Less restrictive Redis sanitization

[luismiramirez]

We recently changed how MongoDB queries are sanitized to show more information to our users. (#819)

We can do the same to Redis commands. Currently, we are sanitizing the whole query and only showing the called command.

Example:

SET key value

SET ? ?

The idea is not to remove the command's first argument to know which keys are set up or retrieved.

appsignal-ruby/lib/appsignal/integrations/redis.rb

Lines 7 to 16 in d73905d

	sanitized_command =
	if command[0] == :eval
	"#{command[1]}#{" ?" * (command.size - 3)}"
	else
	"#{command[0]}#{" ?" * (command.size - 1)}"
	end
	Appsignal.instrument "query.redis", id, sanitized_command do
	super
	end

ToDo

• Allow the first argument of Redis commands
• Test properly with an alpha release to make sure we don't flood the DB with the new unique events

[tombruijn]

First step: can we track which command is run, not just "eval".
Can we stop replacing keys with ? as well, but only replace values? Can we distinguish between them?