"ecs_container_network_transmit_packets_total",
"Cumulative total count of container network packets transmitted.",
containerNetworkLabels, nil)
"ecs_network_transmit_packets_total",
"Cumulative total count of network packets transmitted.",
networkLabels, nil)

networkTxDroppedDesc = prometheus.NewDesc(
"ecs_container_network_transmit_dropped_total",
"Cumulative total count of container network packets dropped in transmit.",
containerNetworkLabels, nil)
"ecs_network_transmit_dropped_total",
"Cumulative total count of network packets dropped in transmit.",
networkLabels, nil)

networkTxErrorsDesc = prometheus.NewDesc(
"ecs_container_network_transmit_errors_total",
"Cumulative total count of container network errors in transmit.",
containerNetworkLabels, nil)
"ecs_network_transmit_errors_total",
"Cumulative total count of network errors in transmit.",
networkLabels, nil)
)

var containerLabels = []string{
Expand All @@ -150,10 +150,9 @@ var taskMetadataLabels = []string{
"launch_type",
}

var containerNetworkLabels = append(
containerLabels,
var networkLabels = []string{
"interface",
)
}

// NewCollector returns a new Collector that queries ECS metadata server
// for ECS task and container metrics.
Expand Down Expand Up @@ -324,7 +323,15 @@ func (c *collector) Collect(ch chan<- prometheus.Metric) {

// Network metrics per interface.
for iface, netStats := range s.Networks {
networkLabelVals := append(containerLabelVals, iface)
// While the API response attaches network stats to each container,
// the container is in fact not a relevant dimension; only the
// interface is. This means that if multiple containers use the same
// network (extremely likely), we are redundantly writing this
// metric with "last one wins" semantics. This is fine: the values
// for an interface are the same across all containers.
networkLabelVals := []string{
iface,
}

for desc, value := range map[*prometheus.Desc]float64{
networkRxBytesDesc: float64(netStats.RxBytes),
Expand Down

0 comments on commit 61f6562

Please sign in to comment.