PingPollerd
The Oracle Communications Unified Assurance Metric Ping Poller is the common ICMP availability and latency collection poller. This poller is device oriented and uses the default Latency, Packet Loss, Ping Jitter and/or Ping Jitter Utilization metric types.
You can run this application as a service using the Services UI.
Ping Poller Setup
-
Devices must exist in the Device Catalog for this application to be able to function. Devices can be created in several different ways:
-
Using the Device Auto Discovery application.
-
Using the Manual Discovery UI:
-
Manually entering all values using the Devices UI:
-
-
Metrics must exist for the devices to be polled. Metrics can be created in several different ways:
-
The default rules for the Metric Custom Poller Discovery Agent application will create the metrics for the Device instance.
-
Using a polling policy to create metrics using whatever restrictions are needed. See the Polling Policies UI and Metric Poller Discovery Agent application documentation:
-
Using the Polling Assignments UI to manually select which devices and instances are processed. See the related documentation:
-
-
Enable the default Service, unless a specific configuration option is needed:
Default Service
The following table shows the settings for the default service. Actual values are in bold, descriptions of values are in plaintext.
| Field | Value |
|---|---|
| Package | coreCollection-app |
| Name | Metric Ping Latency Poller |
| Program | bin/core/collection/PingPollerd |
| Arguments | This field is blank. There is no default value. |
| Description | Metric Poller used to poll Ping (ICMP) Availability & Response Times |
| Failover Type | Standalone (Supported: Standalone, Primary, Redundant/Backup, Cluster) |
| Status | Disabled |
| Privileged | This option is selected. |
See Services in Unified Assurance User's Guide for general information about the settings for services.
See Using Application Primary/Backup Failover and Using Application Clustering for more information about the different failover types.
Default Configuration
The following table shows the default configurations for the application. Actual values are in bold, descriptions of values are in plaintext.
| Name | Default Value | Possible Values | Notes |
|---|---|---|---|
| CheckTime | 900 | An integer | How often (in seconds) where the application checks to see if there are any relevant changes. |
| DeviceZoneID | Default First Zone | All Zones or the name of any available device zone. | The poller will only retrieve devices in the specific zone. |
| LogFile | logs/MetricPingPoller.log | Text, 255 characters | The relative path to the log file. |
| LogLevel | ERROR | OFF, FATAL, ERROR, WARN, INFO, DEBUG | The logging level for the application. |
| PacketSize | 64 | An integer | The size of the packet the poller will send, in bytes. |
| PollTime | 300 | An integer | How often the poller will start a poll cycle, in seconds. |
| Threads | 3 | An integer | Number of process threads created. The aggregator takes a third of this value (rounded up) for database threads unless overridden by the DBThreads application configuration. |
| TimeOut | 1 | An integer | The time out for the ping request. |
| DBThreads | This field is blank. There is no default value. | An integer | (Optional) Number of database threads to be created. If not specified, defaults to a third (rounded up) of Threads application configuration. |
| DeviceGroupID | This field is blank. There is no default value. | The name of any available device group. | (Optional) If specified, the poller will only retrieve devices in the specific group. |
| MaxBatchSize | This field is blank. There is no default value. | An integer | (Optional) The maximum size of each batch handled per thread. By default, batches are sized by dividing the total number of devices to poll over the number of threads. As the number of devices increases, it may be desirable to limit this batch size to prevent certain denial of service limitations on your network. |
| PingCount | 10 | An integer | (Optional) Number of times to ping the device. This is also used for multiple serial polling and packet loss calculation. NOTE We may ping up to an additional 2 times for more accuracy. |
| PollByIP | Enabled | Enabled or Disabled | (Optional) If enabled, the device IP address will be used to ping the device instead of the device DNS name. |
| PreferIPv4 | Enabled | Enabled or Disabled | (Optional) Controls whether or not to prefer IPv4 transport to communicate with Devices. This option is only considered if both IPv4 and IPv6 are available for a device. If this config is missing, IPv6 will be preferred. |
| ThresholdThreads | This field is blank. There is no default value. | An integer | (Optional) Number of threshold threads to be created. Enabled the checking of thresholds in the application instead of the Default Threshold Engine. If not specified, application threshold checking is disabled. |
| SendAllViolations | Disabled | Enabled or Disabled | (Optional) If enabled, thresholds will send violation notifications every poll cycle rather than only on state change. Requires ThresholdThreads |
Best Practices
The following list shows the best practices for working with this application:
-
Ensure the following metric types have been configured for polling:
-
60 - Latency
-
61 - Packet Loss
-
62 - Ping Jitter
-
63 - Ping Jitter Utilization
-
-
When batching multiple devices per thread, ensure that there is potential based on hardware for processing overhead to be included in the Latency metric because pings are requested asynchronously per batch. This overhead can be reduced by either limiting the batch size with MaxBatchSize to reduce the number of devices pinged at once by the threads, but this will cause a longer poll cycle.
-
In cases where devices are reporting high packet loss but DEBUG logs show ping results as failed with code MALFORMED, this can indicate that too many threads are configured or too many devices are being pinged simultaneously. To troubleshoot this issue, you should try the following:
-
Reset Threads to the recommended value of 3 * (number of CPU Cores) if set greater.
-
Set MaxBatchSize to 64 / (number of CPU Cores) if device count to be polled is greater than 200. Note that this can cause the poll cycle to take longer.
-
Administration Details
The following list shows the technical details you will need for advanced administration of the application:
-
Package: coreCollection-app
-
Package:
./PingPollerd [OPTIONS] -
Options:
-c, --AppConfigID N Application Config ID (Service, Job, or Request ID) -?, -h, --Help Print usage and exit -
Threaded: Multithreaded