/mayday

utility for gathering support information

Primary LanguageGoApache License 2.0Apache-2.0

mayday

...man overboard!



     ___ ___   ____  __ __  ___     ____  __ __
    |   |   | /    ||  |  ||   \   /    ||  |  |
    | _   _ ||  o  ||  |  ||    \ |  o  ||  |  |
    |  \_/  ||     ||  ~  ||  D  ||     ||  ~  |
    |   |   ||  _  ||___, ||     ||  _  ||___, |
    |   |   ||  |  ||     ||     ||  |  ||     |
    |___|___||__|__||____/ |_____||__|__||____/




~~~~~~~~~~~~~~~~~~~~~~~~~\o/~~~~~~~~~~~~~~~~~~~~~~~~

                                   ><>
               <><  

overview

Mayday is a tool to simplify gathering support information. It is built in the spirit of sysreport, son of sysreport (sosreport), and similar support tools.
Mayday gathers information about the configuration, hardware, and running state of a system.

goals

The goals of mayday are:

  • simplify gathering information about a running system into a single command
  • collect information into one single file to be transferred to support staff
  • when possible the file should be small enough to be sent via email (<10MB)
  • not collect sensitive information like crypto keys, password hashes, etc
  • extensible through plugin system

usage

In it's most simplistic form, all a user needs to do is run mayday:

$ mayday

This will collect a basic set of data and emit it in a tar archive for transmission to a systems administrator, site reliability engineer, or support technician for further troubleshooting.

In addition, more data can be collected by running as the superuser:

$ sudo mayday

If permissions are missing or if a dependency is not installed, the report files for that check will be empty. Running as superuser will ensure that all permissions are given to access the kernel logs, pstore content, coredump info, and so on.

Even more data can be collected by adding the --danger flag:

$ sudo mayday --danger

what's collected

By default, mayday operates in a "safe" mode. No sensitive information is collected -- only information that support might need. This includes things like:

  • network connections, firewall rules, and hostname
  • information about currently running processes, including open files and ports
  • the system journal with the kernel log via journalctl, depending on the user this needs sudo
  • pstore dmesg logs, normally needs sudo
  • log files from systemd services
  • the list of logged coredumps (depends on systemd-coredump), works best with sudo
  • filesystem and memory usage information
  • information about docker and rkt containers, including network and state but NOT logs

The following is collected only if the --danger flag is activated:

  • logs and environment variables of docker and rkt containers

The following information is never collected:

  • private keys
  • passwords

integration

about

Mayday can be integrated into other projects by defining a default configuration file at either the location /etc/mayday/default.json or /usr/share/mayday/default.json. Through the use of viper YAML and TOML are now supported as well, though Flatcar Container Linux will continue to use JSON as the mechanism of choice. If multiple products are to be supported specialized configurations can be provided as "profiles" located in the above directories (e.g. /etc/mayday/quay.json) and the referenced via:

$ mayday -p quay

configuration syntax

The configuration file is comprised of objects (As of 1.0.0 valid objects are "files" and "commands"). A example of the syntax can be seen in the file default.json. Each top level object contains an array of the relevant items to collect. Optionally items can be annotated with a "link" which will provide an easy to locate pointer for commonly accessed data.

collection

Files are directly retrieved. Commands are executed and the results of standard output (stdout) are collected. Assets are placed into a Go "tarable" interface and then gzipped and serialized out to a file on disk.