Architecture overview

This document outlines basic architectural principles of Uliska scanner.

Commands and output

Main principle is simple: take list of commands, execute and save result in easy to produce and easy to parse format. Preferably output format should be some kind of structured data format (data serialization).

Output format - YAML

• YAML is easy to produce: prepend lines of STDOUT with several spaces;

• it is easy to parse: almost any programming language has parser for YAML;

• as a bonus it's human readable.

Command lists

Command lists in Uliska are plain text files with one command per line, with some additional sugar related to command aliases.

Aliasing gives an ability to have similar output data structure across different varieties of UNIX systems.

Components

Main scanner components are:

• main script uliska.pl

• Perl module ./lib/Uliska.pm

• additional Perl module(s)s ./lib/* (currently only one module included - YAML);

• command lists sub-directory ./cfg/**/*.cfg

• modules-'workers' in ./cfg/**/*.pm

Perl modules for Uliska scanner

See in separate window

Command lists vs Perl modules

Perl modules provide minimal logical glue, while main workhorse is list of shell commands.

Directory structure

All scripts and command lists in Uliska are organized into hierarchical structure of sub-directories (name-space) reflecting scanned systems properties (kernel name and versions, distribution name, hardware vendor, etc).

Execution work-flow

Top level files and directories

Uliska execution starts from calling reading and executing command list from ./cfg/generic.cfg file. This file contains commands, supposed to be present in all supported variants and dialects of UNIX. After this step Uliska must be able to find at least kernel name and version of OS it is being run on.

At the end of execution uliska.pl tries:

• read and execute commands from files cfg/<kernel>.cfg, cfg/<kernel>/<major>.cfg, cfg/<kernel>/<major>/<minor>.cfg;

• and to load additional modules lib/Uliska/<kernel>.pm, lib/Uliska/<kernel>/<major>.pm, lib/Uliska/<kernel>/<major>/<minor>.pm and execute them;

where

• <kernel> - is OS kernel name as returned by uname -s (Linux, SunOS), case sensitive;

• <major> and <minor> - are kernel release numbers, returned by uname -r.

Specific commands and modules

Each loadable module should have subroutine named run(). It is executed after module load has succeeded.

OS specific modules and commands

All modules are built in a similar way:

• module performs some commands to detect specific information about OS and/or host ;

• module load additional modules based on discovered data;

• execute modules by running modules' subroutine run().

For more information see Extending file.

Requirements

Current requirements for Uliska.pm are: strict, warnings, and YAML. First two should not be a problem on any reasonable Perl installation, as for last one -- YAML -- please be sure to include YAML module with Uliska, if it is missing on target hosts.

YAML::Old from CPAN http://search.cpan.org/~ingy/YAML-Old-0.81/ included to make Uliska package self-contained. It can be replaced with any existing Perl YAML dumper.

LocalWords: Uliska YAML uliska uname SunOS