Permalink
305 lines (207 sloc) 10.7 KB

papertrail command-line tail & search client for Papertrail log management service

Build Status

Small standalone binary to retrieve, search, and tail recent app server log and system syslog messages from Papertrail.

Supports optional Boolean search queries and polling for new events (like "tail -f"). Example:

$ papertrail -f "(www OR db) (nginx OR pgsql) -accepted"

Output is line-buffered so it can be fed into a pipe, like for grep. ANSI color codes are rendered in color on suitable terminals; see below for additional colorization options.

The Connection class can be used by other apps to perform one-off API searches or follow (tail) events matching a given query. Interface may change.

Also includes papertrail-add-system, papertrail-remove-system, papertrail-add-group, and papertrail-join-group binaries, which invoke the corresponding Papertrail HTTP API call.

Quick Start

$ [sudo] gem install papertrail
$ echo "token: 123456789012345678901234567890ab" > ~/.papertrail.yml
$ papertrail

Retrieve the token from Papertrail User Profile.

The API token can also be passed in the PAPERTRAIL_API_TOKEN environment variable instead of a configuration file. Example:

$ export PAPERTRAIL_API_TOKEN='abc123'
$ papertrail

Installation

Install the gem (details on RubyGems), which includes a binary called "papertrail":

$ [sudo] gem install papertrail

Configuration

Create ~/.papertrail.yml containing your API token, or specify the path to that file with -c. Example (from examples/papertrail.yml.example):

token: 123456789012345678901234567890ab

Retrieve token from Papertrail User Profile. For compatibility with older config files, username and password keys are also supported.

You may want to alias "pt" to "papertrail", like:

echo "alias pt=papertrail" >> ~/.bashrc

Usage & Examples

$ papertrail --help
papertrail - command-line tail and search for Papertrail log management service
    -h, --help                       Show usage
    -f, --follow                     Continue running and printing new events (off)
        --min-time MIN               Earliest time to search from
        --max-time MAX               Latest time to search from
    -d, --delay SECONDS              Delay between refresh (2)
    -c, --configfile PATH            Path to config (~/.papertrail.yml)
    -g, --group GROUP                Group to search
    -S, --search SEARCH              Saved search to search
    -s, --system SYSTEM              System to search
    -j, --json                       Output raw JSON data (off)
        --color [program|system|all|off]
                                     Attribute(s) to colorize based on (program)
        --force-color                Force use of ANSI color characters even on non-tty outputs (off)
    -V, --version                    Display the version and exit

  Usage:
    papertrail [-f] [--min-time time] [--max-time time] [-g group] [-S search]
      [-s system] [-d seconds] [-c papertrail.yml] [-j] [--color attributes]
      [--force-color] [--] [query]

  Examples:
    papertrail -f
    papertrail something
    papertrail 1.2.3 Failure
    papertrail -s ns1 "connection refused"
    papertrail -f "(www OR db) (nginx OR pgsql) -accepted"
    papertrail -f -g Production --color all "(nginx OR pgsql) -accepted"
    papertrail --min-time 'yesterday at noon' --max-time 'today at 4am' -g Production
    papertrail -- -redis

  More: https://github.com/papertrail/papertrail-cli
        https://papertrailapp.com/

Count, pivot, and summarize

To count the number of matches, pipe to wc -l. For example, count how many logs contained Failure in the last minute:

$ papertrail --min-time '1 minute ago' Failure | wc -l
42

Output only the program/file name (which is output as field 5):

$ papertrail --min-time '1 minute ago' | cut -f 5 -d ' '
passenger.log:
sshd:
app/web.2:

Count by source/system name (field 4):

$ papertrail --min-time '1 minute ago' | cut -f 4 -d ' ' | sort | uniq -c
  98 www42
  39 acmedb-core01
  2 fastly

For sum, mean, and statistics, see datamash and one-liners.

Colors

ANSI color codes are retained, so log messages which are already colorized will automatically render in color on ANSI-capable terminals.

By default, the CLI will colorize the non-body portion of each log message based on the value of the program attribute. 5 colors are available, so colors may not be unique. When the sending system name is more important than the program, use --color=system to colorize based on its value. Use --color=all to colorize based on both together.

For content-based colorization, pipe through lnav. Install lnav from your preferred package repository, such as brew install lnav or apt-get install lnav, then:

$ papertrail -f | lnav
$ papertrail --min-time "1 hour ago" error | lnav

Redirecting output

Since output is line-buffered, pipes and output redirection will automatically work:

$ papertrail | less
$ papertrail --min-time '2016-01-15 10:00:00' > logs.txt

If you frequently pipe output to a certain command, create a function which accepts optional arguments, invokes papertrail with any arguments, and pipes output to that command. For example, this pt function will pipe to lnav:

$ function pt() { papertrail -f -d 5 $* | lnav; }

Add the function line to your ~/.bashrc. It can be invoked with search parameters:

$ pt 1.2.3 Failure

UTF-8 (non-English searches)

When searching in a language other than English, if you get no matches, you may need to explicitly tell Ruby to use UTF-8. Ruby 1.9 honors the LANG shell environment variable, and your shell may not set it to UTF-8.

To test, try:

ruby -E:UTF-8 -S papertrail your_search

If that works, add -E:UTF-8 to the RUBYOPT variable to set the encoding at invocation. For example, to persist that in a .bashrc:

export RUBYOPT="-E:UTF-8"

Negation-only queries

Unix shells handle arguments beginning with hyphens (-) differently (why). Usually this is moot because most searches start with a positive match. To search only for log messages without a given string, use --. For example, to search for -whatever, run:

papertrail -- -whatever

Time zones

Times are interpreted in the client itself, which means it uses the time zone that your local PC is set to. Log timestamps are also output in the same local PC time zone.

When providing absolute times, append UTC to provide the input time in UTC. For example, regardless of the local PC time zone, this will show messages beginning from 1 PM UTC:

papertrail --min-time "2014-04-27 13:00:00 UTC"

Output timestamps will still be in the local PC time zone.

Quoted phrases

Because the Unix shell parses and strips one set of quotes around a phrase, to search for a phrase, wrap the string in both single-quotes and double-quotes. For example:

papertrail -f '"Connection reset by peer"'

Use one set of double-quotes and one set of single-quotes. The order does not matter as long as the pairs are consistent.

Note that many phrases are unique enough that searching for the words yields the same results as searching for the quoted phrase. As a result, quoting strings twice is often not actually necessary. For example, these two searches are likely to yield the same log messages, even though one is for 4 words (AND) while the other is for a phrase:

papertrail -f Connection reset by peer
papertrail -f '"Connection reset by peer"'

Multiple API tokens

To use multiple API tokens (such as for separate home and work Papertrail accounts), create a .papertrail.yml configuration file in each project's working directory and invoke the CLI in that directory. The CLI checks for .papertrail.yml in the current working directory prior to using ~/.papertrail.yml.

Alternatively, use shell aliases with different -c paths. For example:

echo "alias pt1='papertrail -c /path/to/papertrail-home.yml'" >> ~/.bashrc
echo "alias pt2='papertrail -c /path/to/papertrail-work.yml'" >> ~/.bashrc

Add/Remove Systems, Create Group, Join Group

In addition to tail and search with the papertrail binary, the gem includes 4 other binaries which wrap other parts of Papertrail's HTTP API to explicitly add or remove a system, to create a new group, and to join a system to a group.

In most cases, configuration is automatic and these are not not necessary.

To see usage, run any of these commands with --help: papertrail-add-system, papertrail-remove-system, papertrail-add-group, papertrail-join-group.

Releasing

Build

  1. Bump VERSION in lib/papertrail.rb
  2. Build the new gem: $ rake build

Install & Test

  1. Install built gem: $ gem install pkg/papertrail-0.9.17.gem
  2. Check version in rubygems: $ gem list papertrail
  3. Verify installed version matches: $ which papertrail && papertrail --version
  4. Test: $ papertrail test search string
  5. Uninstall local gem $ gem uninstall papertrail

Release

  1. Release: $ rake release
  2. Check latest published version: $ gem list --versions --remote papertrail
  3. Install release version: $ gem install papertrail
  4. Verify installed version matches: $ which papertrail && papertrail --version
  5. Test: $ papertrail test search string
  6. Party! πŸŽ‰ 🎈 🎊

Contribute

Testing:

Run all the tests with rake

To run the tests when files save, run bundle exec guard (requires ruby version >= 2.2.5).

Bug report:

  1. See whether the issue has already been reported: http://github.com/papertrail/papertrail-cli/issues/
  2. If you don't find one, create an issue with a repro case.

Enhancement or fix:

  1. Fork the project: http://github.com/papertrail/papertrail-cli
  2. Make your changes with tests.
  3. Commit the changes without changing the Rakefile or other files unrelated to your enhancement.
  4. Send a pull request.