DX NetOps CA Performance Management (CAPM) Polled Items counts obtained by different means return different results

Products

• CA Infrastructure Management
• CA Performance Management - Usage and Administration
• DX NetOps

Issue/Introduction

Polled Items counts obtained by different means return different results. The total Polled Item counts for a device will differ depending on which method was used to obtain the data.

There are various methods to obtain Polled Item counts for a device and each one is designed to display different data. These methods include:

• The Polled Items Count value when viewing a device in the Monitored Devices view in Performance Center
• Manually summing the polled components by Metric Family through the Monitored Devices view in Performance Center
• Query of the device using the oData Query Builder API
• Querying of the Vertica database for counts of polled items per facet
• The total Polled item count for a device obtained from the Data Aggregator's Polling Configuration by IP query

How can these be checked?

Environment

DX NetOps CAPM 

Resolution

The following lists each method of obtaining Polled Item counts for devices and what items they include or exclude.

1. Under Administration->Data Aggregator->Monitored Devices, after selecting a device and looking at the Details tab, the 'Polled Items Count' will include all components which are listed for each Supported Polled Metric Family. It will also include the following Metric Families which are for the device item itself:

Availability [system Statistics]
Reachability [ICMP]
Device Polling Statistics [SNMP Device Polling Statistics]

• When you view a monitored device and see the list of Polled Metric Families you will notice that by default we display 50 metric families at a time. You will need to go to any additional pages and metric families in order to see/count all items.

2. Under Administration->Data Aggregator->Monitored Devices, after selecting a device and manually counting up all the "Supported" Polled Metric Families, you will arrive at a larger number than what the Details tab shows for polled items. This is because there are several Supported metric families that even though they do not list any components/items, are for the device item itself.

For example, a router or switch will show the following metric families as being 'Supported' but will not list any items. This is because the metric families are for the device itself:

- Availability [system Statistics]
- Device Polling Statistics [SNMP Device Polling Statistics]
- Network Address [IP Address Table (RFC1213)]
- Reachability [ICMP]
- System

• These are just some examples. The number of device-related metric families and which metric families can differ for each device.
• When counting up all the Supported polled metric families, do NOT count all the device-specific metric families EXCEPT for Availability, Reachability and Device Polling Statistics. That number should then match the Details tab.

3. When you go to the Data Aggregator system's dcdebug page (http://<DA_HOST>:8581/dcdebug/searchdebug.html), and specifically to the Polling Configuration by IP for a device, the "total polled item count" at the top will match what the Monitored Devices->Details tab shows. This will also match manually counting up the Metric Families as described above in line item 4.

4. When using the oData query, the returned 'PolledItemCount' DOES NOT include all the device-level Metric Families (Availability, Reachability, Network Address, System or Device Polling Statistics). This only counts the Metric Families with components associated to them (interfaces, CPUs, memory pools, QoS, etc.).

5. VSQL queries to obtain polled item counts.

• There are a couple options - each will present data in a different manner.

A. The vsql query below can be run on the Data Repository to provide a list of polled item counts "per facet". Log into the vertica database as the database admin user. For example:  /opt/vertica/bin/vsql -U dradmin -w <DB_PASSWORD>

select facet_qname,count(*) from dauser.v_item_facet where item_id in (select item_id from dauser.item where tenant_item_id=1) group by facet_qname;

B1. The following will provide a polled item count "per device". This is also run in a vertica vsql session, but this one is run as the DA user account (typically dauser):

select device_item_id,device_name,count(*) from v_poll_item group by device_item_id, device_name order by count(*) desc;

• If for some reason you do not know who the DA user account is, you can run the following query as the dradmin:

select * from users;

B2. You can also run query #2 above as the dradmin user. The syntax is changed to append the name of the DA user account to the table names:

select device_item_id,device_name,count(*) from dauser.v_poll_item group by device_item_id, device_name order by count(*) desc;

Additional Information

• Any components which are "Filtered" are NOT counted toward any of the polled item counts above. If an item is filtered, then it is not being polled and no data is being collected.