tk/Ocelot
1
0
Fork 0
mirror of https://github.com/nsnail/Ocelot.git synced 2025-04-22 18:22:49 +08:00
Code Issues Packages Projects Releases Wiki Activity

moved readme into docs

Browse Source
This commit is contained in:
TomPallister 2017-03-05 17:33:18 +00:00
parent a70a279487
commit c3c37bc785
1 changed files with 4 additions and 504 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

508
README.md
View File

@ -26,7 +26,7 @@ it reaches a request builder middleware where it creates a HttpRequestMessage ob
used to make a request to a downstream service. The middleware that makes the request is used to make a request to a downstream service. The middleware that makes the request is
the last thing in the Ocelot pipeline. It does not call the next middleware. the last thing in the Ocelot pipeline. It does not call the next middleware.
The response from the downstream service is stored in a per request scoped repository The response from the downstream service is stored in a per request scoped repository
and retrived as the requests goes back up the Ocelot pipeline. There is a piece of middleware and retrieved as the requests goes back up the Ocelot pipeline. There is a piece of middleware
that maps the HttpResponseMessage onto the HttpResponse object and that is returned to the client. that maps the HttpResponseMessage onto the HttpResponse object and that is returned to the client.
That is basically it with a bunch of other features. That is basically it with a bunch of other features.
@ -35,34 +35,12 @@ That is basically it with a bunch of other features.
Pull requests, issues and commentary welcome! No special process just create a request and get in Pull requests, issues and commentary welcome! No special process just create a request and get in
touch either via gitter or create an issue. touch either via gitter or create an issue.
## Building Ocelot
You should be able to just build Ocelot using the ./build.ps1 or ./build.sh scripts if you are on
windows or mac. Alternatively you can build the project in VS2015 with the latest .NET Core SDK.
The tests should all just run and work apart from the integration tests which need the following
environmental variables setting. This is a manual step at the moment.
OCELOT_USERNAME=admin
OCELOT_HASH=kE/mxd1hO9h9Sl2VhGhwJUd9xZEv4NP6qXoN39nIqM4=
OCELOT_SALT=zzWITpnDximUNKYLiUam/w==
On windows you can use..
SETX OCELOT_USERNAME admin
On mac..
export OCELOT_USERNAME=admin
I need to work out a nicer way of doing this in the future.
## How to install ## How to install
Ocelot is designed to work with ASP.NET core only and is currently Ocelot is designed to work with ASP.NET core only and is currently
built to netcoreapp1.1 [this](https://docs.microsoft.com/en-us/dotnet/articles/standard/library) documentation may prove helpful when working out if Ocelot would be suitable for you. built to netcoreapp1.1 [this](https://docs.microsoft.com/en-us/dotnet/articles/standard/library) documentation may prove helpful when working out if Ocelot would be suitable for you.
Install Ocelot and it's dependecies using nuget. At the moment Install Ocelot and it's dependencies using NuGet. At the moment
all we have is the pre version. Once we have something working in all we have is the pre version. Once we have something working in
a half decent way we will drop a version. a half decent way we will drop a version.
@ -70,487 +48,9 @@ a half decent way we will drop a version.
All versions can be found [here](https://www.nuget.org/packages/Ocelot/) All versions can be found [here](https://www.nuget.org/packages/Ocelot/)
## Configuration ## Documentation
An example configuration can be found [here](https://github.com/TomPallister/Ocelot/blob/develop/test/Ocelot.ManualTest/configuration.json) Please click [here](https://github.com/TomPallister/Ocelot/wiki) for the Ocleot documentation.
and an explained configuration can be found [here](https://github.com/TomPallister/Ocelot/blob/develop/configuration-explanation.txt). More detailed instructions to come on how to configure this.
There are two sections to the configuration. An array of ReRoutes and a GlobalConfiguration.
The ReRoutes are the objects that tell Ocelot how to treat an upstream request. The Global
configuration is a bit hacky and allows overrides of ReRoute specific settings. It's useful
if you don't want to manage lots of ReRoute specific settings.
```json
{
"ReRoutes": [],
"GlobalConfiguration": {}
}
```
More information on how to use these options is below..
## Startup
An example startup using a json file for configuration can be seen below.
Currently this is the only way to get configuration into Ocelot.
```csharp
public class Startup
{
public Startup(IHostingEnvironment env)
{
var builder = new ConfigurationBuilder()
.SetBasePath(env.ContentRootPath)
.AddJsonFile("appsettings.json", optional: true, reloadOnChange: true)
.AddJsonFile($"appsettings.{env.EnvironmentName}.json", optional: true)
.AddJsonFile("configuration.json")
.AddEnvironmentVariables();
Configuration = builder.Build();
}
public IConfigurationRoot Configuration { get; }
public void ConfigureServices(IServiceCollection services)
{
Action<ConfigurationBuilderCachePart> settings = (x) =>
{
x.WithMicrosoftLogging(log =>
{
log.AddConsole(LogLevel.Debug);
})
.WithDictionaryHandle();
};
services.AddOcelotOutputCaching(settings);
services.AddOcelot(Configuration);
}
public async void Configure(IApplicationBuilder app, IHostingEnvironment env, ILoggerFactory loggerFactory)
{
loggerFactory.AddConsole(Configuration.GetSection("Logging"));
await app.UseOcelot();
}
}
}
```
Then in your Program.cs you will want to have the following. This can be changed if you
don't wan't to use the default url e.g. UseUrls(someUrls) and should work as long as you keep the WebHostBuilder registration.
```csharp
public class Program
{
public static void Main(string[] args)
{
IWebHostBuilder builder = new WebHostBuilder();
builder.ConfigureServices(s => {
s.AddSingleton(builder);
});
builder.UseKestrel()
.UseContentRoot(Directory.GetCurrentDirectory())
.UseStartup<Startup>();
var host = builder.Build();
host.Run();
}
}
```
Sadly we need to inject the IWebHostBuilder interface to get the applications scheme, url and port later. I cannot
find a better way of doing this at the moment without setting this in a static or some kind of config.
This is pretty much all you need to get going.......more to come!
## Routing
Ocelot's primary functionality is to take incomeing http requests and forward them on
to a downstream service. At the moment in the form of another http request (in the future
this could be any transport mechanism.).
Ocelot always adds a trailing slash to an UpstreamPathTemplate.
Ocelot's describes the routing of one request to another as a ReRoute. In order to get
anything working in Ocelot you need to set up a ReRoute in the configuration.
```json
{
"ReRoutes": [
]
}
```
In order to set up a ReRoute you need to add one to the json array called ReRoutes like
the following.
```json
{
"DownstreamPathTemplate": "/api/posts/{postId}",
"DownstreamScheme": "https",
"DownstreamPort": 80,
"DownstreamHost" "localhost"
"UpstreamPathTemplate": "/posts/{postId}",
"UpstreamHttpMethod": "Put"
}
```
The DownstreamPathTemplate,Scheme, Port and Host make the URL that this request will be forwarded to.
The UpstreamPathTemplate is the URL that Ocelot will use to identity which
DownstreamPathTemplate to use for a given request. Finally the UpstreamHttpMethod is used so
Ocelot can distinguish between requests to the same URL and is obviously needed to work :)
In Ocelot you can add placeholders for variables to your Templates in the form of {something}.
The placeholder needs to be in both the DownstreamPathTemplate and UpstreamPathTemplate. If it is
Ocelot will attempt to replace the placeholder with the correct variable value from the
Upstream URL when the request comes in.
At the moment without any configuration Ocelot will default to all ReRoutes being case insensitive.
In order to change this you can specify on a per ReRoute basis the following setting.
```json
"ReRouteIsCaseSensitive": true
```
This means that when Ocelot tries to match the incoming upstream url with an upstream template the
evaluation will be case sensitive. This setting defaults to false so only set it if you want
the ReRoute to be case sensitive is my advice!
## Administration
Ocelot supports changing configuration during runtime via an authenticated HTTP API. The API is authenticated
using bearer tokens that you request from iteself. This is provided by the amazing [IdentityServer](https://github.com/IdentityServer/IdentityServer4)
project that I have been using for a few years now. Check them out.
In order to enable the administration section you need to do a few things. First of all add this to your
initial configuration.json. The value can be anything you want and it is obviously reccomended don't use
a url you would like to route through with Ocelot as this will not work. The administration uses the
MapWhen functionality of asp.net core and all requests to root/administration will be sent there not
to the Ocelot middleware.
```json
"GlobalConfiguration": {
"AdministrationPath": "/administration"
}
```
This will get the admin area set up but not the authentication. Please note that this is a very basic approach to
this problem and if needed we can obviously improve on this!
You need to set 3 environmental variables.
OCELOT_USERNAME
OCELOT_HASH
OCELOT_SALT
These need to be the admin username you want to use with Ocelot and the hash and salt of the password you want to
use given hashing algorythm. When requesting bearer tokens for use with the administration api you will need to
supply username and password.
In order to create a hash and salt of your password please check out HashCreationTests.should_create_hash_and_salt()
this technique is based on [this](https://docs.microsoft.com/en-us/aspnet/core/security/data-protection/consumer-apis/password-hashing)
using SHA256 rather than SHA1.
Now if you went with the configuration options above and want to access the API you can use the postman scripts
called ocelot.postman_collection.json in the solution to change the Ocelot configuration. Obviously these
will need to be changed if you are running Ocelot on a different url to http://localhost:5000.
The scripts show you how to request a bearer token from ocelot and then use it to GET the existing configuration and POST
a configuration.
## Service Discovery
Ocelot allows you to specify a service discovery provider and will use this to find the host and port
for the downstream service Ocelot is forwarding a request to. At the moment this is only supported in the
GlobalConfiguration section which means the same service discovery provider will be used for all ReRoutes
you specify a ServiceName for at ReRoute level.
In the future we can add a feature that allows ReRoute specfic configuration.
At the moment the only supported service discovery provider is Consul. The following is required in the
GlobalConfiguration. The Provider is required and if you do not specify a host and port the Consul default
will be used.
```json
"ServiceDiscoveryProvider": {
"Provider":"Consul",
"Host":"localhost",
"Port":8500
}
```
In order to tell Ocelot a ReRoute is to use the service discovery provider for its host and port you must add the
ServiceName and load balancer you wish to use when making requests downstream. At the moment Ocelot has a RoundRobin
and LeastConnection algorithm you can use. If no load balancer is specified Ocelot will not load balance requests.
```json
{
"DownstreamPathTemplate": "/api/posts/{postId}",
"DownstreamScheme": "https",
"UpstreamPathTemplate": "/posts/{postId}",
"UpstreamHttpMethod": "Put",
"ServiceName": "product",
"LoadBalancer": "LeastConnection"
}
```
When this is set up Ocelot will lookup the downstream host and port from the service discover provider and load balancer
requests across any available services.
## Authentication
Ocelot currently supports the use of bearer tokens with Identity Server (more providers to
come if required). In order to identity a ReRoute as authenticated it needs the following
configuration added.
```json
"AuthenticationOptions": {
"Provider": "IdentityServer",
"ProviderRootUrl": "http://localhost:52888",
"ScopeName": "api",
"AdditionalScopes": [
"openid",
"offline_access"
],
"ScopeSecret": "secret"
}
```
In this example the Provider is specified as IdentityServer. This string is important
because it is used to identity the authentication provider (as previously mentioned in
the future there might be more providers). Identity server requires that the client
talk to it so we need to provide the root url of the IdentityServer as ProviderRootUrl.
IdentityServer requires at least one scope and you can also provider additional scopes.
Finally if you are using IdentityServer reference tokens you need to provide the scope
secret.
Ocelot will use this configuration to build an authentication handler and if
authentication is succefull the next middleware will be called else the response
is 401 unauthorised.
## Authorisation
Ocelot supports claims based authorisation which is run post authentication. This means if
you have a route you want to authorise you can add the following to you ReRoute configuration.
```json
"RouteClaimsRequirement": {
"UserType": "registered"
},
```
In this example when the authorisation middleware is called Ocelot will check to see
if the user has the claim type UserType and if the value of that claim is registered.
If it isn't then the user will not be authorised and the response will be 403 forbidden.
## Claims Tranformation
Ocelot allows the user to access claims and transform them into headers, query string
parameters and other claims. This is only available once a user has been authenticated.
After the user is authenticated we run the claims to claims transformation middleware.
This allows the user to transform claims before the authorisation middleware is called.
After the user is authorised first we call the claims to headers middleware and Finally
the claims to query strig parameters middleware.
The syntax for performing the transforms is the same for each proces. In the ReRoute
configuration a json dictionary is added with a specific name either AddClaimsToRequest,
AddHeadersToRequest, AddQueriesToRequest.
Note I'm not a hotshot programmer so have no idea if this syntax is good..
Within this dictionary the entries specify how Ocelot should transform things!
The key to the dictionary is going to become the key of either a claim, header
or query parameter.
The value of the entry is parsed to logic that will perform the transform. First of
all a dictionary accessor is specified e.g. Claims[CustomerId]. This means we want
to access the claims and get the CustomerId claim type. Next is a greater than (>)
symbol which is just used to split the string. The next entry is either value or value with
and indexer. If value is specifed Ocelot will just take the value and add it to the
transform. If the value has an indexer Ocelot will look for a delimiter which is provided
after another greater than symbol. Ocelot will then split the value on the delimiter
and add whatever was at the index requested to the transform.
#### Claims to Claims Tranformation
Below is an example configuration that will transforms claims to claims
```json
"AddClaimsToRequest": {
"UserType": "Claims[sub] > value[0] > |",
"UserId": "Claims[sub] > value[1] > |"
},
```
This shows a transforms where Ocelot looks at the users sub claim and transforms it into
UserType and UserId claims. Assuming the sub looks like this "usertypevalue|useridvalue".
#### Claims to Headers Tranformation
Below is an example configuration that will transforms claims to headers
```json
"AddHeadersToRequest": {
"CustomerId": "Claims[sub] > value[1] > |"
},
```
This shows a transform where Ocelot looks at the users sub claim and trasnforms it into a
CustomerId header. Assuming the sub looks like this "usertypevalue|useridvalue".
#### Claims to Query String Parameters Tranformation
Below is an example configuration that will transforms claims to query string parameters
```json
"AddQueriesToRequest": {
"LocationId": "Claims[LocationId] > value",
},
```
This shows a transform where Ocelot looks at the users LocationId claim and add its as
a query string parameter to be forwarded onto the downstream service.
## Quality of Service
Ocelot supports one QoS capability at the current time. You can set on a per ReRoute basis if you
want to use a circuit breaker when making requests to a downstream service. This uses the an awesome
.NET library called Polly check them out [here](https://github.com/App-vNext/Polly).
Add the following section to a ReRoute configuration.
```json
"QoSOptions": {
"ExceptionsAllowedBeforeBreaking":3,
"DurationOfBreak":5,
"TimeoutValue":5000
}
```
You must set a number greater than 0 against ExceptionsAllowedBeforeBreaking for this rule to be
implemented. Duration of break is how long the circuit breaker will stay open for after it is tripped.
TimeoutValue means ff a request takes more than 5 seconds it will automatically be timed out.
If you do not add a QoS section QoS will not be used.
## RequestId / CorrelationId
Ocelot supports a client sending a request id in the form of a header. If set Ocelot will
use the requestid for logging as soon as it becomes available in the middleware pipeline.
Ocelot will also forward the request id with the specified header to the downstream service.
I'm not sure if have this spot on yet in terms of the pipeline order becasue there are a few logs
that don't get the users request id at the moment and ocelot just logs not set for request id
which sucks. You can still get the framework request id in the logs if you set
IncludeScopes true in your logging config. This can then be used to match up later logs that do
have an OcelotRequestId.
In order to use the requestid feature in your ReRoute configuration add this setting
```json
"RequestIdKey": "OcRequestId"
```
In this example OcRequestId is the request header that contains the clients request id.
There is also a setting in the GlobalConfiguration section which will override whatever has been
set at ReRoute level for the request id. The setting is as fllows.
```json
"RequestIdKey": "OcRequestId",
```
It behaves in exactly the same way as the ReRoute level RequestIdKey settings.
## Caching
Ocelot supports some very rudimentary caching at the moment provider by
the [CacheManager](http://cachemanager.net/) project. This is an amazing project
that is solving a lot of caching problems. I would reccomend using this package to
cache with Ocelot. If you look at the example [here](https://github.com/TomPallister/Ocelot/blob/develop/test/Ocelot.ManualTest/Startup.cs)
you can see how the cache manager is setup and then passed into the Ocelot
AddOcelotOutputCaching configuration method. You can use any settings supported by
the CacheManager package and just pass them in.
Anyway Ocelot currently supports caching on the URL of the downstream service
and setting a TTL in seconds to expire the cache. More to come!
In orde to use caching on a route in your ReRoute configuration add this setting.
```json
"FileCacheOptions": { "TtlSeconds": 15 }
```
In this example ttl seconds is set to 15 which means the cache will expire after 15 seconds.
## Ocelot Middleware injection and overrides
Warning use with caution. If you are seeing any exceptions or strange behavior in your middleware
pipeline and you are using any of the following. Remove them and try again!
When setting up Ocelot in your Startup.cs you can provide some additonal middleware
and override middleware. This is done as follos.
```csharp
var configuration = new OcelotMiddlewareConfiguration
{
PreErrorResponderMiddleware = async (ctx, next) =>
{
await next.Invoke();
}
};
app.UseOcelot(configuration);
```
In the example above the provided function will run before the first piece of Ocelot middleware.
This allows a user to supply any behaviours they want before and after the Ocelot pipeline has run.
This means you can break everything so use at your own pleasure!
The user can set functions against the following.
+ PreErrorResponderMiddleware - Already explained above.
+ PreAuthenticationMiddleware - This allows the user to run pre authentication logic and then call
Ocelot's authentication middleware.
+ AuthenticationMiddleware - This overrides Ocelots authentication middleware.
+ PreAuthorisationMiddleware - This allows the user to run pre authorisation logic and then call
Ocelot's authorisation middleware.
+ AuthorisationMiddleware - This overrides Ocelots authorisation middleware.
+ PreQueryStringBuilderMiddleware - This alows the user to manipulate the query string on the
http request before it is passed to Ocelots request creator.
Obviously you can just add middleware as normal before the call to app.UseOcelot() It cannot be added
after as Ocelot does not call the next middleware.
## Logging
Ocelot uses the standard logging interfaces ILoggerFactory / ILogger<T> at the moment.
This is encapsulated in IOcelotLogger / IOcelotLoggerFactory with an implementation
for the standard asp.net core logging stuff at the moment.
There are a bunch of debugging logs in the ocelot middlewares however I think the
system probably needs more logging in the code it calls into. Other than the debugging
there is a global error handler that should catch any errors thrown and log them as errors.
The reason for not just using bog standard framework logging is that I could not
work out how to override the request id that get's logged when setting IncludeScopes
to true for logging settings. Nicely onto the next feature.
## Not supported
Ocelot does not support...
+ Chunked Encoding - Ocelot will always get the body size and return Content-Length
header. Sorry if this doesn't work for your use case!
+ Fowarding a host header - The host header that you send to Ocelot will not be
forwarded to the downstream service. Obviously this would break everything :(
## Things that are currently annoying me ## Things that are currently annoying me