Riffdog Terraform scanner - finding 'things' in the Real World (aka AWS) which Terraform didn't put there.

Overview

riffdog

Riffdog Terraform / Reality scanner - finding 'things' in the Real World which Terraform didn't put there.

This project works by firstly loading your terraform state files - not your terraform files, and building a memory object that represents what terraform thinks it has deployed. The second step is then to access your environment and actually look what is there, and it builds another memory object.

Then it compares the two, and looks for:

  • Things that TF put there, but arn't there
  • Things that are there, that TF doesn't know about

For this to work, you must install a resouce pack, for instance riffdog_aws

Command line vs Library

Riffdog is both a command line tool and a python library. The command line tool basically creates a config object and calls the scan method.

To install (for example with the AWS resources)

$ pip install riffdog[aws]

To run:

$ riffdog -b bucketname_containing_states

For full command line reference, please see the docs #FIXME - link to docs!

Light vs Full Scan mode

AWS and Terraform are very complex: for instance, a EC2 instance is a computer, a collection of network cards, volumes, etc. In 'light' mode, we do a high level scan to see if all the main components are there - in 'full' mode, it inspects all elements (as far as RiffDog is aware of, see below) - i.e. checks that all network cards, volumes, etc are mounted as intended.

Ignores

Its important to realise that not all infrastructure is deployed via TF - for instance auto scaling, or service discovery systems may be interacting with your infrastructure, and this is OK, but you don't want RiffDog raising alerts on those.

Caveats:

This is pretty much always a work under development, as AWS adds more features modules etc. The system is designed to be modular, but we're only able to keep up with modules that we use - so if you use a feature of AWS that we don't scan please feel free to add a module. If you need help, please contact us or file a ticket.

Testing

Automated testing of this system is quite difficult. Ultimatly these are here to make sure that changes to internal data structures do not break other functions (e.g. some data objects such as network interfaces are used by multiple features: both EC2 instances and Lambdas). The real test is whether it works against the current Terraform and AWS API's, not a snapshot from previous years.

  • Tests go into the tests folder,
  • Tox is used to make sure compatibility with various versions of python & libraries
Comments
  • Run tests on GitHub actions

    Run tests on GitHub actions

    Since I added a couple of tests, would be nice to actually run them presumably on GH actions if we can.

    Also, I used pytest because it's much nicer than nose (is nose even still used nowadays)?

    opened by dreynolds 2
  • Change the to_scan module process

    Change the to_scan module process

    This is an issue where new modules need to be added - but does it make sense to actually require this setting, i.e. shouldn't it be based on what modules there are in the discovery mode? And then allow (in future) config to --exclude particular modules (or white list), but by default use all available modules?

    enhancement question 
    opened by jmons 2
  • Dynamic Module Loading & Registration

    Dynamic Module Loading & Registration

    This comes with some notes, and items which might want to be changed:

    1. New 'resource' (was called modules) go into the resources folder. EAch file should contain One (or more) classes
    2. Should implement the Resource class - I've steered clear of "aws" terminology here so that in future resources can be for anything - I've used 'terraform' for the terraform and 'real' for things that actually exist (or don't)... (debate naming strategy)
    3. must decorate the class with the register decorator so that it links the state file to the class.

    Nothing else is needed - the class instance management is handled by the scanner, and it will find all classes in that folder.

    What I think is very much up for debate is this particular class interface - I don't think you can get away from the 'process state' (which is process an element in the state, not process the entire state file) being called multiple times, but perhaps the fetch_real_resources shouldn't take 'region' but a config, and in that config if its AWS it looks for aws_regions or something?

    from ..resource import Resource, register
    
    @register("aws-name-in-state-file")
    class MyResource(Resource):
        _local_real_things = {}
        _local_terraform_things = {} 
    
        def fetch_real_resources(self, region):
            # This may be called multiple times for each region in the scan list
            # i.e. append
            raise NotImplementedError()
    
        def process_state_resource(state_resource):
            # This function is called potentilly multiple times as each resource
            # is discovered by the state scanner i.e. append results to local store
            raise NotImplementedError()
    
        def compare(self, depth):
            # this function should be called once, take the local data and return
            # an array of result elements.
            raise NotImplementedError()
    

    More code cleanup is probably possible, I think I got all the debug entries into debug (don't try running with -vvv unless you're very brave (boto debug output becomes manic - fixing that requires a full logging config i.e. waiting for the configure file ticket).

    enhancement 
    opened by jmons 2
  • Fix `--exclude-resource` support and add `--include-resource` support

    Fix `--exclude-resource` support and add `--include-resource` support

    Fixes #45 If anything is in include-resource it completely overrides exclude-resource so it's like starting from a blank list.

    I went for --include-resource instead of --only so that the arguments are logically matched with --exclude-resource

    Happy to change if others disagree

    opened by dreynolds 1
  • Logging initialisation (First & Second Arg Parse)

    Logging initialisation (First & Second Arg Parse)

    Logging is not initialised until the Second argparse, which means logging in the initial import scan is impossible. This means exceptions in the import are completely silent and ignored.

    Some arguments need to be in both first and second - first to be parsed and second so the help works. Perhaps all core arguments?

    bug 
    opened by jmons 1
  • Framework to handle modules arguments

    Framework to handle modules arguments

    --region is really an aws argument - i.e. --aws_region but how does argparser (which hasn't had the module scanner run) know what the extra arguments are? Does it need to run twice, once before and one after loading?

    enhancement 
    opened by jmons 1
  • Adding basics for RDS Clusters (cluster, instances and param groups)

    Adding basics for RDS Clusters (cluster, instances and param groups)

    Implements https://github.com/jmons/riffdog/issues/17

    Added

    • RDS Cluster basics (cluster, instances, parameter groups.

    Fixed

    • Minor bug where the ResourceDirectory didnt have self in its str.
    • In order to support foldered module structure, the dynamic importing has changed a little.
    opened by Jamian 1
  • Documentation & Homepage revamp

    Documentation & Homepage revamp

    Giant homepage and docs revamp

    • making the home page more welcoming / project home page
    • added youtube video link (video needs review as well, see dm's)
    • added mailchimp (this may need a patch) for signup to news etc
    • added core team

    Then

    • fixed some of the help docs with some examples - because --help is quite good but brief, perhaps the command line reference morphs to being some tips and tricks type help? Don't want to really duplicate the wording of --help?
    • updated release notes to include the 0.1.0 release.
    opened by jmons 0
  • Initial command line tests using resource pack and module loader.

    Initial command line tests using resource pack and module loader.

    This is a relatively larger refactor

    1. It removes support for v3 state file (see #55 )
    2. It adds support for a file based state (and folder based state) - these don't have a named arguments but just go on the end of the command.
    3. it also adds support for the -b option to be a single folder or path e.g. -b bucketname/folder/ which speeds up dev/testing, especially with the --include-resource
    4. It fixes a bug in the alias - where before we had a bug that the --include-resource option did not apply to the scanning in state files, that fix introduced a new one where it excluded aliased. This fix now means that it will be included if the 'root' is in the list.

    In terms of testing, it now contains 2 state files, so when testing the core you can now do

    $ pip install -e tests/test_resource_pack
    $ riffdog -i test_resource_pack tests/test_state_files/
    

    The tox tests install the resource pack & then check that a scan finds the right items etc.

    opened by jmons 0
  • Bucket to Scan Hard Coded

    Bucket to Scan Hard Coded

    Looks like last merge we accidentally left in some code that was there to help with local devving. Flipping that back out (commenting it) and uncommenting the actual, release ready code that scans buckets dynamically.

    opened by Jamian 0
  • Revisit `--exclude` and create an `--only` option

    Revisit `--exclude` and create an `--only` option

    Using --exclude is more difficult now there are lots of resources - instead we should consider using an --only element to specify.

    I don't like the word --only though so feedback and ideas here appreciated.

    enhancement hot 
    opened by jmons 0
  • Proposal: Drop State file V3 support

    Proposal: Drop State file V3 support

    Whilst I added explicit support for V3 state files in 0.1.0, it seems that the structure of the individual states in the file are not compatible with V4- I had assumed they were because v4 states have schema_versions set to 0, which I assumed meant the state structure hadn't changed.

    But they have.

    In an upcoming PR, I have removed support for v3- explicitly with a warning.

    I think we need to set an overall output flag and change the table to indicate that there were states which look like tf states but we could not understand - but I think this ties into a statistics output so that a user knows what 'coverage' they are getting of their environment.

    documentation enhancement hot 
    opened by jmons 2
  • Dirty Reasons (not just a flag)

    Dirty Reasons (not just a flag)

    Implement a framework on the FoundItem (and then outputs) to store a Dirty Reason list, so that you know why a thing is dirty, not just given a flag.

    Consider output, and also perhaps guidance how to use it so outputs are concise.

    enhancement hot 
    opened by jmons 0
  • Investigate potential conflict of `depends_on` and `--exclude-resource`

    Investigate potential conflict of `depends_on` and `--exclude-resource`

    Consider the edge case:

    class Foo:
         ...
    
    class Bar:
        depends_on = [Foo]
        ...
    

    and then $ riffdog --exclude-resource Foo

    I think without checking that this means foo's real gets executed even though its states got ignored (?) (or it will error because Foo wasn't registered, not sure where the --exclude kicks in).

    question 
    opened by jmons 0
  • Config File Support

    Config File Support

    At the moment we're heading rapidly towards using just command line arguments, but I think we need to support a config file especially as we head towards more complex arguments such as AWS credentials etc.

    What do we want for this? YAML? (and then problems with yaml library imports - the pyyaml parser seems to have a lot of security patches which are mostly irrelevant in our case, but looks bad if we don't roll updates).

    enhancement 
    opened by jmons 1
  • Output to HTML

    Output to HTML

    It would be nice for this to have the option to generate a pretty output in HTML form. Useful if running this in an automated environment and you want to be able to link to a page/report. Something that is a bit more human friendly.

    One suggestion was to use Jinja and have it as optional?

    enhancement hot 
    opened by Jamian 2
Releases(0.1.0)
  • 0.1.0(Mar 2, 2020)

    Welcome to the beta release of Riffdog.

    To install this, please follow the guide - you will need at least one 'resource pack', and at the time of writing only the riffdog_aws resource pack exists.

    pip install riffdog[aws]

    Then to run:

    riffdog -b bucket_name_containing_states --show-matched

    Major changes:

    • The return data structure is radically different to before, which makes resource pack developers life easier.
    • Streamlined tabulated data outputs.
    • Introduced 'dirty' flag to indicate where resources exist, but do not match in some form of sub-data element.
    Source code(tar.gz)
    Source code(zip)
Owner
Riffdog
Community to support the riffdog app, riffdog plugins and associated repositories.
Riffdog
❄️ Don't waste your money paying for new tokens, once you have used your tokens, clean them up and resell them!

TokenCleaner Don't waste your money paying for new tokens, once you have used your tokens, clean them up and resell them! If you have a very large qua

0xVichy 59 Nov 14, 2022
Projeto de teste para acesso a API SWAPI.

SwapiTest Projeto de teste para acesso a API Swapi com informações sobre Star Wars. Como rodar o programa Foi utilizado o pipenv, então basta clonar o

Gabriel de Souza Alves 1 Nov 23, 2021
Wechat based auto reply with pyautogui

Python-微信 自动回复 练手~ 一直想做个给微信发个消息,就可以跑Python程序,并将结果发送给我的东西,之前看了 B站@不高兴就喝水 的视频,终于有了灵感~ 使用的是模拟点击方案,请求期间是不能操作了。 库 pyautogui 用于模拟鼠标键盘操作和定位操作位置 pyperclip 剪贴板

Vito Song 1 Oct 22, 2022
A Pluggable And Powerful Telegram Manager Bot

FRIDAY-USERBOT 🇮🇳 A Powerful, Smart And Simple Userbot In Pyrogram. Support 🚑 Inspiration & Credits Userge-X Userge Pokurt Pyrogram Code Owners Chs

DevsExpo 371 Mar 26, 2022
Compares and analyzes GCP IAM roles.

gcp-iam-analyzer I wrote this to help in my day to day working in GCP. A lot of the time I am doing role comparisons to see which role has more permis

Jason Dyke 37 Dec 28, 2022
Andrei 1.4k Dec 24, 2022
A discord webhook client written in Python.

DiscordWebhook A discord webhook client written in Python. Installation pip install webhook-client Example from webhook_client import WebhookClient, E

Elijah 4 Nov 28, 2022
Wordnik Python public library

Python 2.7 client for Wordnik.com API Overview This is a Python 2.7 client for the Wordnik.com v4 API. For more information, see http://developer.word

Wordnik 224 Dec 29, 2022
SimpleDCABot is a simple bot that buys crypto with a dollar-cost averaging strategy.

Simple Open Dollar Cost Averaging (DCA) Bot SimpleDCABot is a simple bot that buys crypto on a selected exchange at regular intervals for a prescribed

4 Mar 28, 2022
Crypto trading bot that detects surges in the bitcoin price and executes trades.

The bot will be trading Bitcoin automatically if the price has increased by more than 3% in the last 10 minutes. We will have a stop loss of 5% and t

164 Oct 20, 2022
The official Pushy SDK for Python apps.

pushy-python The official Pushy SDK for Python apps. Pushy is the most reliable push notification gateway, perfect for real-time, mission-critical app

Pushy 1 Dec 21, 2021
Send automated wishes to your contacts at scheduled time through WhatsApp. Written for Raspberry pi.

Whatsapp Automated Wishes Helps to send automated wishes to your contacts in Whatsapp at scheduled time using pywhatkit . Written for Raspberry pi. Wi

Uthayamurthy 2 Dec 13, 2022
TwitchAccountMaker - Twitch Account Maker with python

Twitch Account Creator A Twitch Account Creator, Requires Capmonster.cloud Verif

vanis / 1800 0 Jan 20, 2022
This repository is used to provide data to zzhack,

This repository is used to provide data to zzhack, but you don't have to care about anything, just write your thinking down, and you can see your thinking is rendered in zzhack perfectly

5 Apr 29, 2022
PackMyPayload - Emerging Threat of Containerized Malware

This tool takes a file or directory on input and embeds them into an output file acting as an archive/container.

Mariusz Banach 594 Dec 29, 2022
A Python Script to automate searching of available vaccination centers in the city and hence booking

Cowin Vaccine Availability Notifier Cowin Vaccine Availability Notifier takes your City or PIN code as an input and automatically notifies you via ema

Jayesh Padhiar 7 Sep 05, 2021
An Anime Theme Telegram group management bot. With lot of features.

Emilia Project Emilia-Prjkt is a modular bot running on python3 with anime theme and have a lot features. Easiest Way To Deploy On Heroku This Bot is

ZenitsuID #M•R•T™ 3 Feb 03, 2022
Find songs by lyrics.

LyricSearch Hi, welcome to LyricSearch - a simple (Yes), fast (Maybe), and powerful (Approach) lyric search engine. We support Three search methods to

Dicer_ 1 Dec 13, 2021
This is a Telegram video compress bot repo. By Binary Tech💫

This is a Telegram Video Compress Bot. Prouduct By Binary Tech 💫 Features Compresse videos and generate screenshots too.You can set custom video name

silentz lk 2 Jan 06, 2022
Best DDoS Attack Script Python3, Cyber Attack With 40 Methods

MXDDoS - DDoS Attack Script With 40 Methods (Code Lang - Python 3) Please Don't Attack '.gov' and '.ir' Websites :) Features And Methods 💣 Layer7 GET

7 Mar 07, 2022