Multithreaded parallel network command execution, simplified, for network engineers of the world!
For everything:
pip install netmiko docopt getpass
Clone the repo, and run the scripts using Python 3. This is NOT a library or module, it is meant to be templates and examples that you can use, copy, and make better.
Create your 'customer' file and 'credentials' file based on exampleCSV.md.
This script is the master controller script. It will make the necessary commands to run whatever is needed against whatever is needed.
Usage:
netspark.py -h | --help
netspark.py (--info COMMAND | --config CONFIGFILE) (--csv FILENAME | --db QUERYNAME | --ip IPADDR) [-c CREDFILE] [--debug]
Options:
-h, --help Shows this menu.
-d, --debug Print debug information. This is the most verbose
option.
--info COMMAND This will run regular show commands
--config CONFIGFILE This will run configuration commands
--csv FILENAME Input file if using CSV [default: test.csv]
--db QUERYNAME SQL field: 'groupname' [default: test]
--ip IPADDR Single IP address of test switch
-c CREDFILE Crednetials file [default: credentials.csv]
To run write mem on every host in your csv file switches.csv (assuming you only have one credentials.csv):
python netspark.py --info "wr mem" --csv switches.csv
To make a config change with many lines, such as changing a local account, you'd make a text file with the lines to change it which will then be run against the hosts. For example, lets make a changeuser.txt:
exit
conf t revert time 5
username secretbackdoor privilege 15 password Thiscantbereal!
--config runs in the context of config mode, which is why I use an exit above.
Then you'd run it like so: python netspark.py --config changeuser.txt --csv switches.csv
This command supports all of the netmiko classes of devices, but I designed it to be used with Cisco gear since that's what most of us run. You just need to specify a different device object in your switches.csv file to work with new stuff.
Final example, lets say you have a rogue sysadmin at one branch who refuses to use newer credentials/radius/whatever. So you have a different nonstandard set of creds and routers.
python netspark.py --info "configure confirm" --csv crappyrouters.csv -c crappycreds.csv
This is multithreaded and I haven't added error handling. I force all of my changes to use config revision (the 'conf t revert time 5' above) as a workaround until I have good error handling in place. It's laziness and a lack of time, it would be decently easy to implement.
The multiprocess code I wrote is messy because of context issues. I plan on making that part into a library later on, but something something time and laziness.
The default number of simultaneous executions is 8, I have run this successfully at 50x simultaneous on a network of 400+ devices but keep in mind that they will return output at a rate of 50x, and it spits it out to STDOUT, so you might want to just leave it.
I will add support for --db and --ip later on, probably --db first because I want this to dive into the rest of my open-source stack I'm slowly writing about on https://teamignition.us so that this all ties in beautifully.
This is old code and alternative projects. It is useful reference material so I'm leaving it up, but it will not be updated.
All patches welcome! Please read CONTRIBUTING.md for furthers details...whenever I make it. For now, submit PR's and we'll chat about them.
GNU GPL Version 3 - see the LICENSE file for details