Merge from ci_feature to ci_feature_prod

README.md

@@ -10,6 +10,13 @@ additional questions or comments.
 ## Release History
 
 Note : The agent version(s) below has dates (ciprod<mmddyyyy>), which indicate the agent build dates (not release dates)
+
+### 03/12/2019 - Version microsoft/oms:ciprod03122019
+- Fix for closing response.Body in outoms
+- Update Mem_Buf_Limit to 5m for fluentbit
+- Tail only files that were modified since 5 minutes
+- Remove some unwanted logs that are chatty in outoms
+- Fix for MDM disablement for AKS-Engine
 
 ### 02/21/2019 - Version microsoft/oms:ciprod02212019
 - Container logs enrichment optimization

installer/conf/td-agent-bit.conf

@@ -10,16 +10,17 @@
     Path /var/log/containers/*.log
     DB /var/log/omsagent-fblogs.db
     Parser docker
-    Mem_Buf_Limit 30m
+    Mem_Buf_Limit 5m
     Path_Key filepath
     Skip_Long_Lines On
+    Ignore_Older 5m
 
 [INPUT]
     Name tail
     Tag oms.container.log.flbplugin.*
     Path /var/log/containers/omsagent*.log
     DB /var/opt/microsoft/docker-cimprov/state/omsagent-ai.db
-    Mem_Buf_Limit 30m
+    Mem_Buf_Limit 2m
     Path_Key filepath
     Skip_Long_Lines On
 
@@ -28,6 +29,4 @@
     EnableTelemetry                 true
     TelemetryPushIntervalSeconds    300
     Match                           oms.container.log.*
-    AgentVersion                    ciprod02212019
-
-
+    AgentVersion                    ciprod03122019

source/code/go/src/plugins/oms.go

@@ -246,16 +246,11 @@ func PostDataHelper(tailPluginRecords []map[interface{}]interface{}) int {
 
 		if val, ok := imageIDMap[containerID]; ok {
 			stringMap["Image"] = val
-		} else {
-			Log("ContainerId %s not present in Name Map ", containerID)
-		}
+		} 
 
 		if val, ok := nameIDMap[containerID]; ok {
 			stringMap["Name"] = val
-		} else {
-			Log("ContainerId %s not present in Image Map ", containerID)
-		}
-
+		} 
 
 		dataItem := DataItem{
 			ID:                    stringMap["Id"],
@@ -319,6 +314,8 @@ func PostDataHelper(tailPluginRecords []map[interface{}]interface{}) int {
 			return output.FLB_RETRY
 		}
 
+		defer resp.Body.Close()
+
 		numRecords := len(dataItems)
 		Log("Successfully flushed %d records in %s", numRecords, elapsed)
 		ContainerLogTelemetryMutex.Lock()

source/code/plugin/CustomMetricsUtils.rb

@@ -9,8 +9,8 @@ class << self
         def check_custom_metrics_availability(custom_metric_regions)
             aks_region = ENV['AKS_REGION']
             aks_resource_id = ENV['AKS_RESOURCE_ID']
-            if aks_region.to_s.empty? && aks_resource_id.to_s.empty?
-                false # This will also take care of AKS-Engine Scenario. AKS_REGION/AKS_RESOURCE_ID is not set for AKS-Engine. Only ACS_RESOURCE_NAME is set
+            if aks_region.to_s.empty? || aks_resource_id.to_s.empty?
+                return false # This will also take care of AKS-Engine Scenario. AKS_REGION/AKS_RESOURCE_ID is not set for AKS-Engine. Only ACS_RESOURCE_NAME is set
             end
 
             custom_metrics_regions_arr = custom_metric_regions.split(',')

source/code/plugin/out_mdm.rb

@@ -29,6 +29,7 @@ def initialize
       @cached_access_token = String.new
       @last_post_attempt_time = Time.now
       @first_post_attempt_made = false
+      @can_send_data_to_mdm = true
     end
 
     def configure(conf)
@@ -39,7 +40,13 @@ def configure(conf)
 
     def start
       super
-      file = File.read(@@azure_json_path)
+      begin
+        file = File.read(@@azure_json_path)
+      rescue => e
+        @log.info "Unable to read file #{@@azure_json_path} #{e}"
+        @can_send_data_to_mdm = false
+        return
+      end
       # Handle the case where the file read fails. Send Telemetry and exit the plugin? 
       @data_hash = JSON.parse(file)
       @token_url = @@token_url_template % {tenant_id: @data_hash['tenantId']}
@@ -48,11 +55,13 @@ def start
       aks_region = ENV['AKS_REGION']
       if aks_resource_id.to_s.empty?
         @log.info "Environment Variable AKS_RESOURCE_ID is not set.. "
-        raise Exception.new "Environment Variable AKS_RESOURCE_ID is not set!!" 
+        @can_send_data_to_mdm = false
+        return
       end
       if aks_region.to_s.empty?
         @log.info "Environment Variable AKS_REGION is not set.. "
-        raise Exception.new "Environment Variable AKS_REGION is not set!!" 
+        @can_send_data_to_mdm = false
+        return
       end
 
       @@post_request_url = @@post_request_url_template % {aks_region: aks_region, aks_resource_id: aks_resource_id}
@@ -115,14 +124,18 @@ def format(tag, time, record)
     # 'chunk' is a buffer chunk that includes multiple formatted records
     def write(chunk)
       begin
-        if !@first_post_attempt_made || (Time.now > @last_post_attempt_time + retry_mdm_post_wait_minutes*60)
+        if (!@first_post_attempt_made || (Time.now > @last_post_attempt_time + retry_mdm_post_wait_minutes*60)) && @can_send_data_to_mdm
           post_body = []
           chunk.msgpack_each {|(tag, record)|
             post_body.push(record.to_json)
           }
           send_to_mdm post_body
         else
-          @log.info "Last Failed POST attempt to MDM was made #{((Time.now - @last_post_attempt_time)/60).round(1)} min ago. This is less than the current retry threshold of #{@retry_mdm_post_wait_minutes} min. NO-OP"
+          if !@can_send_data_to_mdm
+            @log.info "Cannot send data to MDM since all required conditions were not met"
+          else
+            @log.info "Last Failed POST attempt to MDM was made #{((Time.now - @last_post_attempt_time)/60).round(1)} min ago. This is less than the current retry threshold of #{@retry_mdm_post_wait_minutes} min. NO-OP"
+          end
         end
       rescue Exception => e
         @log.info "Exception when writing to MDM: #{e}"