-# Examples: -#
-# "PT20.345S" -- parses as "20.345 seconds" -# "PT15M" -- parses as "15 minutes" (where a minute is 60 seconds) -# "PT10H" -- parses as "10 hours" (where an hour is 3600 seconds) -# "P2D" -- parses as "2 days" (where a day is 24 hours or 86400 seconds) -# "P2DT3H4M" -- parses as "2 days, 3 hours and 4 minutes" -# "P-6H3M" -- parses as "-6 hours and +3 minutes" -# "-P6H3M" -- parses as "-6 hours and -3 minutes" -# "-P-6H+3M" -- parses as "+6 hours and -3 minutes" -#-filter: "{ tags -> tags.job_name == 'vm-monitoring' }" # The OpenTelemetry job name -expSuffix: service(['node_identifier_host_name'] , Layer.OS_LINUX) +# +# Linux infrastructure MAL. +# +# This rule intentionally accepts TWO source metric families under the same +# existing job_name "vm-monitoring": +# 1. node-exporter metrics (existing SkyWalking integration) +# 2. OpenTelemetry hostmetrics metrics (new alternative source) +# +# Both sources are mapped by ONE MAL file to the canonical meter_vm_* targets. +# This avoids cross-file metric ownership collisions in OAP 11 runtime rules. +# +# MAL SampleFamily '+' treats an EMPTY family as identity. Because a host sends +# one source family or the other, each expression can safely add the equivalent +# node-exporter and hostmetrics branches. The output metric is registered once. +filter: "{ tags -> tags.job_name == 'vm-monitoring' }" metricPrefix: meter_vm + metricsRules: + # --------------------------------------------------------------------------- + # Canonical metrics: same target names and semantics for both sources. + # --------------------------------------------------------------------------- - #node cpu + # CPU - name: cpu_total_percentage - exp: (node_cpu_seconds_total * 100).tagNotEqual('mode' , 'idle').sum(['node_identifier_host_name']).rate('PT1M') + exp: > + ( + (node_cpu_seconds_total * 100) + .tagNotEqual('mode','idle') + .sum(['node_identifier_host_name']) + .rate('PT1M') + + + (system_cpu_utilization * 100) + .tagNotEqual('mode','idle') + .sum(['node_identifier_host_name']) + ).service(['node_identifier_host_name'], Layer.OS_LINUX) + - name: cpu_average_used - exp: (node_cpu_seconds_total * 100).sum(['node_identifier_host_name' , 'mode']).rate('PT1M') + exp: > + ( + (node_cpu_seconds_total * 100) + .sum(['node_identifier_host_name','mode']) + .rate('PT1M') + + + (system_cpu_utilization * 100) + .sum(['node_identifier_host_name','mode']) + ).service(['node_identifier_host_name'], Layer.OS_LINUX) + - name: cpu_load1 - exp: node_load1 * 100 + exp: "((node_load1 * 100) + (system_cpu_load_average_1m * 100)).service(['node_identifier_host_name'], Layer.OS_LINUX).decorate({ me -> me.attr0 = me.layer.name()})" + - name: cpu_load5 - exp: node_load5 * 100 + exp: "((node_load5 * 100) + (system_cpu_load_average_5m * 100)).service(['node_identifier_host_name'], Layer.OS_LINUX)" + - name: cpu_load15 - exp: node_load15 * 100 + exp: "((node_load15 * 100) + (system_cpu_load_average_15m * 100)).service(['node_identifier_host_name'], Layer.OS_LINUX)" - #node Memory + # Physical memory - name: memory_total - exp: node_memory_MemTotal_bytes + exp: "(node_memory_MemTotal_bytes + system_memory_limit).service(['node_identifier_host_name'], Layer.OS_LINUX)" + - name: memory_available - exp: node_memory_MemAvailable_bytes + exp: "(node_memory_MemAvailable_bytes + system_linux_memory_available).service(['node_identifier_host_name'], Layer.OS_LINUX)" + - name: memory_used - exp: node_memory_MemTotal_bytes - node_memory_MemAvailable_bytes + exp: > + ( + (node_memory_MemTotal_bytes - node_memory_MemAvailable_bytes) + + + (system_memory_limit - system_linux_memory_available) + ).service(['node_identifier_host_name'], Layer.OS_LINUX) + - name: memory_buff_cache - exp: node_memory_Buffers_bytes + node_memory_Cached_bytes + exp: > + ( + (node_memory_Buffers_bytes + node_memory_Cached_bytes) + + + ( + system_memory_usage.tagEqual('state','buffered').sum(['node_identifier_host_name']) + + + system_memory_usage.tagEqual('state','cached').sum(['node_identifier_host_name']) + - + system_memory_usage.tagEqual('state','slab_reclaimable').sum(['node_identifier_host_name']) + ) + ).service(['node_identifier_host_name'], Layer.OS_LINUX) + + # Swap - name: memory_swap_free - exp: node_memory_SwapFree_bytes + exp: > + ( + node_memory_SwapFree_bytes + + + system_paging_usage + .tagEqual('state','free') + .sum(['node_identifier_host_name']) + ).service(['node_identifier_host_name'], Layer.OS_LINUX) + - name: memory_swap_total - exp: node_memory_SwapTotal_bytes + exp: > + ( + node_memory_SwapTotal_bytes + + + ( + system_paging_usage.tagEqual('state','used').sum(['node_identifier_host_name']) + + + system_paging_usage.tagEqual('state','free').sum(['node_identifier_host_name']) + ) + ).service(['node_identifier_host_name'], Layer.OS_LINUX) + - name: memory_swap_percentage - exp: 100 - ((node_memory_SwapFree_bytes * 100) / node_memory_SwapTotal_bytes) + exp: > + ( + ( + 100 - ((node_memory_SwapFree_bytes * 100) / node_memory_SwapTotal_bytes) + ) + + + ( + ( + system_paging_usage.tagEqual('state','used').sum(['node_identifier_host_name']) + * 100 + ) + / + ( + system_paging_usage.tagEqual('state','used').sum(['node_identifier_host_name']) + + + system_paging_usage.tagEqual('state','free').sum(['node_identifier_host_name']) + ) + ) + ).service(['node_identifier_host_name'], Layer.OS_LINUX) - #node filesystem + # Filesystem. hostmetrics free+reserved+used is the full capacity. - name: filesystem_percentage - exp: 100 - ((node_filesystem_avail_bytes * 100).sum(['node_identifier_host_name' , 'mountpoint']) / node_filesystem_size_bytes.sum(['node_identifier_host_name' , 'mountpoint'])) + exp: > + ( + ( + 100 - ( + (node_filesystem_avail_bytes * 100) + .sum(['node_identifier_host_name','mountpoint']) + / + node_filesystem_size_bytes + .sum(['node_identifier_host_name','mountpoint']) + ) + ) + + + ( + 100 - ( + (system_filesystem_usage.tagEqual('state','free') * 100) + .sum(['node_identifier_host_name','mountpoint']) + / + ( + system_filesystem_usage.tagEqual('state','free') + .sum(['node_identifier_host_name','mountpoint']) + + + system_filesystem_usage.tagEqual('state','reserved') + .sum(['node_identifier_host_name','mountpoint']) + + + system_filesystem_usage.tagEqual('state','used') + .sum(['node_identifier_host_name','mountpoint']) + ) + ) + ) + ).service(['node_identifier_host_name'], Layer.OS_LINUX) - #node disk + # Disk throughput - name: disk_read - exp: node_disk_read_bytes_total.sum(['node_identifier_host_name']).rate('PT1M') + exp: > + ( + node_disk_read_bytes_total + .sum(['node_identifier_host_name']) + .rate('PT1M') + + + system_disk_io + .tagEqual('direction','read') + .sum(['node_identifier_host_name']) + .rate('PT1M') + ).service(['node_identifier_host_name'], Layer.OS_LINUX) + - name: disk_written - exp: node_disk_written_bytes_total.sum(['node_identifier_host_name']).rate('PT1M') + exp: > + ( + node_disk_written_bytes_total + .sum(['node_identifier_host_name']) + .rate('PT1M') + + + system_disk_io + .tagEqual('direction','write') + .sum(['node_identifier_host_name']) + .rate('PT1M') + ).service(['node_identifier_host_name'], Layer.OS_LINUX) - #node network + # Network throughput - name: network_receive - exp: node_network_receive_bytes_total.sum(['node_identifier_host_name']).irate() + exp: > + ( + node_network_receive_bytes_total + .sum(['node_identifier_host_name']) + .irate() + + + system_network_io + .tagEqual('direction','receive') + .sum(['node_identifier_host_name']) + .irate() + ).service(['node_identifier_host_name'], Layer.OS_LINUX) + - name: network_transmit - exp: node_network_transmit_bytes_total.sum(['node_identifier_host_name']).irate() + exp: > + ( + node_network_transmit_bytes_total + .sum(['node_identifier_host_name']) + .irate() + + + system_network_io + .tagEqual('direction','transmit') + .sum(['node_identifier_host_name']) + .irate() + ).service(['node_identifier_host_name'], Layer.OS_LINUX) - #node netstat + # TCP connection metrics with true hostmetrics equivalents. + # Collector normalizes system.network.connections state values to lowercase. - name: tcp_curr_estab - exp: node_netstat_Tcp_CurrEstab + exp: > + ( + node_netstat_Tcp_CurrEstab + + + ( + system_network_connections + .tagEqual('protocol','tcp') + .tagEqual('state','established') + .sum(['node_identifier_host_name']) + + + system_network_connections + .tagEqual('protocol','tcp') + .tagEqual('state','close_wait') + .sum(['node_identifier_host_name']) + ) + ).service(['node_identifier_host_name'], Layer.OS_LINUX) + - name: tcp_tw - exp: node_sockstat_TCP_tw + exp: > + ( + node_sockstat_TCP_tw + + + system_network_connections + .tagEqual('protocol','tcp') + .tagEqual('state','time_wait') + .sum(['node_identifier_host_name']) + ).service(['node_identifier_host_name'], Layer.OS_LINUX) + + # --------------------------------------------------------------------------- + # Existing node-exporter-only canonical metrics. + # hostmetrics has no source with equivalent semantics, so these remain node + # exporter only rather than being populated with a misleading approximation. + # --------------------------------------------------------------------------- - name: tcp_alloc - exp: node_sockstat_TCP_alloc + exp: (node_sockstat_TCP_alloc).service(['node_identifier_host_name'], Layer.OS_LINUX) + - name: sockets_used - exp: node_sockstat_sockets_used + exp: (node_sockstat_sockets_used).service(['node_identifier_host_name'], Layer.OS_LINUX) + - name: udp_inuse - exp: node_sockstat_UDP_inuse + exp: (node_sockstat_UDP_inuse).service(['node_identifier_host_name'], Layer.OS_LINUX) - #node filefd - name: filefd_allocated - exp: node_filefd_allocated - + exp: (node_filefd_allocated).service(['node_identifier_host_name'], Layer.OS_LINUX) + # --------------------------------------------------------------------------- + # New hostmetrics-only capabilities. New names are used whenever semantics + # do not exactly match an existing node-exporter target metric. + # --------------------------------------------------------------------------- + - name: cpu_cores_num + exp: (system_cpu_logical_count.sum(['node_identifier_host_name'])).service(['node_identifier_host_name'], Layer.OS_LINUX) + - name: cpu_norm_percentage + exp: > + ( + ( + (system_cpu_utilization * 100) + .tagNotEqual('mode','idle') + .sum(['node_identifier_host_name']) + ) + / + system_cpu_logical_count.sum(['node_identifier_host_name']) + ).service(['node_identifier_host_name'], Layer.OS_LINUX) diff --git a/oap-server/server-starter/src/main/resources/otel-rules/windows.yaml b/oap-server/server-starter/src/main/resources/otel-rules/windows.yaml index de2a651111d5..2e872e86659a 100644 --- a/oap-server/server-starter/src/main/resources/otel-rules/windows.yaml +++ b/oap-server/server-starter/src/main/resources/otel-rules/windows.yaml @@ -13,60 +13,213 @@ # See the License for the specific language governing permissions and # limitations under the License. -# This will parse a textual representation of a duration. The formats -# accepted are based on the ISO-8601 duration format {@code PnDTnHnMn.nS} -# with days considered to be exactly 24 hours. -#
-# Examples: -#
-# "PT20.345S" -- parses as "20.345 seconds" -# "PT15M" -- parses as "15 minutes" (where a minute is 60 seconds) -# "PT10H" -- parses as "10 hours" (where an hour is 3600 seconds) -# "P2D" -- parses as "2 days" (where a day is 24 hours or 86400 seconds) -# "P2DT3H4M" -- parses as "2 days, 3 hours and 4 minutes" -# "P-6H3M" -- parses as "-6 hours and +3 minutes" -# "-P6H3M" -- parses as "-6 hours and -3 minutes" -# "-P-6H+3M" -- parses as "+6 hours and -3 minutes" -#-filter: "{ tags -> tags.job_name == 'windows-monitoring' }" # The OpenTelemetry job name -expSuffix: service(['node_identifier_host_name'] , Layer.OS_WINDOWS) +# Windows infrastructure MAL. +# +# The existing windows-exporter source and the OpenTelemetry hostmetrics source +# intentionally use the same existing job_name "windows-monitoring". One MAL +# file maps either input family to the canonical meter_win_* targets, avoiding +# cross-file metric ownership collisions in OAP 11. +filter: "{ tags -> tags.job_name == 'windows-monitoring' }" +expSuffix: service(['node_identifier_host_name'], Layer.OS_WINDOWS) metricPrefix: meter_win + metricsRules: - #cpu windows don't expose cpu load metrics + # --------------------------------------------------------------------------- + # Canonical windows.yaml metrics, now accepting either source. + # --------------------------------------------------------------------------- + + # CPU - name: cpu_total_percentage - exp: (windows_cpu_time_total * 100).tagNotEqual('mode' , 'idle').sum(['node_identifier_host_name']).rate('PT1M') + exp: > + (windows_cpu_time_total * 100) + .tagNotEqual('mode','idle') + .sum(['node_identifier_host_name']) + .rate('PT1M') + + + (system_cpu_time * 100) + .tagNotEqual('mode','idle') + .sum(['node_identifier_host_name']) + .rate('PT1M') + - name: cpu_average_used - exp: (windows_cpu_time_total * 100).sum(['node_identifier_host_name' , 'mode']).rate('PT1M') + exp: > + (windows_cpu_time_total * 100) + .sum(['node_identifier_host_name','mode']) + .rate('PT1M') + + + (system_cpu_time * 100) + .sum(['node_identifier_host_name','mode']) + .rate('PT1M') - #memory + # Physical memory - name: memory_total - exp: windows_cs_physical_memory_bytes + exp: > + windows_cs_physical_memory_bytes + + + system_memory_limit.sum(['node_identifier_host_name']) + - name: memory_available - exp: windows_os_physical_memory_free_bytes + exp: > + windows_os_physical_memory_free_bytes + + + system_memory_usage + .tagEqual('state','free') + .sum(['node_identifier_host_name']) + - name: memory_used - exp: windows_cs_physical_memory_bytes - windows_os_physical_memory_free_bytes + exp: > + (windows_cs_physical_memory_bytes - windows_os_physical_memory_free_bytes) + + + ( + system_memory_limit.sum(['node_identifier_host_name']) + - + system_memory_usage.tagEqual('state','free').sum(['node_identifier_host_name']) + ) + + # Canonical virtual memory semantics. + # The hostmetrics branch uses Windows Performance Counters: + # Memory\\Commit Limit + # Memory\\Committed Bytes - name: memory_virtual_memory_free - exp: windows_os_virtual_memory_free_bytes + exp: > + windows_os_virtual_memory_free_bytes + + + ( + system_virtual_memory_commit_limit.sum(['node_identifier_host_name']) + - + system_virtual_memory_committed.sum(['node_identifier_host_name']) + ) + - name: memory_virtual_memory_total - exp: windows_os_virtual_memory_bytes + exp: > + windows_os_virtual_memory_bytes + + + system_virtual_memory_commit_limit.sum(['node_identifier_host_name']) + - name: memory_virtual_memory_percentage - exp: 100 - ((windows_os_virtual_memory_free_bytes * 100) / windows_os_virtual_memory_bytes) + exp: > + ( + 100 - ((windows_os_virtual_memory_free_bytes * 100) / windows_os_virtual_memory_bytes) + ) + + + ( + system_virtual_memory_committed.sum(['node_identifier_host_name']) * 100 + / + system_virtual_memory_commit_limit.sum(['node_identifier_host_name']) + ) - #disk + # Disk throughput - name: disk_read - exp: windows_logical_disk_read_bytes_total.sum(['node_identifier_host_name']).rate('PT1M') + exp: > + windows_logical_disk_read_bytes_total + .sum(['node_identifier_host_name']) + .rate('PT1M') + + + system_disk_io + .tagEqual('direction','read') + .sum(['node_identifier_host_name']) + .rate('PT1M') + - name: disk_written - exp: windows_logical_disk_write_bytes_total.sum(['node_identifier_host_name']).rate('PT1M') + exp: > + windows_logical_disk_write_bytes_total + .sum(['node_identifier_host_name']) + .rate('PT1M') + + + system_disk_io + .tagEqual('direction','write') + .sum(['node_identifier_host_name']) + .rate('PT1M') - #network + # Network throughput - name: network_receive - exp: windows_net_bytes_received_total.sum(['node_identifier_host_name']).irate() + exp: > + windows_net_bytes_received_total + .sum(['node_identifier_host_name']) + .irate() + + + system_network_io + .tagEqual('direction','receive') + .sum(['node_identifier_host_name']) + .irate() + - name: network_transmit - exp: windows_net_bytes_sent_total.sum(['node_identifier_host_name']).irate() - - - - + exp: > + windows_net_bytes_sent_total + .sum(['node_identifier_host_name']) + .irate() + + + system_network_io + .tagEqual('direction','transmit') + .sum(['node_identifier_host_name']) + .irate() + + # --------------------------------------------------------------------------- + # Additional hostmetrics capabilities. These are additive and do not change + # the existing windows-exporter metric contract. + # --------------------------------------------------------------------------- + - name: cpu_cores_num + exp: system_cpu_logical_count.sum(['node_identifier_host_name']) + + - name: cpu_norm_percentage + exp: > + ( + (system_cpu_time * 100) + .tagNotEqual('mode','idle') + .sum(['node_identifier_host_name']) + .rate('PT1M') + ) + / + system_cpu_logical_count.sum(['node_identifier_host_name']) + + - name: cpu_load1 + exp: system_cpu_load_average_1m.sum(['node_identifier_host_name']) * 100 + + - name: cpu_load5 + exp: system_cpu_load_average_5m.sum(['node_identifier_host_name']) * 100 + + - name: cpu_load15 + exp: system_cpu_load_average_15m.sum(['node_identifier_host_name']) * 100 + + - name: filesystem_percentage + exp: > + 100 - ( + (system_filesystem_usage.tagEqual('state','free') * 100) + .sum(['node_identifier_host_name','mountpoint']) + / + ( + system_filesystem_usage.tagEqual('state','free') + .sum(['node_identifier_host_name','mountpoint']) + + + system_filesystem_usage.tagEqual('state','reserved') + .sum(['node_identifier_host_name','mountpoint']) + + + system_filesystem_usage.tagEqual('state','used') + .sum(['node_identifier_host_name','mountpoint']) + ) + ) + + - name: filehandles_allocated + exp: system_filehandles_allocated.sum(['node_identifier_host_name']) + # system.paging.usage is the Windows pagefile, not the windows.yaml virtual + # memory metric. Keep it under explicit new names. + - name: memory_pagefile_free + exp: > + system_paging_usage + .tagEqual('state','free') + .sum(['node_identifier_host_name']) + - name: memory_pagefile_total + exp: system_paging_usage.sum(['node_identifier_host_name']) + - name: memory_pagefile_percentage + exp: > + ( + system_paging_usage + .tagEqual('state','used') + .sum(['node_identifier_host_name']) + * 100 + ) + / + system_paging_usage.sum(['node_identifier_host_name']) diff --git a/test/e2e-v2/cases/vm/otel-hostmetrics/Dockerfile.otelcol b/test/e2e-v2/cases/vm/otel-hostmetrics/Dockerfile.otelcol new file mode 100644 index 000000000000..e13d04d7a708 --- /dev/null +++ b/test/e2e-v2/cases/vm/otel-hostmetrics/Dockerfile.otelcol @@ -0,0 +1,27 @@ +# Licensed to the Apache Software Foundation (ASF) under one or more +# contributor license agreements. See the NOTICE file distributed with +# this work for additional information regarding copyright ownership. +# The ASF licenses this file to You under the Apache License, Version 2.0 +# (the "License"); you may not use this file except in compliance with +# the License. You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +ARG OTEL_COLLECTOR_CONTRIB_VERSION + +FROM otel/opentelemetry-collector-contrib:${OTEL_COLLECTOR_CONTRIB_VERSION} AS otel + +FROM alpine:3.22 + +COPY --from=otel /otelcol-contrib /otelcol-contrib +COPY entrypoint.sh /entrypoint.sh + +RUN chmod +x /entrypoint.sh + +ENTRYPOINT ["/entrypoint.sh"] diff --git a/test/e2e-v2/cases/vm/otel-hostmetrics/assert-collector-output.sh b/test/e2e-v2/cases/vm/otel-hostmetrics/assert-collector-output.sh new file mode 100755 index 000000000000..f3e01de4d6d1 --- /dev/null +++ b/test/e2e-v2/cases/vm/otel-hostmetrics/assert-collector-output.sh @@ -0,0 +1,147 @@ +#!/usr/bin/env bash +# Licensed to the Apache Software Foundation (ASF) under one or more +# contributor license agreements. See the NOTICE file distributed with +# this work for additional information regarding copyright ownership. +# The ASF licenses this file to You under the Apache License, Version 2.0 +# (the "License"); you may not use this file except in compliance with +# the License. You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. +set -euo pipefail + +CID="$( + docker ps \ + --filter 'label=com.docker.compose.service=vm-service' \ + --filter 'status=running' \ + --format '{{.ID}}' \ + | head -n 1 +)" + +if [ -z "${CID}" ]; then + echo "vm-service container not found" >&2 + exit 1 +fi + +TMPDIR="$(mktemp -d)" +trap 'rm -rf "${TMPDIR}"' EXIT + +docker exec "${CID}" cat /tmp/process-linux-raw.log \ + > "${TMPDIR}/process-linux.log" + +docker exec "${CID}" cat /tmp/windows-host-raw.log \ + > "${TMPDIR}/windows-host.log" + +docker exec "${CID}" cat /tmp/windows-process-raw.log \ + > "${TMPDIR}/windows-process.log" + + +# --------------------------------------------------------------------------- +# Linux process pre-OTLP contract. +# +# For every observed process.count metric block we require the Collector-side +# result to contain one datapoint with value 3 and the corrected {process} unit. +# A raw value 1 would reproduce the reviewer-reported regression. +# --------------------------------------------------------------------------- +awk ' +function finish_block() { + if (!in_count) { + return + } + + if (points == 1 && value3 && unit_process) { + good = 1 + } + + if (value1) { + bad = 1 + } + + in_count = 0 +} + +/-> Name: process.count[[:space:]]*$/ { + finish_block() + in_count = 1 + points = 0 + value3 = 0 + value1 = 0 + unit_process = 0 + next +} + +in_count && /^[[:space:]]*Metric #[0-9]+/ { + finish_block() + next +} + +in_count && /NumberDataPoints #[0-9]+/ { + points++ +} + +in_count && /-> Unit: \{process\}/ { + unit_process = 1 +} + +in_count && /Value:[[:space:]]+3(\.0+)?[[:space:]]*$/ { + value3 = 1 +} + +in_count && /Value:[[:space:]]+1(\.0+)?[[:space:]]*$/ { + value1 = 1 +} + +END { + finish_block() + exit(good && !bad ? 0 : 1) +} +' "${TMPDIR}/process-linux.log" + + +# --------------------------------------------------------------------------- +# Windows host normalization: +# native system.cpu.time state -> canonical mode. +# --------------------------------------------------------------------------- +grep -q -- '-> Name: system.cpu.time' \ + "${TMPDIR}/windows-host.log" + +grep -q -- '-> mode: Str(idle)' \ + "${TMPDIR}/windows-host.log" + +grep -q -- '-> mode: Str(user)' \ + "${TMPDIR}/windows-host.log" + +if grep -q -- '-> state:' "${TMPDIR}/windows-host.log"; then + echo "Windows system.cpu.time still contains state attribute" >&2 + exit 1 +fi + + +# --------------------------------------------------------------------------- +# Windows process normalization: +# native process.handles -> canonical process.open_handles. +# --------------------------------------------------------------------------- +grep -q -- '-> Name: process.open_handles' \ + "${TMPDIR}/windows-process.log" + +if grep -q -- '-> Name: process.handles[[:space:]]*$' \ + "${TMPDIR}/windows-process.log"; then + echo "Native process.handles survived normalization" >&2 + exit 1 +fi + + +cat <<'YAML' +collector_process_count: + datapoints_per_export: 1 + value: 3 + unit: "{process}" +windows_normalization: + cpu_state_to_mode: true + process_handles_to_open_handles: true +YAML diff --git a/test/e2e-v2/cases/vm/otel-hostmetrics/docker-compose.yml b/test/e2e-v2/cases/vm/otel-hostmetrics/docker-compose.yml new file mode 100644 index 000000000000..88168a2965f7 --- /dev/null +++ b/test/e2e-v2/cases/vm/otel-hostmetrics/docker-compose.yml @@ -0,0 +1,84 @@ +# Licensed to the Apache Software Foundation (ASF) under one or more +# contributor license agreements. See the NOTICE file distributed with +# this work for additional information regarding copyright ownership. +# The ASF licenses this file to You under the Apache License, Version 2.0 +# (the "License"); you may not use this file except in compliance with +# the License. You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. +services: + oap: + extends: + file: ../../../script/docker-compose/base-compose.yml + service: oap + environment: + SW_OTEL_RECEIVER: default + volumes: + - ./otel-rules/vm.yaml:/skywalking/config/otel-rules/vm.yaml + - ./otel-rules/process-hostmetrics-linux.yaml:/skywalking/config/otel-rules/process-hostmetrics-linux.yaml + ports: + - 12800 + banyandb: + extends: + file: ../../../script/docker-compose/base-compose.yml + service: banyandb + ports: + - 17912 + vm-service: + hostname: vm-service + build: + context: . + dockerfile: Dockerfile.otelcol + args: + OTEL_COLLECTOR_CONTRIB_VERSION: ${OTEL_COLLECTOR_CONTRIB_VERSION} + networks: + - e2e + volumes: + - ./otel-collector-config.yaml:/etc/otelcol-contrib/config.yaml + - ./fixtures:/fixtures:ro + depends_on: + oap: + condition: service_healthy + windows-fixture-replay: + image: curlimages/curl:8.10.1 + depends_on: + - vm-service + volumes: + - ./fixtures:/fixtures:ro + networks: + - e2e + command: + - sh + - -c + - | + while true; do + TS="$$(date +%s)000000000" + echo "Replaying Windows OTLP fixtures timestamp=$${TS}" + sed "s/1788474000000000000/$${TS}/g" \ + /fixtures/windows-host.json \ + | curl -sS \ + -o /dev/null \ + -w "windows-host HTTP %{http_code}\n" \ + -X POST \ + -H 'Content-Type: application/json' \ + --data-binary @- \ + http://vm-service:4318/v1/metrics || true + sed "s/1788474000000000000/$${TS}/g" \ + /fixtures/windows-process.json \ + | curl -sS \ + -o /dev/null \ + -w "windows-process HTTP %{http_code}\n" \ + -X POST \ + -H 'Content-Type: application/json' \ + --data-binary @- \ + http://vm-service:4318/v1/metrics || true + sleep 10 + done +networks: + e2e: diff --git a/test/e2e-v2/cases/vm/otel-hostmetrics/e2e.yaml b/test/e2e-v2/cases/vm/otel-hostmetrics/e2e.yaml new file mode 100644 index 000000000000..bb3c28839b8f --- /dev/null +++ b/test/e2e-v2/cases/vm/otel-hostmetrics/e2e.yaml @@ -0,0 +1,82 @@ +# Licensed to the Apache Software Foundation (ASF) under one or more +# contributor license agreements. See the NOTICE file distributed with +# this work for additional information regarding copyright ownership. +# The ASF licenses this file to You under the Apache License, Version 2.0 +# (the "License"); you may not use this file except in compliance with +# the License. You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +setup: + env: compose + file: docker-compose.yml + timeout: 20m + init-system-environment: ../../../script/env + steps: + - name: set PATH + command: export PATH=/tmp/skywalking-infra-e2e/bin:$PATH + + - name: install yq + command: bash test/e2e-v2/script/prepare/setup-e2e-shell/install.sh yq + + - name: install swctl + command: bash test/e2e-v2/script/prepare/setup-e2e-shell/install.sh swctl + +verify: + retry: + count: 30 + interval: 3s + + cases: + # Existing shared VM assertions: + # service exists and canonical memory metrics have values. + - includes: + - ../vm-cases.yaml + + # Canonical metrics produced from native OTel hostmetrics. + - query: swctl --display yaml --base-url=http://${oap_host}:${oap_12800}/graphql metrics exec --expression=meter_vm_cpu_total_percentage --service-name=vm-service + expected: ../expected/metrics-has-value.yml + + - query: swctl --display yaml --base-url=http://${oap_host}:${oap_12800}/graphql metrics exec --expression=meter_vm_cpu_cores_num --service-name=vm-service + expected: ../expected/metrics-has-value.yml + + - query: swctl --display yaml --base-url=http://${oap_host}:${oap_12800}/graphql metrics exec --expression=meter_vm_cpu_norm_percentage --service-name=vm-service + expected: ../expected/metrics-has-value.yml + + # The normalized process name must be usable as a SkyWalking instance. + - query: swctl --display yaml --base-url=http://${oap_host}:${oap_12800}/graphql metrics exec --expression=mp_process_linux_num_procs --service-name=vm-service --instance-name=sleep + expected: expected/process-count-3.yml + + - query: swctl --display yaml --base-url=http://${oap_host}:${oap_12800}/graphql metrics exec --expression=mp_process_linux_num_threads --service-name=vm-service --instance-name=sleep + expected: ../expected/metrics-has-value.yml + + - query: swctl --display yaml --base-url=http://${oap_host}:${oap_12800}/graphql metrics exec --expression=mp_process_linux_memory_resident_bytes --service-name=vm-service --instance-name=sleep + expected: ../expected/metrics-has-value.yml + + - query: swctl --display yaml --base-url=http://${oap_host}:${oap_12800}/graphql metrics exec --expression=mp_process_linux_open_handles --service-name=vm-service --instance-name=sleep + expected: ../expected/metrics-has-value.yml + + - query: swctl --display yaml --base-url=http://${oap_host}:${oap_12800}/graphql metrics exec --expression=mp_process_linux_oldest_process_uptime_seconds --service-name=vm-service --instance-name=sleep + expected: ../expected/metrics-has-value.yml + + # Raw post-processor Collector contract. This distinguishes true + # Collector-side aggregation from a value that only becomes 3 in MAL. + - query: bash test/e2e-v2/cases/vm/otel-hostmetrics/assert-collector-output.sh + expected: expected/collector-contract.yml + + # Synthetic native-Windows OTLP fixtures are normalized by Collector 0.158 + # and then consumed by the production Windows MAL rules. + - query: swctl --display yaml --base-url=http://${oap_host}:${oap_12800}/graphql metrics exec --expression=meter_win_cpu_cores_num --service-name=win-fixture + expected: ../expected/metrics-has-value.yml + + - query: swctl --display yaml --base-url=http://${oap_host}:${oap_12800}/graphql metrics exec --expression=mp_process_windows_num_procs --service-name=win-fixture --instance-name=svchost.exe + expected: ../expected/metrics-has-value.yml + + - query: swctl --display yaml --base-url=http://${oap_host}:${oap_12800}/graphql metrics exec --expression=mp_process_windows_open_handles --service-name=win-fixture --instance-name=svchost.exe + expected: ../expected/metrics-has-value.yml diff --git a/test/e2e-v2/cases/vm/otel-hostmetrics/entrypoint.sh b/test/e2e-v2/cases/vm/otel-hostmetrics/entrypoint.sh new file mode 100644 index 000000000000..004c642e9b68 --- /dev/null +++ b/test/e2e-v2/cases/vm/otel-hostmetrics/entrypoint.sh @@ -0,0 +1,25 @@ +#!/bin/sh + +# Licensed to the Apache Software Foundation (ASF) under one or more +# contributor license agreements. See the NOTICE file distributed with +# this work for additional information regarding copyright ownership. +# The ASF licenses this file to You under the Apache License, Version 2.0 +# (the "License"); you may not use this file except in compliance with +# the License. You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +set -eu + +sleep 3600 & +sleep 3600 & +sleep 3600 & + +exec /otelcol-contrib \ + --config=/etc/otelcol-contrib/config.yaml diff --git a/test/e2e-v2/cases/vm/otel-hostmetrics/expected/collector-contract.yml b/test/e2e-v2/cases/vm/otel-hostmetrics/expected/collector-contract.yml new file mode 100644 index 000000000000..f7c4c0bdbd14 --- /dev/null +++ b/test/e2e-v2/cases/vm/otel-hostmetrics/expected/collector-contract.yml @@ -0,0 +1,22 @@ +# Licensed to the Apache Software Foundation (ASF) under one or more +# contributor license agreements. See the NOTICE file distributed with +# this work for additional information regarding copyright ownership. +# The ASF licenses this file to You under the Apache License, Version 2.0 +# (the "License"); you may not use this file except in compliance with +# the License. You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +collector_process_count: + datapoints_per_export: 1 + value: 3 + unit: "{process}" +windows_normalization: + cpu_state_to_mode: true + process_handles_to_open_handles: true diff --git a/test/e2e-v2/cases/vm/otel-hostmetrics/expected/process-count-3.yml b/test/e2e-v2/cases/vm/otel-hostmetrics/expected/process-count-3.yml new file mode 100644 index 000000000000..738637d60290 --- /dev/null +++ b/test/e2e-v2/cases/vm/otel-hostmetrics/expected/process-count-3.yml @@ -0,0 +1,30 @@ +# Licensed to the Apache Software Foundation (ASF) under one or more +# contributor license agreements. See the NOTICE file distributed with +# this work for additional information regarding copyright ownership. +# The ASF licenses this file to You under the Apache License, Version 2.0 +# (the "License"); you may not use this file except in compliance with +# the License. You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +debuggingtrace: null +type: TIME_SERIES_VALUES +results: + {{- contains .results }} + - metric: + labels: [] + values: + {{- contains .values }} + - id: {{ notEmpty .id }} + value: 3 + owner: null + traceid: null + {{- end}} + {{- end}} +error: null diff --git a/test/e2e-v2/cases/vm/otel-hostmetrics/fixtures/windows-host.json b/test/e2e-v2/cases/vm/otel-hostmetrics/fixtures/windows-host.json new file mode 100644 index 000000000000..048391f8999a --- /dev/null +++ b/test/e2e-v2/cases/vm/otel-hostmetrics/fixtures/windows-host.json @@ -0,0 +1 @@ +{"resourceMetrics":[{"resource":{"attributes":[{"key":"host.name","value":{"stringValue":"win-fixture"}}]},"scopeMetrics":[{"scope":{"name":"windows-host-fixture"},"metrics":[{"name":"system.cpu.logical.count","unit":"{cpu}","gauge":{"dataPoints":[{"timeUnixNano":"1788474000000000000","asInt":"4"}]}},{"name":"system.cpu.time","unit":"s","sum":{"dataPoints":[{"attributes":[{"key":"state","value":{"stringValue":"idle"}}],"timeUnixNano":"1788474000000000000","asDouble":100.0},{"attributes":[{"key":"state","value":{"stringValue":"user"}}],"timeUnixNano":"1788474000000000000","asDouble":20.0}],"aggregationTemporality":2,"isMonotonic":true}}]}]}]} diff --git a/test/e2e-v2/cases/vm/otel-hostmetrics/fixtures/windows-process.json b/test/e2e-v2/cases/vm/otel-hostmetrics/fixtures/windows-process.json new file mode 100644 index 000000000000..6b5edb6e3272 --- /dev/null +++ b/test/e2e-v2/cases/vm/otel-hostmetrics/fixtures/windows-process.json @@ -0,0 +1 @@ +{"resourceMetrics":[{"resource":{"attributes":[{"key":"host.name","value":{"stringValue":"win-fixture"}},{"key":"process.executable.name","value":{"stringValue":"svchost.exe"}}]},"scopeMetrics":[{"scope":{"name":"windows-process-fixture"},"metrics":[{"name":"process.threads","unit":"{threads}","gauge":{"dataPoints":[{"timeUnixNano":"1788474000000000000","asInt":"10"}]}},{"name":"process.handles","unit":"{handles}","gauge":{"dataPoints":[{"timeUnixNano":"1788474000000000000","asInt":"5"}]}}]}]}]} diff --git a/test/e2e-v2/cases/vm/otel-hostmetrics/otel-collector-config.yaml b/test/e2e-v2/cases/vm/otel-hostmetrics/otel-collector-config.yaml new file mode 100644 index 000000000000..e049333b394a --- /dev/null +++ b/test/e2e-v2/cases/vm/otel-hostmetrics/otel-collector-config.yaml @@ -0,0 +1,594 @@ +# Licensed to the Apache Software Foundation (ASF) under one or more +# contributor license agreements. See the NOTICE file distributed with +# this work for additional information regarding copyright ownership. +# The ASF licenses this file to You under the Apache License, Version 2.0 +# (the "License"); you may not use this file except in compliance with +# the License. You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +# +# E2E configuration for OpenTelemetry-native Linux host/process monitoring. +# +# Differences from the production/reference configuration: +# - collection_interval is 5s instead of 30s +# - process allowlist contains only "^sleep$" +# - OTLP endpoint is the Docker Compose service "oap:11800" +# + +receivers: + + otlp/windows-fixture-replay: + protocols: + http: + endpoint: 0.0.0.0:4318 + + # Synthetic OTLP JSON fixtures exercise the Windows normalization contract + # on the Linux E2E runner; no Windows host is required. + # --------------------------------------------------------------------------- + # Host metrics + # --------------------------------------------------------------------------- + hostmetrics/system: + collection_interval: 5s + + scrapers: + + cpu: + metrics: + system.cpu.utilization: + enabled: true + + system.cpu.logical.count: + enabled: true + + memory: + metrics: + system.linux.memory.available: + enabled: true + + system.memory.limit: + enabled: true + + disk: {} + + network: {} + + filesystem: {} + + paging: {} + + load: {} + + + # --------------------------------------------------------------------------- + # Process metrics + # + # The E2E workload starts exactly three "sleep" processes. + # Only those processes are collected. + # --------------------------------------------------------------------------- + hostmetrics/processes: + collection_interval: 5s + + scrapers: + + process: + + include: + match_type: regexp + names: + - "^sleep$" + + mute_process_all_errors: true + + metrics: + + process.cpu.time: + enabled: false + + process.cpu.utilization: + enabled: true + + process.memory.usage: + enabled: true + + process.memory.virtual: + enabled: false + + process.memory.utilization: + enabled: true + + process.threads: + enabled: true + + process.uptime: + enabled: true + + process.open_file_descriptors: + enabled: true + + process.context_switches: + enabled: false + + process.paging.faults: + enabled: false + + process.signals_pending: + enabled: false + + process.disk.io: + enabled: false + + process.disk.operations: + enabled: false + + resource_attributes: + + # This is the only process identity attribute required. + process.executable.name: + enabled: true + + process.executable.path: + enabled: false + + process.command: + enabled: false + + process.command_line: + enabled: false + + process.owner: + enabled: false + + process.pid: + enabled: false + + process.parent_pid: + enabled: false + + +processors: + groupbyattrs/process-linux: + keys: [] + + transform/process-aggregate-linux: + error_mode: propagate + metric_statements: + # hostmetrics/process generates each PID as a separate ResourceMetrics + # and assigns a slightly different timestamp to each process. + # + # groupbyattrs compacts ResourceMetrics with the same normalized + # process identity into common Metric objects. Align timestamps before + # aggregating their datapoints. + - context: datapoint + statements: + - set( + time, + TruncateTime(time, Duration("30s")) + ) where IsMatch(metric.name, "^process\\.") + + - context: metric + statements: + - aggregate_on_attributes( + "sum", + [] + ) where metric.name == "process.count" + + - aggregate_on_attributes( + "sum", + [] + ) where metric.name == "process.threads" + + # CPU keeps state because MAL needs user/system/wait separately. + - aggregate_on_attributes( + "sum", + ["state"] + ) where metric.name == "process.cpu.utilization" + + - aggregate_on_attributes( + "sum", + [] + ) where metric.name == "process.memory.usage" + + - aggregate_on_attributes( + "sum", + [] + ) where metric.name == "process.memory.virtual" + + - aggregate_on_attributes( + "sum", + [] + ) where metric.name == "process.memory.utilization" + + - aggregate_on_attributes( + "sum", + [] + ) where metric.name == "process.open_handles" + + # Logical process uptime is the oldest member of the group. + - aggregate_on_attributes( + "max", + [] + ) where metric.name == "process.uptime" + # --------------------------------------------------------------------------- + # Windows OTLP fixture normalization. These processors intentionally mirror + # the deployable Windows reference configuration. + # --------------------------------------------------------------------------- + transform/host-windows: + error_mode: ignore + metric_statements: + - context: resource + statements: + - set(resource.attributes["node_identifier_host_name"], resource.attributes["host.name"]) + where resource.attributes["host.name"] != nil + - set(resource.attributes["job_name"], "windows-monitoring") + - context: datapoint + statements: + - set(datapoint.attributes["mode"], datapoint.attributes["state"]) + where metric.name == "system.cpu.time" and datapoint.attributes["state"] != nil + - delete_key(datapoint.attributes, "state") + where metric.name == "system.cpu.time" + + transform/process-identity-windows: + error_mode: ignore + metric_statements: + - context: resource + statements: + - set(resource.attributes["node_identifier_host_name"], resource.attributes["host.name"]) + where resource.attributes["host.name"] != nil + - set(resource.attributes["job_name"], "hostmetrics-process-monitoring-windows") + - set(resource.attributes["process_name"], ConvertCase(resource.attributes["process.executable.name"], "lower")) + where resource.attributes["process.executable.name"] != nil + - delete_key(resource.attributes, "process.executable.name") + + filter/process-without-name-windows: + error_mode: ignore + metric_conditions: + - resource.attributes["process_name"] == nil + - resource.attributes["process_name"] == "" + + transform/process-prepare-windows: + error_mode: ignore + metric_statements: + - context: metric + statements: + - set(metric.name, "process.open_handles") + where metric.name == "process.handles" + - copy_metric( + name="process.count", + description="One sample per operating-system process", + unit="{process}" + ) where metric.name == "process.threads" + - context: datapoint + statements: + - set(value_int, 1) where metric.name == "process.count" + + groupbyattrs/process-windows: + keys: [] + + transform/process-aggregate-windows: + error_mode: propagate + metric_statements: + - context: datapoint + statements: + - set( + time, + TruncateTime(time, Duration("30s")) + ) where IsMatch(metric.name, "^process\\.") + - context: metric + statements: + - aggregate_on_attributes("sum", []) where metric.name == "process.count" + - aggregate_on_attributes("sum", []) where metric.name == "process.threads" + - aggregate_on_attributes("sum", []) where metric.name == "process.open_handles" + + # --------------------------------------------------------------------------- + # Detect host.name. + # + # Because docker-compose.yml sets: + # + # hostname: vm-service + # + # this should produce: + # + # host.name = vm-service + # + # --------------------------------------------------------------------------- + resource_detection/system: + detectors: + - system + + override: true + + system: + hostname_sources: + - os + + + # --------------------------------------------------------------------------- + # Host identity / routing + # + # Makes hostmetrics use the same canonical SkyWalking VM model as the + # existing node-exporter integration: + # + # node_identifier_host_name = host.name + # job_name = vm-monitoring + # + # CPU state is renamed to "mode" to preserve the canonical vm.yaml shape. + # --------------------------------------------------------------------------- + transform/host-linux: + error_mode: ignore + + metric_statements: + + - context: resource + statements: + + - set( + resource.attributes["node_identifier_host_name"], + resource.attributes["host.name"] + ) where resource.attributes["host.name"] != nil + + - set( + resource.attributes["job_name"], + "vm-monitoring" + ) + + - context: datapoint + statements: + + - set( + datapoint.attributes["mode"], + datapoint.attributes["state"] + ) where metric.name == "system.cpu.utilization" and datapoint.attributes["state"] != nil + + - set( + datapoint.attributes["mode"], + "irq" + ) where metric.name == "system.cpu.utilization" and datapoint.attributes["mode"] == "interrupt" + + - set( + datapoint.attributes["mode"], + "iowait" + ) where metric.name == "system.cpu.utilization" and datapoint.attributes["mode"] == "wait" + + - delete_key( + datapoint.attributes, + "state" + ) where metric.name == "system.cpu.utilization" + + - set( + datapoint.attributes["protocol"], + ConvertCase(datapoint.attributes["protocol"], "lower") + ) where metric.name == "system.network.connections" and datapoint.attributes["protocol"] != nil + + - set( + datapoint.attributes["state"], + ConvertCase(datapoint.attributes["state"], "lower") + ) where metric.name == "system.network.connections" and datapoint.attributes["state"] != nil + + + # --------------------------------------------------------------------------- + # Process identity + # + # Converts: + # + # process.executable.name = sleep + # + # into: + # + # node_identifier_host_name = vm-service + # job_name = hostmetrics-process-monitoring-linux + # process_name = sleep + # + # PID is deliberately NOT part of the identity. + # --------------------------------------------------------------------------- + transform/process-identity-linux: + error_mode: ignore + + metric_statements: + + - context: resource + statements: + + - set( + resource.attributes["node_identifier_host_name"], + resource.attributes["host.name"] + ) where resource.attributes["host.name"] != nil + + - set( + resource.attributes["job_name"], + "hostmetrics-process-monitoring-linux" + ) + + - set( + resource.attributes["process_name"], + ConvertCase( + resource.attributes["process.executable.name"], + "lower" + ) + ) where resource.attributes["process.executable.name"] != nil + + # Generic Python normalization retained from the reference config. + - replace_pattern( + resource.attributes["process_name"], + "^python[0-9.]*$", + "python" + ) where resource.attributes["process_name"] != nil + + # process_name is now the canonical process identity. + - delete_key( + resource.attributes, + "process.executable.name" + ) + + + # --------------------------------------------------------------------------- + # Do not send process telemetry without a normalized process name. + # --------------------------------------------------------------------------- + filter/process-without-name-linux: + error_mode: ignore + + metric_conditions: + + - resource.attributes["process_name"] == nil + + - resource.attributes["process_name"] == "" + + + # --------------------------------------------------------------------------- + # Prepare individual process metrics for Collector-side aggregation. + # + # Linux FD: + # process.open_file_descriptors -> process.open_handles + # + # Process count: + # create one process.count=1 sample per operating-system process. The + # following groupbyattrs and transform processors aggregate these samples + # by normalized process identity before OTLP export. + # --------------------------------------------------------------------------- + transform/process-prepare-linux: + error_mode: ignore + metric_statements: + + - context: metric + statements: + + - set( + metric.name, + "process.open_handles" + ) where metric.name == "process.open_file_descriptors" + + - copy_metric( + name="process.count", + description="One sample per operating-system process", + unit="{process}" + ) where metric.name == "process.threads" + + - context: datapoint + statements: + - set( + value_int, + 1 + ) where metric.name == "process.count" + + # --------------------------------------------------------------------------- + # Batch host telemetry. + # --------------------------------------------------------------------------- + batch/host-linux: + timeout: 2s + send_batch_size: 8192 + + + # --------------------------------------------------------------------------- + # Final batching before OTLP export. + # --------------------------------------------------------------------------- + batch/export: + timeout: 2s + send_batch_size: 8192 + + +exporters: + + # Detailed post-processor Collector output is asserted by + # assert-collector-output.sh before relying on OAP/MAL results. + debug/process-linux-raw: + verbosity: detailed + use_internal_logger: false + output_paths: + - /tmp/process-linux-raw.log + + debug/windows-host-raw: + verbosity: detailed + use_internal_logger: false + output_paths: + - /tmp/windows-host-raw.log + + debug/windows-process-raw: + verbosity: detailed + use_internal_logger: false + output_paths: + - /tmp/windows-process-raw.log + + # --------------------------------------------------------------------------- + # OAP service defined in docker-compose.yml. + # + # No authentication is required inside this isolated E2E network. + # --------------------------------------------------------------------------- + otlp: + endpoint: oap:11800 + + tls: + insecure: true + + +service: + + pipelines: + + # ------------------------------------------------------------------------- + # Host telemetry -> vm.yaml + # ------------------------------------------------------------------------- + metrics/host-linux: + + receivers: + - hostmetrics/system + + processors: + - resource_detection/system + - transform/host-linux + - batch/host-linux + + exporters: + - otlp + + + # ------------------------------------------------------------------------- + # Process telemetry -> process-hostmetrics-linux.yaml + # ------------------------------------------------------------------------- + metrics/processes-linux: + receivers: + - hostmetrics/processes + processors: + - resource_detection/system + - transform/process-identity-linux + - filter/process-without-name-linux + - transform/process-prepare-linux + - groupbyattrs/process-linux + - transform/process-aggregate-linux + - batch/export + exporters: + - otlp + - debug/process-linux-raw + + # ------------------------------------------------------------------------- + # Synthetic Windows host OTLP fixture -> windows.yaml + # ------------------------------------------------------------------------- + metrics/windows-host-fixture: + receivers: + - otlp/windows-fixture-replay + processors: + - transform/host-windows + exporters: + - otlp + - debug/windows-host-raw + + # ------------------------------------------------------------------------- + # Synthetic Windows process OTLP fixture -> process-hostmetrics-windows.yaml + # ------------------------------------------------------------------------- + metrics/windows-process-fixture: + receivers: + - otlp/windows-fixture-replay + processors: + - transform/process-identity-windows + - filter/process-without-name-windows + - transform/process-prepare-windows + - groupbyattrs/process-windows + - transform/process-aggregate-windows + exporters: + - otlp + - debug/windows-process-raw diff --git a/test/e2e-v2/cases/vm/otel-hostmetrics/otel-rules/process-hostmetrics-linux.yaml b/test/e2e-v2/cases/vm/otel-hostmetrics/otel-rules/process-hostmetrics-linux.yaml new file mode 100644 index 000000000000..71356da371b6 --- /dev/null +++ b/test/e2e-v2/cases/vm/otel-hostmetrics/otel-rules/process-hostmetrics-linux.yaml @@ -0,0 +1,84 @@ +# Licensed to the Apache Software Foundation (ASF) under one or more +# contributor license agreements. See the NOTICE file distributed with +# this work for additional information regarding copyright ownership. +# The ASF licenses this file to You under the Apache License, Version 2.0 +# (the "License"); you may not use this file except in compliance with +# the License. You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +# The Collector normalizes process identity and pre-aggregates operating-system +# processes with the same process_name before OTLP export. MAL maps the +# resulting logical process metrics to the corresponding SkyWalking instance. +filter: "{ tags -> tags.job_name == 'hostmetrics-process-monitoring-linux' }" + +expSuffix: |- + service(['node_identifier_host_name'], '', Layer.OS_LINUX) + .instance( + ['node_identifier_host_name'], + '', + ['process_name'], + '', + Layer.OS_LINUX, + { + tags -> [ + 'process_name': tags.process_name, + 'host': tags.node_identifier_host_name + ] + } + ) + +metricPrefix: mp_process_linux + +metricsRules: + - name: num_procs + exp: "process_count.sum(['node_identifier_host_name','process_name'])" + + - name: num_threads + exp: "process_threads.sum(['node_identifier_host_name','process_name'])" + + - name: cpu_total_percent + exp: > + process_cpu_utilization + .tagNotEqual('state','wait') + .sum(['node_identifier_host_name','process_name']) + * 100 + + - name: cpu_user_percent + exp: > + process_cpu_utilization + .tagEqual('state','user') + .sum(['node_identifier_host_name','process_name']) + * 100 + + - name: cpu_system_percent + exp: > + process_cpu_utilization + .tagEqual('state','system') + .sum(['node_identifier_host_name','process_name']) + * 100 + + - name: memory_resident_bytes + exp: "process_memory_usage.sum(['node_identifier_host_name','process_name'])" + + # process.memory.utilization is emitted as a [0,1] fraction. Multiply by + # 100 before SkyWalking fixed-point storage to preserve two decimal places + # of percentage precision (basis points). UI expressions must divide the + # stored metric by 100. + - name: memory_resident_percent + exp: > + process_memory_utilization + .sum(['node_identifier_host_name','process_name']) + * 100 + + - name: open_handles + exp: "process_open_handles.sum(['node_identifier_host_name','process_name'])" + + - name: oldest_process_uptime_seconds + exp: "process_uptime.max(['node_identifier_host_name','process_name'])" diff --git a/test/e2e-v2/cases/vm/otel-hostmetrics/otel-rules/process-hostmetrics-windows.yaml b/test/e2e-v2/cases/vm/otel-hostmetrics/otel-rules/process-hostmetrics-windows.yaml new file mode 100644 index 000000000000..bde9f5b73580 --- /dev/null +++ b/test/e2e-v2/cases/vm/otel-hostmetrics/otel-rules/process-hostmetrics-windows.yaml @@ -0,0 +1,84 @@ +# Licensed to the Apache Software Foundation (ASF) under one or more +# contributor license agreements. See the NOTICE file distributed with +# this work for additional information regarding copyright ownership. +# The ASF licenses this file to You under the Apache License, Version 2.0 +# (the "License"); you may not use this file except in compliance with +# the License. You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +# The Collector normalizes process identity and pre-aggregates operating-system +# processes with the same process_name before OTLP export. MAL maps the +# resulting logical process metrics to the corresponding SkyWalking instance. +filter: "{ tags -> tags.job_name == 'hostmetrics-process-monitoring-windows' }" + +expSuffix: |- + service(['node_identifier_host_name'], '', Layer.OS_WINDOWS) + .instance( + ['node_identifier_host_name'], + '', + ['process_name'], + '', + Layer.OS_WINDOWS, + { + tags -> [ + 'process_name': tags.process_name, + 'host': tags.node_identifier_host_name + ] + } + ) + +metricPrefix: mp_process_windows + +metricsRules: + - name: num_procs + exp: "process_count.sum(['node_identifier_host_name','process_name'])" + + - name: num_threads + exp: "process_threads.sum(['node_identifier_host_name','process_name'])" + + - name: cpu_total_percent + exp: > + process_cpu_utilization + .tagNotEqual('state','wait') + .sum(['node_identifier_host_name','process_name']) + * 100 + + - name: cpu_user_percent + exp: > + process_cpu_utilization + .tagEqual('state','user') + .sum(['node_identifier_host_name','process_name']) + * 100 + + - name: cpu_system_percent + exp: > + process_cpu_utilization + .tagEqual('state','system') + .sum(['node_identifier_host_name','process_name']) + * 100 + + - name: memory_resident_bytes + exp: "process_memory_usage.sum(['node_identifier_host_name','process_name'])" + + # process.memory.utilization is emitted as a [0,1] fraction. Multiply by + # 100 before SkyWalking fixed-point storage to preserve two decimal places + # of percentage precision (basis points). UI expressions must divide the + # stored metric by 100. + - name: memory_resident_percent + exp: > + process_memory_utilization + .sum(['node_identifier_host_name','process_name']) + * 100 + + - name: open_handles + exp: "process_open_handles.sum(['node_identifier_host_name','process_name'])" + + - name: oldest_process_uptime_seconds + exp: "process_uptime.max(['node_identifier_host_name','process_name'])" diff --git a/test/e2e-v2/cases/vm/otel-hostmetrics/otel-rules/vm.yaml b/test/e2e-v2/cases/vm/otel-hostmetrics/otel-rules/vm.yaml new file mode 100644 index 000000000000..3b50da8afafd --- /dev/null +++ b/test/e2e-v2/cases/vm/otel-hostmetrics/otel-rules/vm.yaml @@ -0,0 +1,296 @@ +# Licensed to the Apache Software Foundation (ASF) under one or more +# contributor license agreements. See the NOTICE file distributed with +# this work for additional information regarding copyright ownership. +# The ASF licenses this file to You under the Apache License, Version 2.0 +# (the "License"); you may not use this file except in compliance with +# the License. You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. +# +# Linux infrastructure MAL. +# +# This rule intentionally accepts TWO source metric families under the same +# existing job_name "vm-monitoring": +# 1. node-exporter metrics (existing SkyWalking integration) +# 2. OpenTelemetry hostmetrics metrics (new alternative source) +# +# Both sources are mapped by ONE MAL file to the canonical meter_vm_* targets. +# This avoids cross-file metric ownership collisions in OAP 11 runtime rules. +# +# MAL SampleFamily '+' treats an EMPTY family as identity. Because a host sends +# one source family or the other, each expression can safely add the equivalent +# node-exporter and hostmetrics branches. The output metric is registered once. +filter: "{ tags -> tags.job_name == 'vm-monitoring' }" +metricPrefix: meter_vm + +metricsRules: + # --------------------------------------------------------------------------- + # Canonical metrics: same target names and semantics for both sources. + # --------------------------------------------------------------------------- + + # CPU + - name: cpu_total_percentage + exp: > + ( + (node_cpu_seconds_total * 100) + .tagNotEqual('mode','idle') + .sum(['node_identifier_host_name']) + .rate('PT1M') + + + (system_cpu_utilization * 100) + .tagNotEqual('mode','idle') + .sum(['node_identifier_host_name']) + ).service(['node_identifier_host_name'], Layer.OS_LINUX) + + - name: cpu_average_used + exp: > + ( + (node_cpu_seconds_total * 100) + .sum(['node_identifier_host_name','mode']) + .rate('PT1M') + + + (system_cpu_utilization * 100) + .sum(['node_identifier_host_name','mode']) + ).service(['node_identifier_host_name'], Layer.OS_LINUX) + + - name: cpu_load1 + exp: "((node_load1 * 100) + (system_cpu_load_average_1m * 100)).service(['node_identifier_host_name'], Layer.OS_LINUX).decorate({ me -> me.attr0 = me.layer.name()})" + + - name: cpu_load5 + exp: "((node_load5 * 100) + (system_cpu_load_average_5m * 100)).service(['node_identifier_host_name'], Layer.OS_LINUX)" + + - name: cpu_load15 + exp: "((node_load15 * 100) + (system_cpu_load_average_15m * 100)).service(['node_identifier_host_name'], Layer.OS_LINUX)" + + # Physical memory + - name: memory_total + exp: "(node_memory_MemTotal_bytes + system_memory_limit).service(['node_identifier_host_name'], Layer.OS_LINUX)" + + - name: memory_available + exp: "(node_memory_MemAvailable_bytes + system_linux_memory_available).service(['node_identifier_host_name'], Layer.OS_LINUX)" + + - name: memory_used + exp: > + ( + (node_memory_MemTotal_bytes - node_memory_MemAvailable_bytes) + + + (system_memory_limit - system_linux_memory_available) + ).service(['node_identifier_host_name'], Layer.OS_LINUX) + + - name: memory_buff_cache + exp: > + ( + (node_memory_Buffers_bytes + node_memory_Cached_bytes) + + + ( + system_memory_usage.tagEqual('state','buffered').sum(['node_identifier_host_name']) + + + system_memory_usage.tagEqual('state','cached').sum(['node_identifier_host_name']) + - + system_memory_usage.tagEqual('state','slab_reclaimable').sum(['node_identifier_host_name']) + ) + ).service(['node_identifier_host_name'], Layer.OS_LINUX) + + # Swap + - name: memory_swap_free + exp: > + ( + node_memory_SwapFree_bytes + + + system_paging_usage + .tagEqual('state','free') + .sum(['node_identifier_host_name']) + ).service(['node_identifier_host_name'], Layer.OS_LINUX) + + - name: memory_swap_total + exp: > + ( + node_memory_SwapTotal_bytes + + + ( + system_paging_usage.tagEqual('state','used').sum(['node_identifier_host_name']) + + + system_paging_usage.tagEqual('state','free').sum(['node_identifier_host_name']) + ) + ).service(['node_identifier_host_name'], Layer.OS_LINUX) + + - name: memory_swap_percentage + exp: > + ( + ( + 100 - ((node_memory_SwapFree_bytes * 100) / node_memory_SwapTotal_bytes) + ) + + + ( + ( + system_paging_usage.tagEqual('state','used').sum(['node_identifier_host_name']) + * 100 + ) + / + ( + system_paging_usage.tagEqual('state','used').sum(['node_identifier_host_name']) + + + system_paging_usage.tagEqual('state','free').sum(['node_identifier_host_name']) + ) + ) + ).service(['node_identifier_host_name'], Layer.OS_LINUX) + + # Filesystem. hostmetrics free+reserved+used is the full capacity. + - name: filesystem_percentage + exp: > + ( + ( + 100 - ( + (node_filesystem_avail_bytes * 100) + .sum(['node_identifier_host_name','mountpoint']) + / + node_filesystem_size_bytes + .sum(['node_identifier_host_name','mountpoint']) + ) + ) + + + ( + 100 - ( + (system_filesystem_usage.tagEqual('state','free') * 100) + .sum(['node_identifier_host_name','mountpoint']) + / + ( + system_filesystem_usage.tagEqual('state','free') + .sum(['node_identifier_host_name','mountpoint']) + + + system_filesystem_usage.tagEqual('state','reserved') + .sum(['node_identifier_host_name','mountpoint']) + + + system_filesystem_usage.tagEqual('state','used') + .sum(['node_identifier_host_name','mountpoint']) + ) + ) + ) + ).service(['node_identifier_host_name'], Layer.OS_LINUX) + + # Disk throughput + - name: disk_read + exp: > + ( + node_disk_read_bytes_total + .sum(['node_identifier_host_name']) + .rate('PT1M') + + + system_disk_io + .tagEqual('direction','read') + .sum(['node_identifier_host_name']) + .rate('PT1M') + ).service(['node_identifier_host_name'], Layer.OS_LINUX) + + - name: disk_written + exp: > + ( + node_disk_written_bytes_total + .sum(['node_identifier_host_name']) + .rate('PT1M') + + + system_disk_io + .tagEqual('direction','write') + .sum(['node_identifier_host_name']) + .rate('PT1M') + ).service(['node_identifier_host_name'], Layer.OS_LINUX) + + # Network throughput + - name: network_receive + exp: > + ( + node_network_receive_bytes_total + .sum(['node_identifier_host_name']) + .irate() + + + system_network_io + .tagEqual('direction','receive') + .sum(['node_identifier_host_name']) + .irate() + ).service(['node_identifier_host_name'], Layer.OS_LINUX) + + - name: network_transmit + exp: > + ( + node_network_transmit_bytes_total + .sum(['node_identifier_host_name']) + .irate() + + + system_network_io + .tagEqual('direction','transmit') + .sum(['node_identifier_host_name']) + .irate() + ).service(['node_identifier_host_name'], Layer.OS_LINUX) + + # TCP connection metrics with true hostmetrics equivalents. + # Collector normalizes system.network.connections state values to lowercase. + - name: tcp_curr_estab + exp: > + ( + node_netstat_Tcp_CurrEstab + + + ( + system_network_connections + .tagEqual('protocol','tcp') + .tagEqual('state','established') + .sum(['node_identifier_host_name']) + + + system_network_connections + .tagEqual('protocol','tcp') + .tagEqual('state','close_wait') + .sum(['node_identifier_host_name']) + ) + ).service(['node_identifier_host_name'], Layer.OS_LINUX) + + - name: tcp_tw + exp: > + ( + node_sockstat_TCP_tw + + + system_network_connections + .tagEqual('protocol','tcp') + .tagEqual('state','time_wait') + .sum(['node_identifier_host_name']) + ).service(['node_identifier_host_name'], Layer.OS_LINUX) + + # --------------------------------------------------------------------------- + # Existing node-exporter-only canonical metrics. + # hostmetrics has no source with equivalent semantics, so these remain node + # exporter only rather than being populated with a misleading approximation. + # --------------------------------------------------------------------------- + - name: tcp_alloc + exp: (node_sockstat_TCP_alloc).service(['node_identifier_host_name'], Layer.OS_LINUX) + + - name: sockets_used + exp: (node_sockstat_sockets_used).service(['node_identifier_host_name'], Layer.OS_LINUX) + + - name: udp_inuse + exp: (node_sockstat_UDP_inuse).service(['node_identifier_host_name'], Layer.OS_LINUX) + + - name: filefd_allocated + exp: (node_filefd_allocated).service(['node_identifier_host_name'], Layer.OS_LINUX) + + # --------------------------------------------------------------------------- + # New hostmetrics-only capabilities. New names are used whenever semantics + # do not exactly match an existing node-exporter target metric. + # --------------------------------------------------------------------------- + - name: cpu_cores_num + exp: (system_cpu_logical_count.sum(['node_identifier_host_name'])).service(['node_identifier_host_name'], Layer.OS_LINUX) + + - name: cpu_norm_percentage + exp: > + ( + ( + (system_cpu_utilization * 100) + .tagNotEqual('mode','idle') + .sum(['node_identifier_host_name']) + ) + / + system_cpu_logical_count.sum(['node_identifier_host_name']) + ).service(['node_identifier_host_name'], Layer.OS_LINUX) diff --git a/test/e2e-v2/cases/vm/otel-hostmetrics/otel-rules/windows.yaml b/test/e2e-v2/cases/vm/otel-hostmetrics/otel-rules/windows.yaml new file mode 100644 index 000000000000..2e872e86659a --- /dev/null +++ b/test/e2e-v2/cases/vm/otel-hostmetrics/otel-rules/windows.yaml @@ -0,0 +1,225 @@ +# Licensed to the Apache Software Foundation (ASF) under one or more +# contributor license agreements. See the NOTICE file distributed with +# this work for additional information regarding copyright ownership. +# The ASF licenses this file to You under the Apache License, Version 2.0 +# (the "License"); you may not use this file except in compliance with +# the License. You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +# Windows infrastructure MAL. +# +# The existing windows-exporter source and the OpenTelemetry hostmetrics source +# intentionally use the same existing job_name "windows-monitoring". One MAL +# file maps either input family to the canonical meter_win_* targets, avoiding +# cross-file metric ownership collisions in OAP 11. +filter: "{ tags -> tags.job_name == 'windows-monitoring' }" +expSuffix: service(['node_identifier_host_name'], Layer.OS_WINDOWS) +metricPrefix: meter_win + +metricsRules: + # --------------------------------------------------------------------------- + # Canonical windows.yaml metrics, now accepting either source. + # --------------------------------------------------------------------------- + + # CPU + - name: cpu_total_percentage + exp: > + (windows_cpu_time_total * 100) + .tagNotEqual('mode','idle') + .sum(['node_identifier_host_name']) + .rate('PT1M') + + + (system_cpu_time * 100) + .tagNotEqual('mode','idle') + .sum(['node_identifier_host_name']) + .rate('PT1M') + + - name: cpu_average_used + exp: > + (windows_cpu_time_total * 100) + .sum(['node_identifier_host_name','mode']) + .rate('PT1M') + + + (system_cpu_time * 100) + .sum(['node_identifier_host_name','mode']) + .rate('PT1M') + + # Physical memory + - name: memory_total + exp: > + windows_cs_physical_memory_bytes + + + system_memory_limit.sum(['node_identifier_host_name']) + + - name: memory_available + exp: > + windows_os_physical_memory_free_bytes + + + system_memory_usage + .tagEqual('state','free') + .sum(['node_identifier_host_name']) + + - name: memory_used + exp: > + (windows_cs_physical_memory_bytes - windows_os_physical_memory_free_bytes) + + + ( + system_memory_limit.sum(['node_identifier_host_name']) + - + system_memory_usage.tagEqual('state','free').sum(['node_identifier_host_name']) + ) + + # Canonical virtual memory semantics. + # The hostmetrics branch uses Windows Performance Counters: + # Memory\\Commit Limit + # Memory\\Committed Bytes + - name: memory_virtual_memory_free + exp: > + windows_os_virtual_memory_free_bytes + + + ( + system_virtual_memory_commit_limit.sum(['node_identifier_host_name']) + - + system_virtual_memory_committed.sum(['node_identifier_host_name']) + ) + + - name: memory_virtual_memory_total + exp: > + windows_os_virtual_memory_bytes + + + system_virtual_memory_commit_limit.sum(['node_identifier_host_name']) + + - name: memory_virtual_memory_percentage + exp: > + ( + 100 - ((windows_os_virtual_memory_free_bytes * 100) / windows_os_virtual_memory_bytes) + ) + + + ( + system_virtual_memory_committed.sum(['node_identifier_host_name']) * 100 + / + system_virtual_memory_commit_limit.sum(['node_identifier_host_name']) + ) + + # Disk throughput + - name: disk_read + exp: > + windows_logical_disk_read_bytes_total + .sum(['node_identifier_host_name']) + .rate('PT1M') + + + system_disk_io + .tagEqual('direction','read') + .sum(['node_identifier_host_name']) + .rate('PT1M') + + - name: disk_written + exp: > + windows_logical_disk_write_bytes_total + .sum(['node_identifier_host_name']) + .rate('PT1M') + + + system_disk_io + .tagEqual('direction','write') + .sum(['node_identifier_host_name']) + .rate('PT1M') + + # Network throughput + - name: network_receive + exp: > + windows_net_bytes_received_total + .sum(['node_identifier_host_name']) + .irate() + + + system_network_io + .tagEqual('direction','receive') + .sum(['node_identifier_host_name']) + .irate() + + - name: network_transmit + exp: > + windows_net_bytes_sent_total + .sum(['node_identifier_host_name']) + .irate() + + + system_network_io + .tagEqual('direction','transmit') + .sum(['node_identifier_host_name']) + .irate() + + # --------------------------------------------------------------------------- + # Additional hostmetrics capabilities. These are additive and do not change + # the existing windows-exporter metric contract. + # --------------------------------------------------------------------------- + - name: cpu_cores_num + exp: system_cpu_logical_count.sum(['node_identifier_host_name']) + + - name: cpu_norm_percentage + exp: > + ( + (system_cpu_time * 100) + .tagNotEqual('mode','idle') + .sum(['node_identifier_host_name']) + .rate('PT1M') + ) + / + system_cpu_logical_count.sum(['node_identifier_host_name']) + + - name: cpu_load1 + exp: system_cpu_load_average_1m.sum(['node_identifier_host_name']) * 100 + + - name: cpu_load5 + exp: system_cpu_load_average_5m.sum(['node_identifier_host_name']) * 100 + + - name: cpu_load15 + exp: system_cpu_load_average_15m.sum(['node_identifier_host_name']) * 100 + + - name: filesystem_percentage + exp: > + 100 - ( + (system_filesystem_usage.tagEqual('state','free') * 100) + .sum(['node_identifier_host_name','mountpoint']) + / + ( + system_filesystem_usage.tagEqual('state','free') + .sum(['node_identifier_host_name','mountpoint']) + + + system_filesystem_usage.tagEqual('state','reserved') + .sum(['node_identifier_host_name','mountpoint']) + + + system_filesystem_usage.tagEqual('state','used') + .sum(['node_identifier_host_name','mountpoint']) + ) + ) + + - name: filehandles_allocated + exp: system_filehandles_allocated.sum(['node_identifier_host_name']) + + # system.paging.usage is the Windows pagefile, not the windows.yaml virtual + # memory metric. Keep it under explicit new names. + - name: memory_pagefile_free + exp: > + system_paging_usage + .tagEqual('state','free') + .sum(['node_identifier_host_name']) + + - name: memory_pagefile_total + exp: system_paging_usage.sum(['node_identifier_host_name']) + + - name: memory_pagefile_percentage + exp: > + ( + system_paging_usage + .tagEqual('state','used') + .sum(['node_identifier_host_name']) + * 100 + ) + / + system_paging_usage.sum(['node_identifier_host_name']) diff --git a/test/e2e-v2/cases/vm/prometheus-node-exporter/otel-rules/vm.yaml b/test/e2e-v2/cases/vm/prometheus-node-exporter/otel-rules/vm.yaml index a0921ad99d91..3b50da8afafd 100644 --- a/test/e2e-v2/cases/vm/prometheus-node-exporter/otel-rules/vm.yaml +++ b/test/e2e-v2/cases/vm/prometheus-node-exporter/otel-rules/vm.yaml @@ -1,9 +1,9 @@ # Licensed to the Apache Software Foundation (ASF) under one or more -# contributor license agreements. See the NOTICE file distributed with +# contributor license agreements. See the NOTICE file distributed with # this work for additional information regarding copyright ownership. # The ASF licenses this file to You under the Apache License, Version 2.0 # (the "License"); you may not use this file except in compliance with -# the License. You may obtain a copy of the License at +# the License. You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # @@ -12,85 +12,285 @@ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. - -# This will parse a textual representation of a duration. The formats -# accepted are based on the ISO-8601 duration format {@code PnDTnHnMn.nS} -# with days considered to be exactly 24 hours. -#
-# Examples: -#
-# "PT20.345S" -- parses as "20.345 seconds" -# "PT15M" -- parses as "15 minutes" (where a minute is 60 seconds) -# "PT10H" -- parses as "10 hours" (where an hour is 3600 seconds) -# "P2D" -- parses as "2 days" (where a day is 24 hours or 86400 seconds) -# "P2DT3H4M" -- parses as "2 days, 3 hours and 4 minutes" -# "P-6H3M" -- parses as "-6 hours and +3 minutes" -# "-P6H3M" -- parses as "-6 hours and -3 minutes" -# "-P-6H+3M" -- parses as "+6 hours and -3 minutes" -#-filter: "{ tags -> tags.job_name == 'vm-monitoring' }" # The OpenTelemetry job name +# +# Linux infrastructure MAL. +# +# This rule intentionally accepts TWO source metric families under the same +# existing job_name "vm-monitoring": +# 1. node-exporter metrics (existing SkyWalking integration) +# 2. OpenTelemetry hostmetrics metrics (new alternative source) +# +# Both sources are mapped by ONE MAL file to the canonical meter_vm_* targets. +# This avoids cross-file metric ownership collisions in OAP 11 runtime rules. +# +# MAL SampleFamily '+' treats an EMPTY family as identity. Because a host sends +# one source family or the other, each expression can safely add the equivalent +# node-exporter and hostmetrics branches. The output metric is registered once. +filter: "{ tags -> tags.job_name == 'vm-monitoring' }" metricPrefix: meter_vm + metricsRules: + # --------------------------------------------------------------------------- + # Canonical metrics: same target names and semantics for both sources. + # --------------------------------------------------------------------------- - #node cpu + # CPU - name: cpu_total_percentage - exp: (node_cpu_seconds_total * 100).tagNotEqual('mode' , 'idle').sum(['node_identifier_host_name']).rate('PT1M').service(['node_identifier_host_name'] , Layer.OS_LINUX) + exp: > + ( + (node_cpu_seconds_total * 100) + .tagNotEqual('mode','idle') + .sum(['node_identifier_host_name']) + .rate('PT1M') + + + (system_cpu_utilization * 100) + .tagNotEqual('mode','idle') + .sum(['node_identifier_host_name']) + ).service(['node_identifier_host_name'], Layer.OS_LINUX) + - name: cpu_average_used - exp: (node_cpu_seconds_total * 100).sum(['node_identifier_host_name' , 'mode']).rate('PT1M').service(['node_identifier_host_name'] , Layer.OS_LINUX) + exp: > + ( + (node_cpu_seconds_total * 100) + .sum(['node_identifier_host_name','mode']) + .rate('PT1M') + + + (system_cpu_utilization * 100) + .sum(['node_identifier_host_name','mode']) + ).service(['node_identifier_host_name'], Layer.OS_LINUX) + - name: cpu_load1 - exp: (node_load1 * 100).service(['node_identifier_host_name'] , Layer.OS_LINUX).decorate({ me -> me.attr0 = me.layer.name()}) + exp: "((node_load1 * 100) + (system_cpu_load_average_1m * 100)).service(['node_identifier_host_name'], Layer.OS_LINUX).decorate({ me -> me.attr0 = me.layer.name()})" + - name: cpu_load5 - exp: (node_load5 * 100).service(['node_identifier_host_name'] , Layer.OS_LINUX) + exp: "((node_load5 * 100) + (system_cpu_load_average_5m * 100)).service(['node_identifier_host_name'], Layer.OS_LINUX)" + - name: cpu_load15 - exp: (node_load15 * 100).service(['node_identifier_host_name'] , Layer.OS_LINUX) + exp: "((node_load15 * 100) + (system_cpu_load_average_15m * 100)).service(['node_identifier_host_name'], Layer.OS_LINUX)" - #node Memory + # Physical memory - name: memory_total - exp: node_memory_MemTotal_bytes.service(['node_identifier_host_name'] , Layer.OS_LINUX) + exp: "(node_memory_MemTotal_bytes + system_memory_limit).service(['node_identifier_host_name'], Layer.OS_LINUX)" + - name: memory_available - exp: node_memory_MemAvailable_bytes.service(['node_identifier_host_name'] , Layer.OS_LINUX) + exp: "(node_memory_MemAvailable_bytes + system_linux_memory_available).service(['node_identifier_host_name'], Layer.OS_LINUX)" + - name: memory_used - exp: (node_memory_MemTotal_bytes - node_memory_MemAvailable_bytes).service(['node_identifier_host_name'] , Layer.OS_LINUX) + exp: > + ( + (node_memory_MemTotal_bytes - node_memory_MemAvailable_bytes) + + + (system_memory_limit - system_linux_memory_available) + ).service(['node_identifier_host_name'], Layer.OS_LINUX) + - name: memory_buff_cache - exp: (node_memory_Buffers_bytes + node_memory_Cached_bytes).service(['node_identifier_host_name'] , Layer.OS_LINUX) + exp: > + ( + (node_memory_Buffers_bytes + node_memory_Cached_bytes) + + + ( + system_memory_usage.tagEqual('state','buffered').sum(['node_identifier_host_name']) + + + system_memory_usage.tagEqual('state','cached').sum(['node_identifier_host_name']) + - + system_memory_usage.tagEqual('state','slab_reclaimable').sum(['node_identifier_host_name']) + ) + ).service(['node_identifier_host_name'], Layer.OS_LINUX) + + # Swap - name: memory_swap_free - exp: node_memory_SwapFree_bytes.service(['node_identifier_host_name'] , Layer.OS_LINUX) + exp: > + ( + node_memory_SwapFree_bytes + + + system_paging_usage + .tagEqual('state','free') + .sum(['node_identifier_host_name']) + ).service(['node_identifier_host_name'], Layer.OS_LINUX) + - name: memory_swap_total - exp: node_memory_SwapTotal_bytes.service(['node_identifier_host_name'] , Layer.OS_LINUX) + exp: > + ( + node_memory_SwapTotal_bytes + + + ( + system_paging_usage.tagEqual('state','used').sum(['node_identifier_host_name']) + + + system_paging_usage.tagEqual('state','free').sum(['node_identifier_host_name']) + ) + ).service(['node_identifier_host_name'], Layer.OS_LINUX) + - name: memory_swap_percentage - exp: (100 - ((node_memory_SwapFree_bytes * 100) / node_memory_SwapTotal_bytes)).service(['node_identifier_host_name'] , Layer.OS_LINUX) + exp: > + ( + ( + 100 - ((node_memory_SwapFree_bytes * 100) / node_memory_SwapTotal_bytes) + ) + + + ( + ( + system_paging_usage.tagEqual('state','used').sum(['node_identifier_host_name']) + * 100 + ) + / + ( + system_paging_usage.tagEqual('state','used').sum(['node_identifier_host_name']) + + + system_paging_usage.tagEqual('state','free').sum(['node_identifier_host_name']) + ) + ) + ).service(['node_identifier_host_name'], Layer.OS_LINUX) - #node filesystem + # Filesystem. hostmetrics free+reserved+used is the full capacity. - name: filesystem_percentage - exp: (100 - ((node_filesystem_avail_bytes * 100).sum(['node_identifier_host_name' , 'mountpoint']) / node_filesystem_size_bytes.sum(['node_identifier_host_name' , 'mountpoint']))).service(['node_identifier_host_name'] , Layer.OS_LINUX) + exp: > + ( + ( + 100 - ( + (node_filesystem_avail_bytes * 100) + .sum(['node_identifier_host_name','mountpoint']) + / + node_filesystem_size_bytes + .sum(['node_identifier_host_name','mountpoint']) + ) + ) + + + ( + 100 - ( + (system_filesystem_usage.tagEqual('state','free') * 100) + .sum(['node_identifier_host_name','mountpoint']) + / + ( + system_filesystem_usage.tagEqual('state','free') + .sum(['node_identifier_host_name','mountpoint']) + + + system_filesystem_usage.tagEqual('state','reserved') + .sum(['node_identifier_host_name','mountpoint']) + + + system_filesystem_usage.tagEqual('state','used') + .sum(['node_identifier_host_name','mountpoint']) + ) + ) + ) + ).service(['node_identifier_host_name'], Layer.OS_LINUX) - #node disk + # Disk throughput - name: disk_read - exp: (node_disk_read_bytes_total.sum(['node_identifier_host_name']).rate('PT1M')).service(['node_identifier_host_name'] , Layer.OS_LINUX) + exp: > + ( + node_disk_read_bytes_total + .sum(['node_identifier_host_name']) + .rate('PT1M') + + + system_disk_io + .tagEqual('direction','read') + .sum(['node_identifier_host_name']) + .rate('PT1M') + ).service(['node_identifier_host_name'], Layer.OS_LINUX) + - name: disk_written - exp: (node_disk_written_bytes_total.sum(['node_identifier_host_name']).rate('PT1M')).service(['node_identifier_host_name'] , Layer.OS_LINUX) + exp: > + ( + node_disk_written_bytes_total + .sum(['node_identifier_host_name']) + .rate('PT1M') + + + system_disk_io + .tagEqual('direction','write') + .sum(['node_identifier_host_name']) + .rate('PT1M') + ).service(['node_identifier_host_name'], Layer.OS_LINUX) - #node network + # Network throughput - name: network_receive - exp: (node_network_receive_bytes_total.sum(['node_identifier_host_name']).irate()).service(['node_identifier_host_name'] , Layer.OS_LINUX) + exp: > + ( + node_network_receive_bytes_total + .sum(['node_identifier_host_name']) + .irate() + + + system_network_io + .tagEqual('direction','receive') + .sum(['node_identifier_host_name']) + .irate() + ).service(['node_identifier_host_name'], Layer.OS_LINUX) + - name: network_transmit - exp: (node_network_transmit_bytes_total.sum(['node_identifier_host_name']).irate()).service(['node_identifier_host_name'] , Layer.OS_LINUX) + exp: > + ( + node_network_transmit_bytes_total + .sum(['node_identifier_host_name']) + .irate() + + + system_network_io + .tagEqual('direction','transmit') + .sum(['node_identifier_host_name']) + .irate() + ).service(['node_identifier_host_name'], Layer.OS_LINUX) - #node netstat + # TCP connection metrics with true hostmetrics equivalents. + # Collector normalizes system.network.connections state values to lowercase. - name: tcp_curr_estab - exp: node_netstat_Tcp_CurrEstab.service(['node_identifier_host_name'] , Layer.OS_LINUX) + exp: > + ( + node_netstat_Tcp_CurrEstab + + + ( + system_network_connections + .tagEqual('protocol','tcp') + .tagEqual('state','established') + .sum(['node_identifier_host_name']) + + + system_network_connections + .tagEqual('protocol','tcp') + .tagEqual('state','close_wait') + .sum(['node_identifier_host_name']) + ) + ).service(['node_identifier_host_name'], Layer.OS_LINUX) + - name: tcp_tw - exp: node_sockstat_TCP_tw.service(['node_identifier_host_name'] , Layer.OS_LINUX) + exp: > + ( + node_sockstat_TCP_tw + + + system_network_connections + .tagEqual('protocol','tcp') + .tagEqual('state','time_wait') + .sum(['node_identifier_host_name']) + ).service(['node_identifier_host_name'], Layer.OS_LINUX) + + # --------------------------------------------------------------------------- + # Existing node-exporter-only canonical metrics. + # hostmetrics has no source with equivalent semantics, so these remain node + # exporter only rather than being populated with a misleading approximation. + # --------------------------------------------------------------------------- - name: tcp_alloc - exp: node_sockstat_TCP_alloc.service(['node_identifier_host_name'] , Layer.OS_LINUX) + exp: (node_sockstat_TCP_alloc).service(['node_identifier_host_name'], Layer.OS_LINUX) + - name: sockets_used - exp: node_sockstat_sockets_used.service(['node_identifier_host_name'] , Layer.OS_LINUX) + exp: (node_sockstat_sockets_used).service(['node_identifier_host_name'], Layer.OS_LINUX) + - name: udp_inuse - exp: node_sockstat_UDP_inuse.service(['node_identifier_host_name'] , Layer.OS_LINUX) + exp: (node_sockstat_UDP_inuse).service(['node_identifier_host_name'], Layer.OS_LINUX) - #node filefd - name: filefd_allocated - exp: node_filefd_allocated.service(['node_identifier_host_name'] , Layer.OS_LINUX) - + exp: (node_filefd_allocated).service(['node_identifier_host_name'], Layer.OS_LINUX) + # --------------------------------------------------------------------------- + # New hostmetrics-only capabilities. New names are used whenever semantics + # do not exactly match an existing node-exporter target metric. + # --------------------------------------------------------------------------- + - name: cpu_cores_num + exp: (system_cpu_logical_count.sum(['node_identifier_host_name'])).service(['node_identifier_host_name'], Layer.OS_LINUX) + - name: cpu_norm_percentage + exp: > + ( + ( + (system_cpu_utilization * 100) + .tagNotEqual('mode','idle') + .sum(['node_identifier_host_name']) + ) + / + system_cpu_logical_count.sum(['node_identifier_host_name']) + ).service(['node_identifier_host_name'], Layer.OS_LINUX)