hw2-reports

Homework 2: Create Summary Reports

The second homework (or hands-on assignment) is to create summary reports. The idea is to collect information from the network devices and turn that into an easily readable report that is useful on its own.

Creating a Network Topology Diagram

This course does not require to solve a set of specific problems, it rather sets a theme to follow and provides the student with ideas. One of the given ideas is to create a network graph based on neighbor information collected from the devices. I like this idea, because it provides me with an opportunity to look into Graphviz which is a suggested tool to visualize (i.e. generate an image file of) the connectivity graph.

This problem has some relation to the Prescriptive Topology Manager from Cumulus Networks (PTM Documentation) which uses a file in Graphviz format named topology.dot to describe the intended network topolgy and uses the Link Layer Discovery Protocol (LLDP) to compare the switch's neighbors with the given topology.

Since I am using virtual Cisco IOS devices in my lab, I am using the Cisco Discovery Protocol (CDP, also known as CiscoDP on Enterasys (now ExtremeEOS) switches, sometimes known as Industry Standard Discovery Protocol (ISDP) as well) as data source.

Approaching a Solution

I intend to use the ios_command module to collect CDP data from the virtual routers. The output is then parsed by an Ansible filter (one of the parse_cli or parse_cli_textfsm filters) into variables. The variables are used together with a jinja2 template to generate a network topology description in Graphviz format, i.e. using the DOT language. The dot program is used to create an image file in PNG format.

---

Interlude 1: Beating Ansible Installation into Shape

When I tried to use the ios_command module, it would fail with the error message: unable to open shell. Please see: https://docs.ansible.com/ansible/network_debug_troubleshooting.html#unable-to-open-shell. Root cause was that pip install ansible did not install all requirements for the ios_command module, it missed the two Python packages enum34 and ipaddress. For all the gory details see the separate Interlude 1 document.

---

Ansible Playbook

The solution approach provides a general structure for the Ansible playbook topology.yml. I have used tags to allow invoking parts of the playbook that belong together:

• get retrieves CDP neighbor information from network devices and writes the raw CLI output to a file
• parse reads the raw CLI output from file and parses it into a data structure
• graph uses the data structure from the parse step to generate a DOT language description of the graph (connectivity.gv) and a PNG image visualizing it (connectivity.png)
• debug_* tags allow debugging of the respective steps

---

Interlude 2: Lab Configuration

The lab as configured for homework 1 lacks interesting connectivity, thus I needed add some configuration. This is described in the separate Interlude 2 document.

---

Playbook Description

The following sections provide more information about the Ansible playbook to create a visualization of the network connectivity.

One Play

The playbook comprises one play with several tasks. The network devices used are specified at play level (the OOB group using the OOB interfaces, since in-band connectivity is not yet established). Additionally, use of the local connection is specified at play level, because all tasks are executed on localhost:

• Files for intermediate steps and end result need to be created on the Ansible host.
• Ansible network modules are executed on the Ansible host, the module then connects to the network device.

A few variables to control where generated files are stored and how they are named are defined at play level as well.

Tasks to Retrieve CDP Information

Four tasks (with the tag get) are used to retrieve CDP neighbor information from the network devices. The first uses the ios_command module to issue show cdp neighbors detail on each router. This information is written to a text file in the fouth get task, after the second and third get tasks have deleted old CDP information and (re-)created the directory to store it in.

Tasks to Parse CDP Information

Two tasks (with tag parse) are used to parse the CDP data to fill a data structure that can be used in a Jinja2 template. The first reads the raw CDP output using the Ansible lookup plugin file. The second uses the parse_cli_textfsm filter with the TextFSM template cdp_neighbors_detail.textfsm to create a list with neighbor data.

Tasks to Graph the Topology

Four tasks (with tag graph) are used to generate an image that shows the network topology. The first two tasks prepare the output directory, i.e. they delete contents from previous playbook executions and (re-)create the output directory. The third task uses the Ansible template module to generate a DOT language description with the connectivity.j2 Jinja2 template. The fourth task invokes the dot program using the Ansible command module to generate a PNG image of the connectivity graph.

---

Network Topology Visualization

The Ansible playbook topology.yml generates to result files:

1. The DOT language graph description connectivity.gv
2. The PNG image connectivity.png of the topology

Topology Image

---

References

• Ansible Filters
  • parse_cli
  • parse_cli_textfsm
• Ansible Lookup Plugins
• Ansible Modules
  • ios_command
  • ios_config
• Ansible Templating
• DOT language
• DOT language (Wikipedia)
• dot program
• Graphviz
• Jinja2
• PNG
• Prescriptive Topology Manager
• PTM Documentation
• TextFSM
• TextFSM documentation
• TextFSM examples
• YAML

---

BNAS2018 GitHub repository | My GitHub user page | My home page