Skip to end of banner
Go to start of banner

How to select Nodes (and Interfaces) for reporting

Skip to end of metadata
Go to start of metadata

You are viewing an old version of this page. View the current version.

Compare with Current View Page History

« Previous Version 7 Next »

In opReports 3.0 the various methods for selecting what to report on have been consolidated and simplified.

This page describes what mechanisms are available, and how to control them in the GUI, with opreports-cli and for report schedules.

Which selection mechanisms are supported by what report types?

Certain reports do have specific requirements which are shown in the table below.

In general, however, providing more precision than necessary is allowed: For example, you can use a node and interfaces (and even types) list file for all reports; for reports where interfaces are not relevant, the extra information will simply be discarded and just the listed nodes will be used. Reports that don't need the summary type will simply use the nodes (and possibly the interfaces).

On the other hand, providing insufficiently precise information to reports that require it will result in an error message (e.g. trying to run a summary report with just group=X).

 

Report TypeSelection MechanismsNotes
Nodesingle node onlyThe node report supports a single node only.
If your selection contains more nodes, then the report is created for the first listed node.
(Node) Healthall, interfaces irrelevantInterface selections are not relevant for this report.
WANall, interfaces ignoredAs of version 3.0.4 the WAN report ignores interface selections;
It operates on any of the selected nodes whose net type is "wan".
QoSall, interfaces ignoredAs of version 3.0.4 the QoS report ignores interface selections;
It operates on the selected nodes' active interfaces that have QoS configured.
Uptimeall, interfaces irrelevantInterface selections are not relevant for this report.
Response Timeall, interfaces irrelevantInterface selections are not relevant for this report.
Interface UtilisationallInterface selections are honored but the "type" component (in selections by node+interface+type) are ignored as not relevant.
CPUall, interfaces irrelevantInterface selections are not relevant for this report.
Free Memoryall, interfaces irrelevantInterface selections are not relevant for this report.
Memory Poolall, interfaces irrelevantInterface selections are not relevant for this report.
Memory Bufferall, interfaces irrelevantInterface selections are not relevant for this report.
Traffic UsageallInterface selections are honored but the "type" component (in selections by node+interface+type) are ignored as not relevant.
Traffic Summarynode+interface+type

This report strictly requires this specific selection format.

Traffic SnapshotopCharts Business Services,
selected per page
This report type strictly requires lists of Business Services for each of its (multiple) page definitions.

 

Choosing the selection mechanism for scheduled reports

For scheduled reports your schedule must contain a property named sources, with one of the following values: "everything", "node_regexp", "node_group", "nodes", "node_list",  "node_intf_list", "node_intf_type_list", or "business_services" (in 3.0.14 an newer). Those mechanisms are described below.

If you use the opReports schedule editing GUI then this property will be managed on your behalf.

The Simplest Choice: Everything

If you do not make an explicit selection, then opReports will work on all active nodes (and all their active interfaces, for report types that handle interfaces).

In the GUI this  choice is shown as "All Active Nodes".

Nodes that belong to a specific group

In NMIS every node belongs to precisely one group, and this concept therefore applies to opReports as well.

With opreport-cli you have to give the argument group=<groupname>. In a report schedule this is expressed using the property node_group. In the GUI this choice is presented as "by Group".

There are two "wildcard" groups available:

  • Group "All" is equivalent to the default choice, all active nodes. This wildcard should not be used as we will likely retire it in a future version of opReports.
  • Group "Each" is available for scheduled reports only, and causes the generation of a separate report for each of the known groups.

Nodes whose name matches a regular expression

In the GUI this choice is called "by Regular Expression", opreports-cli uses the command line argument noderegex=<regular expression>, and for scheduled reports you'd specify this with the property  node_regexp.

The regular expression syntax is Perl's standard, described in detail in this Perl Regexp Tutorial.

Explicitly listed Nodes

In the GUI this choice is called "Pick from Node List".

To use this mechanism with opreports-cli, you have to list each node you want in a separate nodes=<nodename> argument.

In a report schedule the property nodes would have to be set to an array of node names, like in this example:

"nodes" : [
  "ASGARD",
  "midgard"
],

Nodes listed in a file

opReports expects a node list file to contain one node name per line.  Whitespace before or after the node name is removed.

In the GUI this choice is called "from Node List File (Upload)", and you need to select a suitable file for uploading.

For opreports-cli, the command line argument node_list=<path to node list file> would be used. In that case the node list file must already reside on the opReports server.

In a report schedule definition, you'd use the property node_list, with the path to the list file as value:

"node_list" : "/tmp/my_list_of_lotsa_nodes.txt"

Nodes and specific Interfaces, listed in a file

Certain reports allow a more precise selection of nodes and just some of their interfaces. This is implemented using a list file.

In the GUI you'd select "from Node and Interfaces List File (Upload)" and upload the file of choice. For opreports-cli you use node_intf_list=<path to listfile>, and for a scheduled report you would set the property node_intf_list with the value being the path to the list file.

The node and interface list can be in one of  two formats, JSON or plain text:

  • JSON: it must be a valid JSON document, consisting of a hash of the node name as key, and the value being a list of the interfaces in question.

     

    { "testnode": [ "eth0" ], "othernode" : [ 1, 2, "Tunnel20" ] }
  • Plain Text: a text file, one entry per line.
    Each entry must start with the node name, followed by one or more TAB characters, and one or more interfaces (again separated by TAB characters). If you list a node on multiple lines then all  listed  interfaces will be combined into a single list. Blank lines and lines starting with the "#" sign are treated as comments and are ignored.

    testnode   2   14    eth0
    othernode  Dialer1
    testnode   17

For both JSON and Plain Text formats, interfaces can be identified by the numeric SNMP interface index, or by the SNMP ifDescr property.

Nodes, specific Interfaces and Types, listed in a file

Certain reports offer a refinement of the above, with the added notion of a "Type" for grouping of Node+Interface into particular reporting classes.

This is implemented again using a list file, but  with a very specific format - which has a few inherent limitations; Starting with version 3.0.16, opReports also supports CSV as format for this  type of input.

The relevant GUI choice is called "from Node, Interfaces and Type List File (Upload)",  for opreports-cli the parameter is node_intf_type_list=<path to listfile>, and in a report schedule the controlling property is node_intf_type_list (value again the path to the list file).

Plain Text Format

Please note: as of version 3.0.16 it's recommended that you use CSV as a safer alternative to this format.

The list file format is plain text, and each line must consist of precisely one node name, one of its interfaces and a "type" declaration. The interface must be identified by its SNMP Interface Index.

Node name and interface must be separated by "_", and this must be separated by the "type" by one or more spaces, like in the example below:

nodeA_1 groupA
nodeB_41 groupB
nodeC_41_eth0 groupB

The example also shows an optional format variant: the node+interface stanza may include a trailing "_<ifdescr>", but  the interface description is not used for the selection logic: only the SNMP Interface Index is relevant. It is recommended that you do not use this flavour as it's ambiguous: interface descriptions can (and often do) include both _ and space characters.

The "type" will be used to group all nodes and interfaces with the same type value into a group for summary reporting.

CSV Format (3.0.16 and newer)

opReports now also supports CSV (with comma as the separator character) for this kind of input.
The lines in your file must contain at least the following four columns, in the following order:

  1. The  node name,
  2. the SNMP Interface Index (must be present but may be empty if the Interface Description is given)
  3. the Interface Description (must be present but is ignored if an SNMP Interface Index is given),
  4. the "type" declaration.

Extra columns are ignored; files with fewer columns are rejected. Empty lines and comment lines (ie. lines starting with a "#" character) are ignored.

For extra convenience you may now also specify interfaces by their Interface Description instead of the Interface Index; both columns must be present in your input, but one of them may be blank.
The SNMP Interface Index is considered authoritative, if given. If it is not, then opReports looks for an interface with the given Interface Description. Nonexistent interfaces are skipped, and a warning message is logged.

Here is an example CSV file:

# comment, ignored. columns: nodename,interface index,interface description,type name
"some node","1","FastEthernet0/0","categoryA"
"not_the_greatest_name","10","Dialer1","catB"
"pleasefindme",,"Dialer1194","categoryA"
"iknowtheindex",12,,"catB"

Nodes and Interfaces that are part of an opCharts Business Service

If you have opReports version 3.0.14 and newer and opCharts is installed on the same system, then you can make use of Business Services to declare nodes and interfaces for reporting.

Configuration

The following three configuration options (in conf/opCommon.nmis) are vital for opReports accessing opCharts:

# base of opCharts server url, eg http://localhost or http://localhost:80 - no slash at the end
'opreports_opcharts_url_base' => "http://127.0.0.1:8042",
'opreports_opcharts_user' => "nmis", # opreports needs a user with readonly-access
'opreports_opcharts_password' => "nm1888",

If you've changed the password for the default nmis user (or disabled it altogether), then these configuration items need to be adjusted accordingly.
Once that's done you need to restart the OMK webserver (using sudo service omkd restart) to activate the changed configuration.

Usage

In the GUI you will be presented with a list of known Business Service names, which supports multiple selections.

When using opreports-cli, the parameter is called business_services and you need to pass in each business service name separately, e.g. opreports-cli.pl type=health business_services=first_service business_services=second_service.

In a report schedule file, the property is called business_services and its value must be a list of business service names.

Business service memberships are expanded at report creation time.

  • No labels