overhaul monitoring

This commit is contained in:
cventastic
2022-06-23 14:30:17 +02:00
parent a825b27ccd
commit 1b1913d335
8 changed files with 15 additions and 495 deletions

View File

@@ -52,8 +52,6 @@ services:
- 9090
ports:
- 127.0.0.1:9090:9090
networks:
- chains
labels:
org.label-schema.group: "monitoring"
depends_on:
@@ -87,6 +85,9 @@ services:
- 9100
labels:
org.label-schema.group: "monitoring"
- "prometheus-scrape.enabled=true"
- "prometheus-scrape.port=9100"
- "prometheus-scrape.job_name=nodeexporter"
cadvisor:
image: gcr.io/cadvisor/cadvisor:v0.42.0
@@ -105,6 +106,9 @@ services:
- 8080
labels:
org.label-schema.group: "monitoring"
- "prometheus-scrape.enabled=true"
- "prometheus-scrape.port=8080"
- "prometheus-scrape.job_name=cadvisor"
### VOLUMES

View File

@@ -32,94 +32,6 @@ services:
labels:
- "traefik.enable=true"
### WIREGUARD
wireguard:
image: lscr.io/linuxserver/wireguard
container_name: wireguard
cap_add:
- NET_ADMIN
- SYS_MODULE
environment:
- PUID=$PUID
- PGID=$PGID
volumes:
- ./wireguard/config/wg0.conf:/config/wg0.conf
- /lib/modules:/lib/modules
# Expose prometheus port
expose:
- 9090
ports:
- $SERVERPORT:$SERVERPORT/udp
sysctls:
- net.ipv4.conf.all.src_valid_mark=1
restart: unless-stopped
### MONITORING
prometheus:
image: prom/prometheus:v2.30.3
container_name: prometheus
volumes:
- ./prometheus:/etc/prometheus
- prometheus_data:/prometheus
command:
- '--config.file=/etc/prometheus/prometheus.yml'
- '--storage.tsdb.path=/prometheus'
- '--web.console.libraries=/etc/prometheus/console_libraries'
- '--web.console.templates=/etc/prometheus/consoles'
- '--storage.tsdb.retention.time=200h'
- '--web.enable-lifecycle'
restart: unless-stopped
network_mode: "service:wireguard"
labels:
org.label-schema.group: "monitoring"
depends_on:
- wireguard
nodeexporter:
image: prom/node-exporter:v1.2.2
container_name: nodeexporter
volumes:
- /proc:/host/proc:ro
- /sys:/host/sys:ro
- /:/rootfs:ro
command:
- '--path.procfs=/host/proc'
- '--path.rootfs=/rootfs'
- '--path.sysfs=/host/sys'
- '--collector.filesystem.mount-points-exclude=^/(sys|proc|dev|host|etc)($$|/)'
restart: unless-stopped
expose:
- 9100
labels:
org.label-schema.group: "monitoring"
cadvisor:
image: gcr.io/cadvisor/cadvisor:v0.42.0
container_name: cadvisor
privileged: true
devices:
- /dev/kmsg:/dev/kmsg
volumes:
- /:/rootfs:ro
- /var/run:/var/run:ro
- /sys:/sys:ro
- /var/lib/docker:/var/lib/docker:ro
#- /cgroup:/cgroup:ro #doesn't work on MacOS only for Linux
restart: unless-stopped
expose:
- 8080
labels:
org.label-schema.group: "monitoring"
pushgateway:
image: prom/pushgateway:v1.4.2
container_name: pushgateway
restart: unless-stopped
expose:
- 9091
labels:
org.label-schema.group: "monitoring"
### POKT
pocket-de-1:
build: pokt

View File

@@ -32,94 +32,6 @@ services:
labels:
- "traefik.enable=true"
### WIREGUARD
wireguard:
image: lscr.io/linuxserver/wireguard
container_name: wireguard
cap_add:
- NET_ADMIN
- SYS_MODULE
environment:
- PUID=$PUID
- PGID=$PGID
volumes:
- ./wireguard/config/wg0.conf:/config/wg0.conf
- /lib/modules:/lib/modules
# Expose prometheus port
expose:
- 9090
ports:
- $SERVERPORT:$SERVERPORT/udp
sysctls:
- net.ipv4.conf.all.src_valid_mark=1
restart: unless-stopped
### MONITORING
prometheus:
image: prom/prometheus:v2.30.3
container_name: prometheus
volumes:
- ./prometheus:/etc/prometheus
- prometheus_data:/prometheus
command:
- '--config.file=/etc/prometheus/prometheus.yml'
- '--storage.tsdb.path=/prometheus'
- '--web.console.libraries=/etc/prometheus/console_libraries'
- '--web.console.templates=/etc/prometheus/consoles'
- '--storage.tsdb.retention.time=200h'
- '--web.enable-lifecycle'
restart: unless-stopped
network_mode: "service:wireguard"
labels:
org.label-schema.group: "monitoring"
depends_on:
- wireguard
nodeexporter:
image: prom/node-exporter:v1.2.2
container_name: nodeexporter
volumes:
- /proc:/host/proc:ro
- /sys:/host/sys:ro
- /:/rootfs:ro
command:
- '--path.procfs=/host/proc'
- '--path.rootfs=/rootfs'
- '--path.sysfs=/host/sys'
- '--collector.filesystem.mount-points-exclude=^/(sys|proc|dev|host|etc)($$|/)'
restart: unless-stopped
expose:
- 9100
labels:
org.label-schema.group: "monitoring"
cadvisor:
image: gcr.io/cadvisor/cadvisor:v0.42.0
container_name: cadvisor
privileged: true
devices:
- /dev/kmsg:/dev/kmsg
volumes:
- /:/rootfs:ro
- /var/run:/var/run:ro
- /sys:/sys:ro
- /var/lib/docker:/var/lib/docker:ro
#- /cgroup:/cgroup:ro #doesn't work on MacOS only for Linux
restart: unless-stopped
expose:
- 8080
labels:
org.label-schema.group: "monitoring"
pushgateway:
image: prom/pushgateway:v1.4.2
container_name: pushgateway
restart: unless-stopped
expose:
- 9091
labels:
org.label-schema.group: "monitoring"
### POKT
pocket-fr-1:
build: pokt
@@ -146,6 +58,10 @@ services:
- "traefik.http.routers.pocket-fr-1.entrypoints=websecure"
- "traefik.http.routers.pocket-fr-1.tls.certresolver=myresolver"
- "traefik.http.routers.pocket-fr-1.rule=Host(`$POKT_FR_1_DOMAIN`) && Path(`/v1`, `/v1/client/{dispatch|relay|challenge|sim}`)"
- "prometheus-scrape.enabled=true"
- "prometheus-scrape.port=8083"
- "prometheus-scrape.port=26660"
- "prometheus-scrape.job_name=pocket-fr-1"
pocket-fr-2:
build: pokt
@@ -172,6 +88,10 @@ services:
- "traefik.http.routers.pocket-fr-2.entrypoints=websecure"
- "traefik.http.routers.pocket-fr-2.tls.certresolver=myresolver"
- "traefik.http.routers.pocket-fr-2.rule=Host(`$POKT_FR_2_DOMAIN`) && Path(`/v1`, `/v1/client/{dispatch|relay|challenge|sim}`)"
- "prometheus-scrape.enabled=true"
- "prometheus-scrape.port=8083"
- "prometheus-scrape.port=26660"
- "prometheus-scrape.job_name=pocket-fr-2"
haproxy:
build: haproxy
@@ -194,5 +114,4 @@ volumes:
pocket-mainnet:
pocket-mainnet-addrbook:
pocket-fr-2:
pocket-fr-2-addrbook:
prometheus_data:
pocket-fr-2-addrbook:

View File

@@ -1,65 +0,0 @@
version: '3.1'
services:
#pocket-mainnet:
# image: poktnetwork/pocket-core:stagenet-latest
# ports:
# #- "127.0.0.1:8081:8081"
# - "26656:26656"
# expose:
# - 26656
# - 8081
# command: /home/app/.pocket/pokt_mainnet.sh && pocket start --seeds=$POCKET_MAIN_SEEDS --mainnet
# #command: pocket start --simulateRelay
# environment:
# - POCKET_CORE_KEY=$POCKET_CORE_KEY
# - POCKET_CORE_PASSPHRASE=$POCKET_CORE_PASSPHRASE
# - POCKET_SNAPSHOT=$POCKET_SNAPSHOT
# volumes:
# - ./bootstrap_skript/pokt_mainnet.sh:/home/app/.pocket/pokt_mainnet.sh
# - ./chains/chains_mainnet.json:/home/app/.pocket/config/chains.json
# - ./pokt/config.json:/home/app/.pocket/config/config.json
# - pocket-mainnet:/home/app/.pocket
# networks:
# - chains
# labels:
# - "traefik.enable=true"
# - "traefik.http.middlewares.pocket-mainnet-stripprefix.stripprefix.prefixes=/pokt"
# - "traefik.http.services.pocket-mainner.loadbalancer.server.port=8081"
# - "traefik.http.routers.pocket-mainnet.entrypoints=websecure"
# - "traefik.http.routers.pocket-mainnet.tls.certresolver=myresolver"
# - "traefik.http.routers.pocket-mainnet.rule=Host(`$DOMAIN`) && PathPrefix(`/pokt`)"
# - "traefik.http.routers.pocket-mainnet.middlewares=pocket-mainnet-stripprefix, ipwhitelist"
# restart: unless-stopped
pocket-mainnet:
build: pokt
ports:
- "127.0.0.1:8081:8081"
- "26657:26656"
expose:
- 26656
- 26660
- 8081
- 8083
environment:
- POCKET_CORE_KEY=$POCKET_CORE_KEY
- POCKET_CORE_PASSPHRASE=$POCKET_CORE_PASSPHRASE
- POCKET_SNAPSHOT=$POCKET_SNAPSHOT
volumes:
- pocket-mainnet:/home/app/.pocket/data
- pocket-mainnet-config:/home/app/.pocket/config
restart: unless-stopped
labels:
- "traefik.enable=true"
- "traefik.http.middlewares.pocket-mainnet-stripprefix.stripprefix.prefixes=/pokt"
- "traefik.http.services.pocket-mainner.loadbalancer.server.port=8081"
- "traefik.http.routers.pocket-mainnet.entrypoints=websecure"
- "traefik.http.routers.pocket-mainnet.tls.certresolver=myresolver"
- "traefik.http.routers.pocket-mainnet.rule=Host(`$DOMAIN`) && PathPrefix(`/pokt`)"
- "traefik.http.routers.pocket-mainnet.middlewares=pocket-mainnet-stripprefix, ipwhitelist"
volumes:
pocket-mainnet:
pocket-mainnet-config:

View File

@@ -1,41 +0,0 @@
global:
scrape_interval: 15s
evaluation_interval: 15s
# Attach these labels to any time series or alerts when communicating with
# external systems (federation, remote storage, Alertmanager).
external_labels:
monitor: 'docker-host-alpha'
# Load and evaluate rules in this file every 'evaluation_interval' seconds.
rule_files:
- "alert.rules"
# A scrape configuration containing exactly one endpoint to scrape.
scrape_configs:
- job_name: 'nodeexporter'
scrape_interval: 5s
static_configs:
- targets: ['nodeexporter:9100']
- job_name: 'cadvisor'
scrape_interval: 5s
static_configs:
- targets: ['cadvisor:8080']
- job_name: 'prometheus'
scrape_interval: 10s
static_configs:
- targets: ['localhost:9090']
- job_name: 'harmony-mainnet'
scrape_interval: 10s
honor_labels: true
static_configs:
- targets: ['harmony-mainnet:9900']
- job_name: 'traefik'
scrape_interval: 10s
honor_labels: true
static_configs:
- targets: ['traefik:8082']

View File

@@ -1,69 +0,0 @@
global:
scrape_interval: 15s
evaluation_interval: 15s
# Attach these labels to any time series or alerts when communicating with
# external systems (federation, remote storage, Alertmanager).
external_labels:
monitor: 'docker-host-alpha'
# Load and evaluate rules in this file every 'evaluation_interval' seconds.
rule_files:
- "alert.rules"
# A scrape configuration containing exactly one endpoint to scrape.
scrape_configs:
- job_name: 'nodeexporter'
scrape_interval: 5s
static_configs:
- targets: ['nodeexporter:9100']
- job_name: 'cadvisor'
scrape_interval: 5s
static_configs:
- targets: ['cadvisor:8080']
- job_name: 'prometheus'
scrape_interval: 10s
static_configs:
- targets: ['localhost:9090']
- job_name: 'pushgateway'
scrape_interval: 10s
honor_labels: true
static_configs:
- targets: ['pushgateway:9091']
- job_name: 'pocket-lt-1'
scrape_interval: 10s
honor_labels: true
static_configs:
- targets: ['pocket-lt-1:8083']
- targets: ['pocket-lt-1:26660']
- job_name: 'pocket-lt-2'
scrape_interval: 10s
honor_labels: true
static_configs:
- targets: ['pocket-lt-2:8083']
- targets: ['pocket-lt-2:26660']
- job_name: 'pocket-lt-3'
scrape_interval: 10s
honor_labels: true
static_configs:
- targets: ['pocket-lt-3:8083']
- targets: ['pocket-lt-3:26660']
- job_name: 'pocket-lt-4'
scrape_interval: 10s
honor_labels: true
static_configs:
- targets: ['pocket-lt-4:8083']
- targets: ['pocket-lt-4:26660']
- job_name: 'traefik'
scrape_interval: 10s
honor_labels: true
static_configs:
- targets: ['traefik:8082']

View File

@@ -1,85 +0,0 @@
global:
scrape_interval: 15s
evaluation_interval: 15s
# Attach these labels to any time series or alerts when communicating with
# external systems (federation, remote storage, Alertmanager).
external_labels:
monitor: 'docker-host-alpha'
# Load and evaluate rules in this file every 'evaluation_interval' seconds.
rule_files:
- "alert.rules"
# A scrape configuration containing exactly one endpoint to scrape.
scrape_configs:
- job_name: 'nodeexporter'
scrape_interval: 5s
static_configs:
- targets: ['nodeexporter:9100']
- job_name: 'cadvisor'
scrape_interval: 5s
static_configs:
- targets: ['cadvisor:8080']
- job_name: 'prometheus'
scrape_interval: 10s
static_configs:
- targets: ['localhost:9090']
- job_name: 'pushgateway'
scrape_interval: 10s
honor_labels: true
static_configs:
- targets: ['pushgateway:9091']
- job_name: 'traefik'
scrape_interval: 10s
honor_labels: true
static_configs:
- targets: ['traefik:8082']
- job_name: 'geth-mainnet'
scrape_interval: 10s
static_configs:
- targets: ['geth-mainnet:6060']
relabel_configs:
- source_labels: [__address__]
action: replace
regex: (.*)
replacement: $1
target_label: geth_mainnet
- job_name: 'geth-goerli'
scrape_interval: 10s
static_configs:
- targets: ['geth-goerli:6060']
relabel_configs:
- source_labels: [__address__]
action: replace
regex: (.*)
replacement: $1
target_label: geth_goerli
- job_name: 'geth-rinkeby'
scrape_interval: 10s
static_configs:
- targets: ['geth-rinkeby:6060']
relabel_configs:
- source_labels: [__address__]
action: replace
regex: (.*)
replacement: $1
target_label: geth_rinkeby
- job_name: 'geth-ropsten'
scrape_interval: 10s
static_configs:
- targets: ['geth-ropsten:6060']
relabel_configs:
- source_labels: [__address__]
action: replace
regex: (.*)
replacement: $1
target_label: geth_ropsten

View File

@@ -1,55 +0,0 @@
global:
scrape_interval: 15s
evaluation_interval: 15s
# Attach these labels to any time series or alerts when communicating with
# external systems (federation, remote storage, Alertmanager).
external_labels:
monitor: 'docker-host-alpha'
# Load and evaluate rules in this file every 'evaluation_interval' seconds.
rule_files:
- "alert.rules"
# A scrape configuration containing exactly one endpoint to scrape.
scrape_configs:
- job_name: 'nodeexporter'
scrape_interval: 5s
static_configs:
- targets: ['nodeexporter:9100']
- job_name: 'cadvisor'
scrape_interval: 5s
static_configs:
- targets: ['cadvisor:8080']
- job_name: 'prometheus'
scrape_interval: 10s
static_configs:
- targets: ['localhost:9090']
- job_name: 'pushgateway'
scrape_interval: 10s
honor_labels: true
static_configs:
- targets: ['pushgateway:9091']
- job_name: 'pocket-fr-1'
scrape_interval: 10s
honor_labels: true
static_configs:
- targets: ['pocket-fr-1:8083']
- targets: ['pocket-fr-1:26660']
- job_name: 'pocket-fr-2'
scrape_interval: 10s
honor_labels: true
static_configs:
- targets: ['pocket-fr-2:8083']
- targets: ['pocket-fr-2:26660']
- job_name: 'traefik'
scrape_interval: 10s
honor_labels: true
static_configs:
- targets: ['traefik:8082']