Introduction

TAD (Test And Diagnostic) monitors the amount of free memory available in the system at run time. It triggers the memory/Resource Reclamation (RR) process in TDM, when the amount of free memory drops below a configurable threshold. RR process can also be triggered by memory allocation failures which results in a notification being sent to TDM to try to reclaim memory.

Selfheal is another feature implemented in Test And Diagnostic Component.

Self-heal monitors:

• CPU usage
• Memory Usage
• Critical RDK-B processes

Self-heal stores Reset Count and Reboot Count.
Self-heal takes required action like: Rebooting the device, Restarting required process based on predefined conditions.
Self-heal does connectivity test.

Feature

Selfheal – Resource Monitoring

Monitors the resources periodically (eg: 15 mins). If "Average Memory Used" reaches threshold value, necessary action will be executed. 

"resource_monitor.sh" script is used for monitoring Memory and CPU usage.
Located in the device path: "/fss/gw/usr/ccsp/tad/resource_monitor.sh".

Selfheal – Process Monitoring

Monitors the processes periodically (eg: 15 mins) based on process id (pid). Based on the process id availability, required action will be taken such as restarting the process, rebooting the device.

"task_health_monitor.sh" script is used for monitoring RDK-B processes. This is located at path: "/fss/gw/usr/ccsp/tad/task_health_monitor.sh". We can monitor any RDK-B processes by adding the process pid in this script.

Self-heal stores Reset Count and Reboot Count

Selfheal – Connectivity Test

Self-heal does connectivity test. Ping test will be done through server IP/URI (this needs to be configured). If server IP/URI is not configured, Ping test won't be executed and no action will be taken. If server is configured and ping test fails, reboot action will be executed.

"self_heal_connectivity_test.sh" script is used for ping test

Selfheal – Action

Self-heal takes the required action through "corrective_action.sh" script. This script has implementation of the actions.

Some of the actions are:

rebootNeeded - Reboots the device
resetNeeded - Restarts the required process
storeInformation - Stores Memory and CPU usage

In Raspberry Pi the functionality of self-heal feature is provided by systemd.

Code Flow

Resource Monitoring - resource_monitor.sh

• resource_monitor.sh monitors the Memory and CPU usage
• Average memory and CPU thresholds will be obtained from syscfg.db (default avg_cpu_threshold:100, avg_memory_threshold:100)

• Memory Usage Monitor
  • Gets the total, free and used memory details using free command
  • AvgMemUsed = usedMem*100 / totalMem
  • if AvgMemUsed > memory_threshold, device will be rebooted
• CPU Usage Monitor
  • Active CPU is considered as sum of user, system, iowait, irq, softirq, steal cpu
  • CPU usage difference in every 30 seconds for a period of 5 mins, is considered as an Curr_CPULoad_Avg.
  • If Curr_CPULoad_Avg > cpu_threshold, corrective action will be taken

Process Monitoring - task_health_monitor.sh

• task_health_monitor.sh monitors the status of various taks periodically and takes the corrective action
• Default monitoring interval is 15mins and can be modified using resource_monitor_interval in syscfg.db
• Monitors
  • Health of peer processor, in case of dual core processors
  • Other tasks added as part of the script
• New tasks can be added by editing the script

Connectivity Test - self_heal_connectivity_test.sh

• Self_heal_connectivity_test.sh will run Ping and DNS tests.
• ConnTest_PingInterval in syscfg.db specifies the frequency of the connectivity test.
• If nothing specified, it is 60seconds by default.

• runPingTest
  • Gets the IP (default_router IP) from syscfg.db
  • If no IP specified, it will try pinging to default gw
  • If ping fails, takes the corrective action, which is none by default.
• runDNSPingTest
  • This is disabled by default. Can be enabled by selfheal_dns_pingtest_enable in syscfg.db
  • Gets the urlToVerify from syscfg.db
  • If nslookup fails, takes the corrective action, which is none by default

Objects

Self heal objects in its DML layer: 

Self heal can be Enabled/disabled by the below data model. By default, it is enabled

$ dmcli eRT getv Device.SelfHeal.X_RDKCENTRAL-COM_Enable
CR component name is: eRT.com.cisco.spvtg.ccsp.CR
subsystem_prefix eRT.
getv from/to component(eRT.com.cisco.spvtg.ccsp.tdm): Device.SelfHeal.X_RDKCENTRAL-COM_Enable
Execution succeed.
Parameter    1 name: Device.SelfHeal.X_RDKCENTRAL-COM_Enable
               type:       bool,    value: true

Verify the selfheal feature running status

$ ps -Af | grep -i self
 4449 root       0:00 {self_heal_conne} /bin/sh /usr/ccsp/tad/self_heal_connectivity_test.sh
18921 root       0:00 grep -i self

Resource monitoring

The Below DM is used to verify the Average CPU threshold. By default the value is set to 100

$ dmcli eRT getv Device.SelfHeal.ResourceMonitor.X_RDKCENTRAL-COM_AvgCPUThreshold
CR component name is: eRT.com.cisco.spvtg.ccsp.CR
subsystem_prefix eRT.
getv from/to component(eRT.com.cisco.spvtg.ccsp.tdm): Device.SelfHeal.ResourceMonitor.X_RDKCENTRAL-COM_AvgCPUThreshold
Execution succeed.
Parameter    1 name: Device.SelfHeal.ResourceMonitor.X_RDKCENTRAL-COM_AvgCPUThreshold
               type:       uint,    value: 100

The Below DM is used to verify the Average Memory threshold. By default the value is set to 100

$ dmcli eRT getv Device.SelfHeal.ResourceMonitor.X_RDKCENTRAL-COM_AvgMemoryThreshold
CR component name is: eRT.com.cisco.spvtg.ccsp.CR
subsystem_prefix eRT.
getv from/to component(eRT.com.cisco.spvtg.ccsp.tdm): Device.SelfHeal.ResourceMonitor.X_RDKCENTRAL-COM_AvgMemoryThreshold
Execution succeed.
Parameter    1 name: Device.SelfHeal.ResourceMonitor.X_RDKCENTRAL-COM_AvgMemoryThreshold
               type:       uint,    value: 100