v22.1.1

Upgrades

• pushgateway to v0.5.1

Deployment

releases:
  - name: prometheus
    version: 22.1.1
    url: https://github.com/bosh-prometheus/prometheus-boshrelease/releases/download/v22.1.1/prometheus-22.1.1.tgz
    sha1: ecfb666f3ef3241b28f9b47c6702fd1bfdd0f92a

v22.1.0

Features

• Add latency and error rate to UAA dashboard.
• Add client_header_buffers properties to nginx job
• Add cf_component_metrics_v2 grafana dashboard

Fixes

• Fix kube_state_metrics exporter property
• Fix a typo error at p_rabbitmq alerts
• Fix alertmanager route_prefix configuration in Prometheus config

Manifests

• Fix az name in the prometheus v1 migration ops file
• Make bosh_exporter UAA client_id fconfigurable via ops file
• Add op-file to allow configuring an external LB for the nginx job

Upgrades

• cadvisor to v0.29.0
• Grafana to v5.1.3
• postgres_exporter to v0.4.6
• pushgateway to v0.5.0
• rabbitmq_exporter to v0.28.0
• redis_exporter to v0.18.0
• Grafana Pie Chart plugin to v1.3.1

Deployment

releases:
  - name: prometheus
    version: 22.1.0
    url: https://github.com/bosh-prometheus/prometheus-boshrelease/releases/download/v22.1.0/prometheus-22.1.0.tgz
    sha1: 6715a8870656c86c5ad0604d6bf840e639115a5f

v22.0.2

Fixes

• Removes a deprecated kube-state-metrics flag

Deployment

releases:
  - name: prometheus
    version: 22.0.2
    url: https://github.com/bosh-prometheus/prometheus-boshrelease/releases/download/v22.0.2/prometheus-22.0.2.tgz
    sha1: 5c39bb9d1bb9923b4453551533ae4cf35f9cdbf5

v22.0.1

Yanked: please ignore the v22.0.1 bosh-release

v22.0.0

Breaking changes

Prometheus v2

Prometheus has been upgraded to v2, and contains a number of backwards incompatible changes:

• The recording and alerting rules have been migrated from a custom format to the ubiquitous YAML format. All alert rules in this BOSH release have been updated to the new format. If you are using your own rules and/or alerts (via the prometheus.custom_rules property or collocating your own BOSH release and adding them via the prometheus.rule_files property), you will need to adapt them to the new format. See this example for more information.

• Prometheus v2 comes with a new storage subsystem that is incompatible with the Prometheus v1 data format. To retain access to your historic monitoring data, the official recommendation is to run a non-scraping Prometheus instance running at least version 1.8.1 in parallel with your Prometheus 2.0 instance, and have the new server read existing data from the old one via the remote read protocol. A migrate_from_prometheus_1.yml op-file has been created based on this recommendation. Please follow these steps in order to upgrade your deployment to Prometheus v2 if you want to retain access to historical data:

  1. Upgrade your existing deployment to use the Prometheus BOSH release v21.1.1.
  2. After the above deployment succeeds, add the migrate_from_prometheus_1.yml op-file to your deployment. This will create a new prometheus2 instance running Prometheus v2 and all the exporters, and it will preserve the old prometheus instance updating it to be a non-scrapping Prometheus. The new prometheus2 instance will be automatically configured to use the prometheus instance as a remote read instance.
  3. Once the Prometheus v1 data is no longer relevant (typically after 15 days, unless you are using a custom prometheus.storage.local.retention property value or if you want to keep data longer than that), remove the migrate_from_prometheus_1.yml op-file from your deployment and the old prometheus instance (and the persistent disk holding the historical data) will be automatically deleted alongside the remote read.

Grafana v5

Grafana has been upgraded to v5. There are two new features that may affect dashboards, specially if you are using your own dashboards:

• The dashboard grid layout engine has changed. All dashboards will be automatically upgraded to a new positioning system when you load them in v5. Dashboards saved in v5 will not work in older versions of Grafana.

• Grafana allows organizing dashboards in folders. A new property in this BOSH release has been added to support this feature, and the manifest files have been updated accordingly to store all dashboards inside folders. If you are collocating dashboards from an external BOSH release and using the grafana.prometheus.dashboard_files property, all those dashboards will be located at the default folder, but if you prefer to use a different folder, adapt your manifest files to use the new property:

  grafana.prometheus.dashboard_folders:
    description: "Array of grafana folders and dashboard json file locations or glob patterns"
    example:
      - name: "My Dashboards"
        files:
          - "/var/vcap/packages/my_dashboards/*"
  

Misc

• The github_exporter has been removed

Features

• Upgraded to Prometheus v2
• Upgraded to Grafana v5
• Allow organizing grafana dashboard in folders
• Allow setting custom headers in nginx config
• Allow prometheus job to use a proxy
• Allow kube-state-metrics job to use a proxy
• Allow kube-state-metrics job to get properties from cfcf via BOSH links
• Updated kubernetes dashboards
• Added alerts for RabbitMQ for PCF queue depth
• Added dashboards compatible with loggregator version 101.4
• Service name for CF Apps alerts is now configurable

Fixes

• Fixes app request at cloudfoundry dashboards
• Fixes expr for unassigned shards and node count at elasticsearch dashboards
• Fixes a problem with a grafana script not being compatible with CentOS stemcells

Manifests

• Added migrate_from_prometheus_1.yml op-file to allows migrations from Prometheus v1
• All grafana dashboards are now store inside folders
• The alertmanager slack configuration now sends alerts when they are resolved
• monitor-kubernetes op file now enables kubernetes API servers and kubelets service discovery
• Added op files to enable proxy properties on several components
• cf_exporter client uses now read only privileges (cloud_controller.admin_read_only)
• Bump cf-routing release to v0.174.0
• Bump postgres relase to v28

Upgrades

• alertmanager to v0.14.0
• blackbox_exporter to v0.12.0
• bosh_exporter to v3.0.1
• cf_exporter to v0.6.2
• firehose_exporter to v5.0.3
• grafana to v5.0.4
• Grafana Status panel to v1.0.8
• Grafana Pie Chart panel to a07f6cf
• kube-state-metrics to v1.3.0
• memcached_exporter to v0.4.1
• pcre to v8.42
• postgres_exporter to v0.4.5
• prometheus to v2.2.1
• rabbitmq_exporter to 0.27.0
• redis_exporter to 0.17.2
• stackdriver_exporter to v0.5.1

Deployment

releases:
  - name: prometheus
    version: 22.0.0
    url: https://github.com/bosh-prometheus/prometheus-boshrelease/releases/download/v22.0.0/prometheus-22.0.0.tgz
    sha1: b2c97721e1f8db7ffce2f849e02afe62e3005de7

v21.1.1

Fixes

• Fixes a firehose_exporter problem when reporting too many metrics

Upgrades

• firehose_exporter to v5.0.1

Deployment

releases:
  - name: prometheus
    version: 21.1.1
    url: https://github.com/bosh-prometheus/prometheus-boshrelease/releases/download/v21.1.1/prometheus-21.1.1.tgz
    sha1: a680187c5fb5df9337657e41e9d11b178a43edfa

v21.1.0

Features

• Added basic grootfs alerts
• Use BOSH vendored golang-1.9 package

Fixes

• Fix CFLRPAuctionsFailedTooHigh alert
• Removed nginx URL path for grafana

Upgrades

• alertmanager to v0.13.0
• collectd_exporter to v0.4.0
• elasticsearch_exporter to v1.0.2
• haproxy_exporter to v0.9.0
• memcached_exporter to v0.4.0
• redis_exporter to v0.15.0
• statsd_exporter to v0.6.0

Deployment

releases:
  - name: prometheus
    version: 21.1.0
    url: https://github.com/bosh-prometheus/prometheus-boshrelease/releases/download/v21.1.0/prometheus-21.1.0.tgz
    sha1: ce6cb58b121a977de5d2d591e348683f3aa50450

v21.0.1

Fixes

• Fixes a problem with cf_exporter not being able to authenticate when UAA client ID and/or Secret contains special characters

Upgrades

• cf_exporter to v0.6.1

Deployment

releases:
  - name: prometheus
    version: 21.0.1
    url: https://github.com/bosh-prometheus/prometheus-boshrelease/releases/download/v21.0.1/prometheus-21.0.1.tgz
    sha1: f26a98b9005683b5080b0dfaa2ba52cb81a33735

v21.0.0

Breaking changes

• Control scripts no longer tees the output to syslog and the log files are now written inside the job's log directory.
• If you were rusing the enable-cf-route-registrar.yml op file, a new variable cf_deployment_name is now required

Features

• The test alert sent periodically (if enabled) will be tagged with the name TestAlert
• Updated concourse dashboards with new panels

Fixes

• Increased prometheus stop wait time to allow proper shut down
• Fixed typo at BOSHJobExtendedUnhealthy, BOSHJobProcessExtendedUnhealthy, and BOSHTSDBJobExtendedUnhealthy alert names
• Fixed ElasticHeapUsage alert summary and description
• Exclude ephemeral Pivotal Cloud Foundry Healthwatch vms from BOSH alerts and dashboards

Manifests

• prometheus.yml manifest file now uses persistent_disk instead of persistent_disk_types
• Added enable-prometheus-metrics.yml op file to be applied to your concourse-deployment to allow gathering metrics directly from concourse (>= v3.8.0)
• Added prometheus-web-external-url.yml op file to allow configuring the URL under which prometheus is externally reachable
• Fixed alertmanager-victorops-receiver.yml op file to require a routing_key variable
• Updated enable-cf-route-registrar.yml op file to require the cf_deployment_name variable (tipically cf but may change if you are using Pivotal Application Service)
• Removed postgres release when using use-sqlite3.yml op file
• Upgraded postgres release to v23
• Upgraded routing release to v0.169.0

Updates

• alertmanager to v0.12.0
• blackbox_exporter to v0.11.0
• bosh_exporter to v3.0.0
• bosh_tsdb_exporter to v0.2.0
• cadvisor to v0.28.3
• cf_exporter to v0.6.0
• firehose_exporter to v5.0.0
• grafana to v4.6.3
• grafana status panel to v1.0.7
• grafana diagram panel to v1.4.4
• grafana pie chart panel to v1.1.6
• postgres_exporter to v0.4.1
• rabbitmq_exporter to v0.25.2
• redis_exporter to v0.14
• shield_exporter to v0.3.0

Deployment

releases:
  - name: prometheus
    version: 21.0.0
    url: https://github.com/bosh-prometheus/prometheus-boshrelease/releases/download/v21.0.0/prometheus-21.0.0.tgz
    sha1: 606757c5cd5192e875768da3c581970d3146328c

v20.0.0

Breaking changes

• nginx auth configuration for alertmanager, grafana and prometheus has been converted to an array to allow configuring multiple users. Use

  • nginx.alertmanager.auth_users instead of nginx.alertmanager.auth_username and nginx.alertmanager.auth_password
  • nginx.grafana.auth_users instead of nginx.grafana.auth_username and nginx.grafana.auth_password
  • nginx.prometheus.auth_users instead of nginx.prometheus.auth_username and nginx.prometheus.auth_password

  An example configuration is:

  properties:
    nginx:
      alertmanager:
        auth_users:
          - name: admin
            password: ((alertmanager_password))
      grafana:
        auth_users:
          - name: admin
            password: ((grafana_password))
      prometheus:
        auth_users:
          - name: admin
            password: ((prometheus_password))
  

• statsd_exporter mapping configuration requires now a YAML format. You must convert your mappings configuration to the new format when you upgrade. Please refer to the statsd_exporter release notes.

Updates

• alertmanager to v0.11.0
• grafana to v4.6.2
• rabbitmq_exporter to v0.24.0
• statsd_exporter to v0.5.0

Deployment

releases:
- name: prometheus
  version: 20.0.0
  url: https://github.com/cloudfoundry-community/prometheus-boshrelease/releases/download/v20.0.0/prometheus-20.0.0.tgz
  sha1: ed8f42ab7dddc4b660a0f884748a80c2dfc470c7