Merge pull request #5897 from jcantrill/1506073_cpu_request_match_limit

Automatic merge from submit-queue.

bug 1506073. Lower cpu request for logging when it exceeds limit

This PR fixes https://bugzilla.redhat.com/show_bug.cgi?id=1506073 by:
* Lowering the CPU request to match the limit when the request is greater then a specified limit

I have an open question on if this is an acceptable change of if it makes the outcome unexpected.  Should we prefer to exit during the deployment and advise the operator to correct their inventory?

roles/openshift_logging/README.md

@@ -297,6 +297,8 @@ oc delete pod --selector=<ds_selector>
 
 Changelog
 ---------
+Tue Oct 26, 2017
+- Make CPU request equal limit if limit is greater then request
 
 Tue Oct 10, 2017
 - Default imagePullPolicy changed from Always to IfNotPresent 

roles/openshift_logging/filter_plugins/openshift_logging.py

@@ -3,6 +3,7 @@
 '''
 
 import random
+import re
 
 
 def es_storage(os_logging_facts, dc_name, pvc_claim, root='elasticsearch'):
@@ -17,6 +18,31 @@ def es_storage(os_logging_facts, dc_name, pvc_claim, root='elasticsearch'):
     return dict(kind='emptydir')
 
 
+def min_cpu(left, right):
+    '''Return the minimum cpu value of the two values given'''
+    message = "Unable to evaluate {} cpu value is specified correctly '{}'. Exp whole, decimal or int followed by M"
+    pattern = re.compile(r"^(\d*\.?\d*)([Mm])?$")
+    millis_per_core = 1000
+    if not right:
+        return left
+    m_left = pattern.match(left)
+    if not m_left:
+        raise RuntimeError(message.format("left", left))
+    m_right = pattern.match(right)
+    if not m_right:
+        raise RuntimeError(message.format("right", right))
+    left_value = float(m_left.group(1))
+    right_value = float(m_right.group(1))
+    if m_left.group(2) not in ["M", "m"]:
+        left_value = left_value * millis_per_core
+    if m_right.group(2) not in ["M", "m"]:
+        right_value = right_value * millis_per_core
+    response = left
+    if left_value != min(left_value, right_value):
+        response = right
+    return response
+
+
 def walk(source, path, default, delimiter='.'):
     '''Walk the sourch hash given the path and return the value or default if not found'''
     if not isinstance(source, dict):
@@ -87,6 +113,7 @@ class FilterModule(object):
             'random_word': random_word,
             'entry_from_named_pair': entry_from_named_pair,
             'map_from_pairs': map_from_pairs,
+            'min_cpu': min_cpu,
             'es_storage': es_storage,
             'serviceaccount_name': serviceaccount_name,
             'serviceaccount_namespace': serviceaccount_namespace,

roles/openshift_logging/filter_plugins/test

@@ -1,7 +1,22 @@
 import unittest
 from openshift_logging import walk
+from openshift_logging import min_cpu
 
 class TestFilterMethods(unittest.TestCase):
+    
+
+    def test_min_cpu_for_none(self):
+        source = "1000M"
+        self.assertEquals(min_cpu(source, None), "1000M")
+
+    def test_min_cpu_for_millis(self):
+        source = "1"
+        self.assertEquals(min_cpu(source, "0.1"), "0.1")
+
+
+    def test_min_cpu_for_whole(self):
+        source = "120M"
+        self.assertEquals(min_cpu(source, "2"), "120M")
 
 
     def test_walk_find_key(self):

roles/openshift_logging_curator/tasks/main.yaml

@@ -90,7 +90,7 @@
     es_host: "{{ openshift_logging_curator_es_host }}"
     es_port: "{{ openshift_logging_curator_es_port }}"
     curator_cpu_limit: "{{ openshift_logging_curator_cpu_limit }}"
-    curator_cpu_request: "{{ openshift_logging_curator_cpu_request }}"
+    curator_cpu_request: "{{ openshift_logging_curator_cpu_request | min_cpu(openshift_logging_curator_cpu_limit | default(none)) }}"
     curator_memory_limit: "{{ openshift_logging_curator_memory_limit }}"
     curator_replicas: "{{ openshift_logging_curator_replicas | default (1) }}"
     curator_node_selector: "{{openshift_logging_curator_nodeselector | default({})}}"

roles/openshift_logging_elasticsearch/tasks/main.yaml

@@ -354,7 +354,7 @@
     image: "{{ openshift_logging_elasticsearch_image_prefix }}logging-elasticsearch:{{ openshift_logging_elasticsearch_image_version }}"
     proxy_image: "{{ openshift_logging_elasticsearch_proxy_image_prefix }}oauth-proxy:{{ openshift_logging_elasticsearch_proxy_image_version }}"
     es_cpu_limit: "{{ openshift_logging_elasticsearch_cpu_limit | default('') }}"
-    es_cpu_request: "{{ openshift_logging_elasticsearch_cpu_request }}"
+    es_cpu_request: "{{ openshift_logging_elasticsearch_cpu_request | min_cpu(openshift_logging_elasticsearch_cpu_limit | default(none)) }}"
     es_memory_limit: "{{ openshift_logging_elasticsearch_memory_limit }}"
     es_node_selector: "{{ openshift_logging_elasticsearch_nodeselector | default({}) }}"
     es_storage_groups: "{{ openshift_logging_elasticsearch_storage_group | default([]) }}"

roles/openshift_logging_fluentd/tasks/main.yaml

@@ -154,7 +154,6 @@
       path: "{{ generated_certs_dir }}/system.logging.fluentd.crt"
 
 # create Fluentd daemonset
-
 # this should change based on the type of fluentd deployment to be done...
 # TODO: pass in aggregation configurations
 - name: Generate logging-fluentd daemonset definition
@@ -173,7 +172,7 @@
     fluentd_nodeselector_key: "{{ openshift_logging_fluentd_nodeselector.keys()[0] }}"
     fluentd_nodeselector_value: "{{ openshift_logging_fluentd_nodeselector.values()[0] }}"
     fluentd_cpu_limit: "{{ openshift_logging_fluentd_cpu_limit }}"
-    fluentd_cpu_request: "{{ openshift_logging_fluentd_cpu_request }}"
+    fluentd_cpu_request: "{{ openshift_logging_fluentd_cpu_request | min_cpu(openshift_logging_fluentd_cpu_limit | default(none)) }}"
     fluentd_memory_limit: "{{ openshift_logging_fluentd_memory_limit }}"
     audit_container_engine: "{{ openshift_logging_fluentd_audit_container_engine | default(False) | bool }}"
     audit_log_file: "{{ openshift_logging_fluentd_audit_file | default() }}"

roles/openshift_logging_kibana/tasks/main.yaml

@@ -230,10 +230,10 @@
     es_host: "{{ openshift_logging_kibana_es_host }}"
     es_port: "{{ openshift_logging_kibana_es_port }}"
     kibana_cpu_limit: "{{ openshift_logging_kibana_cpu_limit }}"
-    kibana_cpu_request: "{{ openshift_logging_kibana_cpu_request }}"
+    kibana_cpu_request: "{{ openshift_logging_kibana_cpu_request | min_cpu(openshift_logging_kibana_cpu_limit | default(none)) }}"
     kibana_memory_limit: "{{ openshift_logging_kibana_memory_limit }}"
     kibana_proxy_cpu_limit: "{{ openshift_logging_kibana_proxy_cpu_limit }}"
-    kibana_proxy_cpu_request: "{{ openshift_logging_kibana_proxy_cpu_request }}"
+    kibana_proxy_cpu_request: "{{ openshift_logging_kibana_proxy_cpu_request | min_cpu(openshift_logging_kibana_proxy_cpu_limit | default(none)) }}"
     kibana_proxy_memory_limit: "{{ openshift_logging_kibana_proxy_memory_limit }}"
     kibana_replicas: "{{ openshift_logging_kibana_replicas | default (1) }}"
     kibana_node_selector: "{{ openshift_logging_kibana_nodeselector | default({}) }}"

roles/openshift_logging_mux/tasks/main.yaml

@@ -171,7 +171,7 @@
     ops_host: "{{ openshift_logging_mux_ops_host }}"
     ops_port: "{{ openshift_logging_mux_ops_port }}"
     mux_cpu_limit: "{{ openshift_logging_mux_cpu_limit }}"
-    mux_cpu_request: "{{ openshift_logging_mux_cpu_request }}"
+    mux_cpu_request: "{{ openshift_logging_mux_cpu_request | min_cpu(openshift_logging_mux_cpu_limit | default(none)) }}"
     mux_memory_limit: "{{ openshift_logging_mux_memory_limit }}"
     mux_replicas: "{{ openshift_logging_mux_replicas | default(1) }}"
     mux_node_selector: "{{ openshift_logging_mux_nodeselector | default({}) }}"