Snazy is a simple tool to parse json or other type of logs and output them in a nice format with nice colors.
As a tekton
and pipelines-as-code
developer who has to dig into controller/webhook logs all the time I wanted something that is a bit easier to look in the eyes and identify error/info/warning statements easily.
You do not have to use it only with tekton
but work well with projects that uses
go-uber/zap
library like
knative
and many others.
It can work as a super-"tail" too if you want to stream logs and being able to have "actions" or "highlights" on a regexp match.
Go to the release page and grab the archive or package targeting your platform.
brew tap chmouel/snazy https://github.com/chmouel/snazy
brew install snazy
cargo install snazy
With your favourite aurhelper for example yay :
yay -S snazy-bin
Snazy is available from nixpkgs
.
nix-env -iA snazy
nix run nixpkgs#snazy -- --help # your args are here
bin install github.com/chmouel/snazy
kubectl logs deployment/pod foo|docker run -i ghcr.io/chmouel/snazy
Snazy is built with rust, if you want to compile it directly you just need to
grab the source and run cargo build
. (assuming you have the rust tool chain installed)
snazy
by "piping" logs into it :kubectl logs deployment/controller|snazy
It supports streaming too. When you have a kubectl logs -f
it will just wait
for input and snazzily print your logs from the stream (one line at a time).
you can pass one or many files on the command line to snazy
and it will
parse them rather than using the standard input.
If you do not pass a file and your input comes from
https://github.com/boz/kail it will automatically detect it and print the
namespace/pod[container]
as prefix :
The flag "--kail-prefix-format" let you customize how to display the kail
format, the templates {namespace}
, {pod}
, {container}
will be replaced
by its value and a "\n" will be replaced by a newline. As an example if you
want to only show the current pod followed by a newline you can use the
following template:
--kail-prefix-format "{pod}\n"
the environment variable SNAZY_KAIL_PREFIX_FORMAT
let you make this setting permanent.
If you do not any prefix for kail you can pass the --kail-no-prefix
flag.
If you want to highlight some patterns you can add the option -r/--regexp
followed by a REGEXP and snazy
will highlight it. You can have many -r
switches with many regexps, and you get different highlight for each match.
If you want to have the highlight forced to some colors you can add the color
at the beginning of the regexp followed by a colon. The colors can be one of
yellow
, red
, green
, blue
, magenta
, cyan
, white
, black
or an RGB
color model e.g: 88,48,235
. For example if you want to highlight ERROR in red
and WARNING in yellow you can do:
% kubectl log pod|snazy -r red:ERROR -r yellow:WARNING -r green:INFO -r 88,48,235:MITIGATED
% kubectl log pod|snazy -r fg=black,bg=yellow:ERROR
If snazy
don't recognize the line as JSON it will simply straight print
it. Either way it will still apply regexp highlighting of the -r
option or
do the action commands matching (see below). This let you use it for any logs
to do some regexp highlighting and action on pattern.
If you want to only show some specific levels, you can add the flag
-f
/--filter-levels
to filter by level or many -f
for many levels, for
example, this only show warning and error from the log:
% kubectl log pod|snazy -f warning -f error
-l/--level-symbols
or set the environment variable
SNAZY_LEVEL_SYMBOLS
, snazy will show some pretty emojis rather than plain log
level label :You can customize the time printed with the --time-format
flag (or the environment
variable SNAZY_TIME_FORMAT
), the variable respect the UNIX
strftime
format
strings.
If you want to skip showing some lines you can specify the flag
-S/--skip-line-regexp
. When it matches the word or regexp in
this value it will simply skipping printing the line. You can have multiple flags
if you want to skip multiple lines.
You can do your own field matching with the -k/--json-keys
flag, you need to pass the fields msg
, level
and ts
.
The fields target a key in a json payload specified as JSON Object notation. The description of the fileds are:
msg
: The message text (string)level
: The log level (eg: info) (string)ts
: The timestamp, a float or a datetime.If any of those fields are missing the parser will fails.
Example:
echo '{"the": {"msg": {"is": "message"}, "level": {"is": "INFO"}, "ts": [{"is": "2022-04-25T14:20:32.505637358Z"}]}}' | \
snazy -k msg=/the/msg/is -k level=/the/level/is -k ts=/the/ts/0/is
# => INFO 14:20:32 message
Snazy support action command on regexp, which mean if you have a regexp
matching a message it will run an action on it. It currently supports only one
action one regexp. If you specify the string "{}"
it will be expanded to
the matched string. For example on MacOS this command will display a
notification with the pipelinerun that has succeeded:
snazy --action-regexp "pipelinerun(s)?\s*.*has success" --action-command "osascript -e 'display notification \"{}\"'"
You can go even further with UNIX shell pipelines, and feed snazy to fzf for interactive filtering of the stream. for example to stream everything on a kubernetes cluster with kail, transforming the logs via snazy and finally using fzf to interactively select the patter to match:
kail --since=1h | snazy | fzf +s --ansi
This will give you a prompt with fzf
where you can type the query you want.
An handy script located here let you show the log
of a GitHub action runs through snazy and the
bat
pager.
You will need to setup the gh cli and install fzf to make it works.
Here is a video showing the feature https://streamable.com/7sf1hq
Shell completions are available for most shells using the command --shell-completion
for example --shell-completion=zsh
. Many different shell are supported. I let the reader figure out how to use them with their respective shells. Brew and RPM packages should have them auto configured for bash/fish/zsh.
Chmouel Boudjnah [@chmouel](https://twitter.com/chmouel)