Introduces a new db key "<service>-previous" to track which instance was
last selected for a particular service. This allows for filtering the
list of available instances to exclude the instance that was last
picked, to ensure a (slightly) more even distribution of traffic.
There's still the possiblity of the following scenario, however:
:service instances > 2
/:service request #1 -> instance #1
/:service request #2 -> instance #2
/:service request #3 -> instance #1
/:service request #4 -> instance #2
where there are many ignored instances for a particular service. One
possible solution would be to implement the "<service>-previous" value
to be a list, rather than a single value, and push to that list until
only one element is left in the original "instance" array after
filtering, and then delete the "<service>-previous" key.
The process of selecting a random (working) instance for a specified
service has been moved out of the router and into lib/farside.ex. Moving
forward, the router itself should have very simple and easy to follow
logic for all paths.
Service struct now defined in lib/service.ex
This makes a bit more sense now that its a shared resource, rather than
just defining it only in the update.exs script.
This introduces a number of new changes:
- Services are now inserted into redis with a prefix prepended to the
key name. This allows for easier filtering to get only live instances.
- The home page now uses an eex template for displaying all live
instances for every service, determined by the last update
- A "last_updated" field was added
- farside.ex was added to contain all functionality related to querying
for instances (WIP)
- Other improvements
The name of the project is being refactored from Privacy Revolver to
Farside. The reasoning behind this is:
1. A shorter name is easier to remember
2. It can stand for "FOSS alternative redirecting service" (which I know
doesn't encapsulate all letters from "farside", but it's close enough).
This commit also includes improvements to the update script for
determining how far along the script is.
Not sure if this is the Elixir-y way to do this, but seems more logical
than hardcoding values such as redis connection.
Also went through and improved how string formatting was performed
throughout the app. Rather than "combining" <> "strings" this way, I'm
now just doing "#${variable}#{formatting}", which looks a lot cleaner.
Router has been updated with a basic /:service/*glob endpoint, which
retrieves a list of instances for the requested service and picks a
random one to forward the user to (or falls back to a default instance
if none are found).
Should probably add a check to make sure the requested service exists
first.
Rather than use a full blown framework*, adding basic routing with
Plug.Router seems to make more sense, since I'm not planning on hosting
any content through this app. The app itself will just be endpoints for
all available services that redirect the user to an available instance
for the requested service.
Note that I might change my mind about this, but that's unlikely. At
most there would just be a home page with info about available
instances, but even then that seems kinda pointless. Trying to keep this
as absolutely simple as possible.
*like Phoenix
