Registry Migration - review and update retagger synced images - Atlas

[weatherhog]

For the first step of the registry migration story (see the RFC more detailed information) we need every team to review and update the retagger synced images that are used by the team. You can find the configures images here.

Therefore you need to do two things:

• remove images that are not needed anymore at all
• bump versions for images that are known to not use certain old versions any more

This will help us to keep the registry cleaner and in the end also cheaper.

[QuentinBisson]

List of atlas apps to check:

• kube-state-metrics/kube-state-metrics
• metrics-server/metrics-server
• prometheus-operator/prometheus-config-reloader:
• prometheus-operator/prometheus-operator:
• prometheus/alertmanager
• prometheus/node-exporter:
• prometheuscommunity/elasticsearch-exporter:
• thanos/thanos
• google-cloud-tools/kube-eagle
• coreos/prometheus-config-reloader:
• coreos/prometheus-operator
• coreos/configmap-reload:
• slok/sloth
• janeczku/go-dnsmasq:
• jgsqware/fluentd-loki-plugin:
• jollinshead/journald-cloudwatch-logs:
• k8spin/loki-multi-tenant-proxy:
• prom/cloudwatch-exporter:
• amazon/opendistro-for-elasticsearch:
• amazon/opendistro-for-elasticsearch-kibana:
• bats/bats: (used in grafana)
• fluent/fluent-bit:
• grafana/grafana:
• grafana/loki:
• grafana/loki-canary:
• grafana/mimir:
• grafana/mimir-continuous-test:
• grafana/promtail:
• grafana/tempo:
• jettech/kube-webhook-certgen: # Used in prometheus-operator-app
• jimmidyson/configmap-reload:
• justwatch/elasticsearch_exporter:
• kiwigrid/k8s-sidecar:
• memcached:
• omnition/opencensus-collector:
• opensearchproject/opensearch:
• opensearchproject/opensearch-dashboards:
• prom/memcached-exporter:
• prom/prometheus:
• prom/pushgateway:
• elasticsearch/elasticsearch-oss:
• kibana/kibana-oss:
• bitnami/memcached
• bitnami/memcached-exporter
• bobrik/curator
• caicloud/event-exporter
• grafana/agent
• fluent-bit-aws-plugins
• fluent-bit-systemd247-aws-plugins
• jaeger-all-in-one
• macropower-analytics-panel-server
• elasticsearch-untergeek-curator

[QuentinBisson]

Well that's a bit too much :D

[QuentinBisson]

I'd advocate that we use a recording rules to push all images in use in grafana

[QuentinBisson]

@giantswarm/team-atlas wdyt about:

label_replace(
  label_replace(
      count by (pod, image) (kube_pod_init_container_info{image=~".*giantswarm.*"})
    or
        count by (pod, image) (kube_pod_container_info{image=~".*giantswarm.*"})
      * on (pod) group_left (label_application_giantswarm_io_team)
        kube_pod_labels,
    "team",
    "$2",
    "label_application_giantswarm_io_team",
    "(team-)?(.*)"
  ),
  "label_application_giantswarm_io_team",
  "",
  "label_application_giantswarm_io_team",
  ".*"
)

[QuentinBisson]

See https://giantswarm.grafana.net/explore?panes=%7B%22-6v%22:%7B%22datasource%22:%22000000006%22,%22queries%22:%5B%7B%22refId%22:%22A%22,%22expr%22:%22count%28aggregation:container:images%29%20by%20%28image%29%22,%22range%22:true,%22instant%22:true,%22datasource%22:%7B%22type%22:%22prometheus%22,%22uid%22:%22000000006%22%7D,%22editorMode%22:%22code%22,%22legendFormat%22:%22__auto%22%7D%5D,%22range%22:%7B%22from%22:%22now-15m%22,%22to%22:%22now%22%7D%7D%7D&schemaVersion=1&orgId=1

[kubasobon]

👋 Hello friends, what's the status on this?

[QuantumEnigmaa]

I'm beginning to work on it right now using Quentin's dashboard

[QuentinBisson]

I did a lot already :D

[QuentinBisson]

@QuantumEnigmaa maybe you can double check giantswarm/retagger#941 ?

[QuentinBisson]

I should only be left with fluent-bits but your eyes would definitely help :)

[QuentinBisson]

Done :)