Merge pull request #5748 from portante/fix-cpu-limits

Use "requests" for CPU resources instead of limits

3 files changed, 36 insertions, 21 deletions

diff --git a/roles/openshift_logging/README.md b/roles/openshift_logging/README.md
index 69eb9283d..280d7d24c 100644
--- a/roles/openshift_logging/README.md
+++ b/roles/openshift_logging/README.md
@@ -44,23 +44,23 @@ When `openshift_logging_install_logging` is set to `False` the `openshift_loggin
 - `openshift_logging_curator_run_timezone`: The timezone that Curator uses for figuring out its run time. Defaults to 'UTC'.
 - `openshift_logging_curator_script_log_level`: The script log level for Curator. Defaults to 'INFO'.
 - `openshift_logging_curator_log_level`: The log level for the Curator process. Defaults to 'ERROR'.
-- `openshift_logging_curator_cpu_limit`: The amount of CPU to allocate to Curator. Default is '100m'.
+- `openshift_logging_curator_cpu_request`: The minimum amount of CPU to allocate to Curator. Default is '100m'.
 - `openshift_logging_curator_memory_limit`: The amount of memory to allocate to Curator. Unset if not specified.
 - `openshift_logging_curator_nodeselector`: A map of labels (e.g. {"node":"infra","region":"west"} to select the nodes where the curator pod will land.
 - `openshift_logging_image_pull_secret`: The name of an existing pull secret to link to the logging service accounts
 
 - `openshift_logging_kibana_hostname`: The Kibana hostname. Defaults to 'kibana.example.com'.
-- `openshift_logging_kibana_cpu_limit`: The amount of CPU to allocate to Kibana or unset if not specified.
+- `openshift_logging_kibana_cpu_request`: The minimum amount of CPU to allocate to Kibana or unset if not specified.
 - `openshift_logging_kibana_memory_limit`: The amount of memory to allocate to Kibana or unset if not specified.
 - `openshift_logging_kibana_proxy_debug`: When "True", set the Kibana Proxy log level to DEBUG. Defaults to 'false'.
-- `openshift_logging_kibana_proxy_cpu_limit`: The amount of CPU to allocate to Kibana proxy or unset if not specified.
+- `openshift_logging_kibana_proxy_cpu_request`: The minimum amount of CPU to allocate to Kibana proxy or unset if not specified.
 - `openshift_logging_kibana_proxy_memory_limit`: The amount of memory to allocate to Kibana proxy or unset if not specified.
 - `openshift_logging_kibana_replica_count`: The number of replicas Kibana should be scaled up to. Defaults to 1.
 - `openshift_logging_kibana_nodeselector`: A map of labels (e.g. {"node":"infra","region":"west"} to select the nodes where the pod will land.
 - `openshift_logging_kibana_edge_term_policy`: Insecure Edge Termination Policy. Defaults to Redirect.
 
 - `openshift_logging_fluentd_nodeselector`: The node selector that the Fluentd daemonset uses to determine where to deploy to. Defaults to '"logging-infra-fluentd": "true"'.
-- `openshift_logging_fluentd_cpu_limit`: The CPU limit for Fluentd pods. Defaults to '100m'.
+- `openshift_logging_fluentd_cpu_request`: The minimum amount of CPU to allocate for Fluentd collector pods. Defaults to '100m'.
 - `openshift_logging_fluentd_memory_limit`: The memory limit for Fluentd pods. Defaults to '512Mi'.
 - `openshift_logging_fluentd_use_journal`: *DEPRECATED - DO NOT USE* Fluentd will automatically detect whether or not Docker is using the journald log driver.
 - `openshift_logging_fluentd_journal_read_from_head`: If empty, Fluentd will use its internal default, which is false.
@@ -80,7 +80,7 @@ When `openshift_logging_install_logging` is set to `False` the `openshift_loggin
 - `openshift_logging_es_client_key`: The location of the client key Fluentd uses for openshift_logging_es_host. Defaults to '/etc/fluent/keys/key'.
 
 - `openshift_logging_es_cluster_size`: The number of ES cluster members. Defaults to '1'.
-- `openshift_logging_es_cpu_limit`:  The amount of CPU limit for the ES cluster.  Unused if not set
+- `openshift_logging_es_cpu_request`: The minimum amount of CPU to allocate for an ES pod cluster member. Defaults to 1 CPU.
 - `openshift_logging_es_memory_limit`: The amount of RAM that should be assigned to ES. Defaults to '8Gi'.
 - `openshift_logging_es_log_appenders`: The list of rootLogger appenders for ES logs which can be: 'file', 'console'. Defaults to 'file'.
 - `openshift_logging_es_pv_selector`: A key/value map added to a PVC in order to select specific PVs.  Defaults to 'None'.
@@ -107,7 +107,7 @@ same as above for their non-ops counterparts, but apply to the OPS cluster insta
 - `openshift_logging_es_ops_client_cert`: /etc/fluent/keys/cert
 - `openshift_logging_es_ops_client_key`: /etc/fluent/keys/key
 - `openshift_logging_es_ops_cluster_size`: 1
-- `openshift_logging_es_ops_cpu_limit`:  The amount of CPU limit for the ES cluster.  Unused if not set
+- `openshift_logging_es_ops_cpu_request`: The minimum amount of CPU to allocate for an ES ops pod cluster member. Defaults to 1 CPU.
 - `openshift_logging_es_ops_memory_limit`: 8Gi
 - `openshift_logging_es_ops_pvc_dynamic`: False
 - `openshift_logging_es_ops_pvc_size`: ""
@@ -115,9 +115,9 @@ same as above for their non-ops counterparts, but apply to the OPS cluster insta
 - `openshift_logging_es_ops_recover_after_time`: 5m
 - `openshift_logging_es_ops_storage_group`: 65534
 - `openshift_logging_kibana_ops_hostname`: The Operations Kibana hostname. Defaults to 'kibana-ops.example.com'.
-- `openshift_logging_kibana_ops_cpu_limit`: The amount of CPU to allocate to Kibana or unset if not specified.
+- `openshift_logging_kibana_ops_cpu_request`: The minimum amount of CPU to allocate to Kibana or unset if not specified.
 - `openshift_logging_kibana_ops_memory_limit`: The amount of memory to allocate to Kibana or unset if not specified.
-- `openshift_logging_kibana_ops_proxy_cpu_limit`: The amount of CPU to allocate to Kibana proxy or unset if not specified.
+- `openshift_logging_kibana_ops_proxy_cpu_request`: The minimum amount of CPU to allocate to Kibana proxy or unset if not specified.
 - `openshift_logging_kibana_ops_proxy_memory_limit`: The amount of memory to allocate to Kibana proxy or unset if not specified.
 - `openshift_logging_kibana_ops_replica_count`: The number of replicas Kibana ops should be scaled up to. Defaults to 1.
 
@@ -176,7 +176,7 @@ Elasticsearch OPS too, if using an OPS cluster:
   clients will use to connect to mux, and will be used in the TLS server cert
   subject.
 - `openshift_logging_mux_port`: 24284
-- `openshift_logging_mux_cpu_limit`: 100m
+- `openshift_logging_mux_cpu_request`: 100m
 - `openshift_logging_mux_memory_limit`: 512Mi
 - `openshift_logging_mux_default_namespaces`: Default `["mux-undefined"]` - the
  first value in the list is the namespace to use for undefined projects,
diff --git a/roles/openshift_logging/defaults/main.yml b/roles/openshift_logging/defaults/main.yml
index 6aff89dbf..626732d16 100644
--- a/roles/openshift_logging/defaults/main.yml
+++ b/roles/openshift_logging/defaults/main.yml
@@ -18,20 +18,24 @@ openshift_logging_curator_run_minute: 0
 openshift_logging_curator_run_timezone: UTC
 openshift_logging_curator_script_log_level: INFO
 openshift_logging_curator_log_level: ERROR
-openshift_logging_curator_cpu_limit: 100m
-openshift_logging_curator_memory_limit: null
+openshift_logging_curator_cpu_limit: null
+openshift_logging_curator_memory_limit: 256Mi
+openshift_logging_curator_cpu_request: 100m
 openshift_logging_curator_nodeselector: {}
 
-openshift_logging_curator_ops_cpu_limit: 100m
-openshift_logging_curator_ops_memory_limit: null
+openshift_logging_curator_ops_cpu_limit: null
+openshift_logging_curator_ops_memory_limit: 256Mi
+openshift_logging_curator_ops_cpu_request: 100m
 openshift_logging_curator_ops_nodeselector: {}
 
 openshift_logging_kibana_hostname: "{{ 'kibana.' ~ (openshift_master_default_subdomain | default('router.default.svc.cluster.local', true)) }}"
 openshift_logging_kibana_cpu_limit: null
 openshift_logging_kibana_memory_limit: 736Mi
+openshift_logging_kibana_cpu_request: 100m
 openshift_logging_kibana_proxy_debug: false
 openshift_logging_kibana_proxy_cpu_limit: null
 openshift_logging_kibana_proxy_memory_limit: 256Mi
+openshift_logging_kibana_proxy_cpu_request: 100m
 openshift_logging_kibana_replica_count: 1
 openshift_logging_kibana_edge_term_policy: Redirect
 
@@ -53,9 +57,11 @@ openshift_logging_kibana_ca: ""
 openshift_logging_kibana_ops_hostname: "{{ 'kibana-ops.' ~ (openshift_master_default_subdomain | default('router.default.svc.cluster.local', true)) }}"
 openshift_logging_kibana_ops_cpu_limit: null
 openshift_logging_kibana_ops_memory_limit: 736Mi
+openshift_logging_kibana_ops_cpu_request: 100m
 openshift_logging_kibana_ops_proxy_debug: false
 openshift_logging_kibana_ops_proxy_cpu_limit: null
 openshift_logging_kibana_ops_proxy_memory_limit: 256Mi
+openshift_logging_kibana_ops_proxy_cpu_request: 100m
 openshift_logging_kibana_ops_replica_count: 1
 
 #The absolute path on the control node to the cert file to use
@@ -71,13 +77,14 @@ openshift_logging_kibana_ops_key: ""
 openshift_logging_kibana_ops_ca: ""
 
 openshift_logging_fluentd_nodeselector: {'logging-infra-fluentd': 'true'}
-openshift_logging_fluentd_cpu_limit: 100m
+openshift_logging_fluentd_cpu_limit: null
 openshift_logging_fluentd_memory_limit: 512Mi
+openshift_logging_fluentd_cpu_request: 100m
 openshift_logging_fluentd_journal_source: ""
 openshift_logging_fluentd_journal_read_from_head: ""
 openshift_logging_fluentd_hosts: ['--all']
-openshift_logging_fluentd_buffer_queue_limit: 1024
-openshift_logging_fluentd_buffer_size_limit: 1m
+openshift_logging_fluentd_buffer_queue_limit: 32
+openshift_logging_fluentd_buffer_size_limit: 8m
 
 openshift_logging_es_host: logging-es
 openshift_logging_es_port: 9200
@@ -85,7 +92,8 @@ openshift_logging_es_ca: /etc/fluent/keys/ca
 openshift_logging_es_client_cert: /etc/fluent/keys/cert
 openshift_logging_es_client_key: /etc/fluent/keys/key
 openshift_logging_es_cluster_size: 1
-openshift_logging_es_cpu_limit: 1000m
+openshift_logging_es_cpu_limit: null
+openshift_logging_es_cpu_request: "1"
 # the logging appenders for the root loggers to write ES logs. Valid values: 'file', 'console'
 openshift_logging_es_log_appenders: ['file']
 openshift_logging_es_memory_limit: "8Gi"
@@ -124,8 +132,9 @@ openshift_logging_es_ops_ca: /etc/fluent/keys/ca
 openshift_logging_es_ops_client_cert: /etc/fluent/keys/cert
 openshift_logging_es_ops_client_key: /etc/fluent/keys/key
 openshift_logging_es_ops_cluster_size: "{{ openshift_logging_elasticsearch_ops_cluster_size | default(1) }}"
-openshift_logging_es_ops_cpu_limit: 1000m
-openshift_logging_es_ops_memory_limit: "8Gi"
+openshift_logging_es_ops_cpu_limit: null
+openshift_logging_es_ops_memory_limit: 8Gi
+openshift_logging_es_ops_cpu_request: "1"
 openshift_logging_es_ops_pv_selector: "{{ openshift_loggingops_storage_labels | default('') }}"
 openshift_logging_es_ops_pvc_dynamic: "{{ openshift_logging_elasticsearch_ops_pvc_dynamic | default(False) }}"
 openshift_logging_es_ops_pvc_size: "{{ openshift_logging_elasticsearch_ops_pvc_size | default('') }}"
@@ -158,8 +167,9 @@ openshift_logging_mux_allow_external: False
 openshift_logging_use_mux: "{{ openshift_logging_mux_allow_external | default(False) }}"
 openshift_logging_mux_hostname: "{{ 'mux.' ~ (openshift_master_default_subdomain | default('router.default.svc.cluster.local', true)) }}"
 openshift_logging_mux_port: 24284
-openshift_logging_mux_cpu_limit: 500m
-openshift_logging_mux_memory_limit: 1Gi
+openshift_logging_mux_cpu_limit: null
+openshift_logging_mux_memory_limit: 512Mi
+openshift_logging_mux_cpu_request: 100m
 # the namespace to use for undefined projects should come first, followed by any
 # additional namespaces to create by default - users will typically not need to set this
 openshift_logging_mux_default_namespaces: ["mux-undefined"]
diff --git a/roles/openshift_logging/tasks/install_logging.yaml b/roles/openshift_logging/tasks/install_logging.yaml
index efc3b97aa..76627acf2 100644
--- a/roles/openshift_logging/tasks/install_logging.yaml
+++ b/roles/openshift_logging/tasks/install_logging.yaml
@@ -134,6 +134,7 @@
     openshift_logging_elasticsearch_pvc_pv_selector: "{{ openshift_logging_es_ops_pv_selector }}"
     openshift_logging_elasticsearch_memory_limit: "{{ openshift_logging_es_ops_memory_limit }}"
     openshift_logging_elasticsearch_cpu_limit: "{{ openshift_logging_es_ops_cpu_limit }}"
+    openshift_logging_elasticsearch_cpu_request: "{{ openshift_logging_es_ops_cpu_request }}"
     openshift_logging_elasticsearch_nodeselector: "{{ openshift_logging_es_ops_nodeselector if item.0.nodeSelector | default(None) is none else item.0.nodeSelector }}"
     openshift_logging_elasticsearch_storage_group: "{{ [openshift_logging_es_ops_storage_group] if item.0.storageGroups | default([]) | length == 0 else item.0.storageGroups }}"
     openshift_logging_es_key: "{{ openshift_logging_es_ops_key }}"
@@ -171,6 +172,7 @@
     openshift_logging_elasticsearch_pvc_pv_selector: "{{ openshift_logging_es_ops_pv_selector }}"
     openshift_logging_elasticsearch_memory_limit: "{{ openshift_logging_es_ops_memory_limit }}"
     openshift_logging_elasticsearch_cpu_limit: "{{ openshift_logging_es_ops_cpu_limit }}"
+    openshift_logging_elasticsearch_cpu_request: "{{ openshift_logging_es_ops_cpu_request }}"
     openshift_logging_elasticsearch_nodeselector: "{{ openshift_logging_es_ops_nodeselector }}"
     openshift_logging_es_key: "{{ openshift_logging_es_ops_key }}"
     openshift_logging_es_cert: "{{ openshift_logging_es_ops_cert }}"
@@ -211,11 +213,13 @@
     openshift_logging_kibana_es_port: "{{ openshift_logging_es_ops_port }}"
     openshift_logging_kibana_nodeselector: "{{ openshift_logging_kibana_ops_nodeselector }}"
     openshift_logging_kibana_cpu_limit: "{{ openshift_logging_kibana_ops_cpu_limit }}"
+    openshift_logging_kibana_cpu_request: "{{ openshift_logging_kibana_ops_cpu_request }}"
     openshift_logging_kibana_memory_limit: "{{ openshift_logging_kibana_ops_memory_limit }}"
     openshift_logging_kibana_hostname: "{{ openshift_logging_kibana_ops_hostname }}"
     openshift_logging_kibana_replicas: "{{ openshift_logging_kibana_ops_replica_count }}"
     openshift_logging_kibana_proxy_debug: "{{ openshift_logging_kibana_ops_proxy_debug }}"
     openshift_logging_kibana_proxy_cpu_limit: "{{ openshift_logging_kibana_ops_proxy_cpu_limit }}"
+    openshift_logging_kibana_proxy_cpu_request: "{{ openshift_logging_kibana_ops_proxy_cpu_request }}"
     openshift_logging_kibana_proxy_memory_limit: "{{ openshift_logging_kibana_ops_proxy_memory_limit }}"
     openshift_logging_kibana_cert: "{{ openshift_logging_kibana_ops_cert }}"
     openshift_logging_kibana_key: "{{ openshift_logging_kibana_ops_key }}"
@@ -247,6 +251,7 @@
     openshift_logging_curator_master_url: "{{ openshift_logging_master_url }}"
     openshift_logging_curator_image_pull_secret: "{{ openshift_logging_image_pull_secret }}"
     openshift_logging_curator_cpu_limit: "{{ openshift_logging_curator_ops_cpu_limit }}"
+    openshift_logging_curator_cpu_request: "{{ openshift_logging_curator_ops_cpu_request }}"
     openshift_logging_curator_memory_limit: "{{ openshift_logging_curator_ops_memory_limit }}"
     openshift_logging_curator_nodeselector: "{{ openshift_logging_curator_ops_nodeselector }}"
   when: