Remove upstream prometheus cookbook, migrate to our own

site-cookbooks/kosmos_prometheus/recipes/server.rb

@@ -5,45 +5,101 @@
 
 include_recipe "firewall"
 
-prometheus_install "prometheus" do
-  version node["prometheus"]["version"]
-  checksum node["prometheus"]["checksum"]
+version = node["kosmos_prometheus"]["version"]
+checksum = node["kosmos_prometheus"]["checksum"]
+tarball = "#{Chef::Config[:file_cache_path]}/prometheus-#{version}.linux-amd64.tar.gz"
+binary_url = "https://github.com/prometheus/prometheus/releases/download/v#{version}/prometheus-#{version}.linux-amd64.tar.gz"
+
+group "prometheus"
+
+user "prometheus" do
+  gid "prometheus"
+  system true
+  shell "/bin/false"
+  home "/nonexistent"
 end
 
-prometheus_config "prometheus" do
-  global_config(
-    "scrape_interval" => "30s",
-    "evaluation_interval" => "30s"
+directory "/var/lib/prometheus" do
+  owner "prometheus"
+  group "prometheus"
+  mode "0755"
+  recursive true
+end
+
+directory "/etc/prometheus" do
+  owner "root"
+  group "root"
+  mode "0755"
+  recursive true
+end
+
+directory "/etc/prometheus/rules" do
+  owner "root"
+  group "root"
+  mode "0755"
+  recursive true
+end
+
+package %w(tar bzip2)
+
+remote_file tarball do
+  source binary_url
+  checksum checksum
+  action :create
+  notifies :run, "execute[install_prometheus]", :immediately
+end
+
+execute "install_prometheus" do
+  command "tar -xzf #{tarball} -C /usr/local/bin --strip-components=1 prometheus-#{version}.linux-amd64/prometheus"
+  action :nothing
+  notifies :restart, "service[prometheus]", :delayed
+end
+
+file "/usr/local/bin/prometheus" do
+  owner "root"
+  group "root"
+  mode "0755"
+  notifies :restart, "service[prometheus]", :delayed
+end
+
+template "/etc/prometheus/prometheus.yml" do
+  source "prometheus.yml.erb"
+  owner "root"
+  group "prometheus"
+  mode "0644"
+  variables(
+    global_config: node["kosmos_prometheus"]["global"],
+    jobs: node["kosmos_prometheus"]["jobs"],
+    rule_files: node["kosmos_prometheus"]["rule_files"]
   )
+  notifies :reload, "service[prometheus]", :delayed
 end
 
-prometheus_job "prometheus" do
-  target "localhost:9090"
-end
-
-prometheus_job "node" do
-  target "localhost:9100"
-end
-
-with_run_context :root do
-  execute "reload prometheus config" do
-    command "systemctl reload prometheus.service"
-    action :nothing
-    subscribes :run, "template[/opt/prometheus/prometheus.yml]", :delayed
-  end
-end
-
-prometheus_service "prometheus" do
-  cli_options({
-    "config.file" => "/opt/prometheus/prometheus.yml",
-    "log.level" => "info",
-    "query.max-concurrency" => 20,
-    "query.lookback-delta" => "5m",
-    "query.timeout" => "2m",
-    "storage.tsdb.path" => "/var/lib/prometheus",
-    "storage.tsdb.retention.time" => "15d",
-    "web.listen-address" => ":9090"
+systemd_unit "prometheus.service" do
+  content({
+    Unit: {
+      Description: "Prometheus",
+      After: "network.target",
+    },
+    Service: {
+      Type: "simple",
+      User: "prometheus",
+      Group: "prometheus",
+      ExecStart: "/usr/local/bin/prometheus --config.file=/etc/prometheus/prometheus.yml --storage.tsdb.path=/var/lib/prometheus --storage.tsdb.retention.time=15d --web.listen-address=:9090 --web.enable-lifecycle",
+      ExecReload: "/bin/kill -HUP $MAINPID",
+      Restart: "on-failure",
+      RestartSec: "5",
+    },
+    Install: {
+      WantedBy: "multi-user.target",
+    },
   })
+  triggers_reload true
+  action :create
+end
+
+service "prometheus" do
+  action [:enable, :start]
 end
 
 firewall_rule "prometheus web" do