Discovering Inventory using SNMP
The cloudshell-snmp package comes bundled with common SNMP MIBs which are pretty generic across most devices. You can find the full list of MIBs that are provided as a part of the cloudshell-snmp package in the package repository.
Here’s a short explanation of the MIBs typically used in discovery:
IF-MIB: In this MIB we are using the table called “ifTable”, which contains information about the ports like interface full address (e.g. GigabitEthrnet0/0/1), the type of the port, speed etc.
Entity-MIB: Contains information about Chassis, modules, ports and describes the structure of the entities. It also contains the physical index of each entity, using the index we can then map between port to its associated interface.
IP-MIB: Contains the IP (IPV6 or IPV4) address of each port.
SNMPv2-MIB: Contains information about the serial number of the device, system description etc.
Etherlike-Mib: Used by CloudShell Shells typically to get the duplex value of interfaces.
In the above example, we initialize the QualiSnmp module which requires at least the device IP and the community string. We then use the get_property function to get the value of the sysName property using one of the MIBs which are included with cloudshell-snmp.
Here we used the same class to retrieve the entire IF table. The get_table function returns a Quali object which inherits from Python’s ordered_dictionary with additional handy sorting and filtering functions to get the table’s rows and columns.
Since the SNMP parameters are included as attributes in the Resource Standard they can be easily retrieved by the driver when implementing the get_inventory function:
Supporting custom vendor MIBs
Compiling and downloading the vendor MIBs
Some vendors will use their own custom MIBs which you’ll want to include in the discovery. For example, this CISCO site provides access to some common MIBS you can download: http://tools.cisco.com/ITDIT/MIBS/MainServlet?ReleaseSel=3&PlatformSel=236&fsSel=368. Other vendors have their own MIBs download links.
To compile the MIBs we recommend using the mibdump.py script which is part of the PySMI SNMP libraries and tools. The mibdump.py script supports both compiling MIB sources you’ve downloaded locally as well as using remote MIB sources. For example, the snmplabs repository provides a wealth of publicly available MIBs. Notice, if you are trying to compile MIB sources you’ve downloaded locally, you should be sure to download all of the dependencies to the same folder (e.g. IF-MIB depends on SNMPV2-MIB). However, you can specify more than one MIB source, so you can include both your local MIB sources and remote repositories.
The following example will compile the IF-MIB source from the snmplabs repository:
If you’re a Linux user, you can take advantage of the smidump package. This package is available as an RPM/APT package for various Linux distros. The smidump package is somewhat easier to use with a clear syntax. For example, to use the smidump package to compile the IF-MIB to Python, simply run these two lines from the terminal:
Including the compiled MIBs in your Shell project
After downloading the MIB files using either the Linux or Windows option, you’ll need to include them in your shell driver project, save the MIBs Python files in a sub-package folder and add the following code to your Shell:
After loading the MIB, you’ll be able to use it like any of the MIBs included in cloudshell-snmp and retrieve properties and tables.
Converting the MIB data to get_inventory results
In order to successfully load the internal structure and attributes of a resource, the get_inventory function need to return lists of AutoLoadResource and AutoLoadAttribute objects.
The following driver implements basic discovery using generic MIBs only and converts the result to the objects required by get_inventory. To view the full source click the ‘view’ or ‘raw’ links above the code snippet. The code can be used as a reference for creating similar procedures:
To use this module, simply initialize an instance of GenericSNMPDiscovery and call the discover method: