I'm working on a live client-server system using TCP. The system must ready 24/7 with cut-off time 30 minutes per day.
For some reasons like testing, I want to add some 'shadow servers' to the system, so that clients will send request to main server and shadow servers. While the request should be sent to all server, only the reply from main server that matters thus should be sent back to client. Reply from shadow servers should be ignored.
Moreover, starting and stopping shadows server mustn't do any harm to the system. If a shadows is up, it will receive request, if it's down, nothing will happen to client-main server.
Is there a way for me to do this without changing either client or main server code?
You can't do this in client or server code because TCP is point-to-point only.
Perhaps what you want is a "frontend load balancer" that keeps track of what servers are running and forwards incoming requests to one of those.
On the high end, a NetScaler does all this. Though I haven't personally configured one of these devices, I expect it can have both "primary" and "backup" servers. When I used them, it distributed the load evenly between many servers. It can also be arranged in "high availability pairs" where two of them are connected with one as a "hot standby" in the case where the primary fails.
You probably don't need anything this sophisticated but it serves to illustrate the point.
Related
I had the following issue on a system that I supported ~7 years ago. We never got to the bottom of it, and focus shifted onto other issues. I was recently reminded of it, and wondered if anyone would know what was going on. But alas I'll be a little short on details. Sorry.
The Setup
I had a farm of web servers sitting behind a load balancer. The servers were hosting a system that would receive HTTP requests (XML &/or SOAP) from clients, then for each one kick-off a bunch of further HTTP requests to 3rd-party-suppliers, wait for the suppliers' responses, process and combine the results and respond to the client's request.
Think insurance comparison, but as Business-To-Business XML service.
The whole processing would take 5s of seconds, from receiving the initial client request to them sending back a response to that original HTTP request, and the server would be processing 10s or 100s of requests in parallel (i.e. at any given point, a given webserver would have many client Requests that had come in, and been logged, but not yet been responded to.)
We had detailed logging which record the reciept of the requests, including origin IP and which server was processing the request, and record when a response was sent.
All client requests were sent to a single IP address (well, URL), which was the address of the loadbalancer, which would then forward requests to the webservers, which weren't individually accessible to the internet (they didn't have public IP addresses).
Our load balancer would allow us to take individual web-servers out of rotation, for maintenance.
When we did that we could watch the DB logs, and see new requests stop coming in, and the existing request gradually get completed, until there were now outstanding requests and the server was idle.
The problem
We found that sometimes, when we took a server out of rotation ... it wouldn't entirely stop receiving requests. You could see the large bulk of request suddenly stop coming in, but it would still receive a trickle of fresh requests (I don't know ... maybe 0.1% of normal load, maybe less?). I think the longest we left it going was maybe ... 10 minutes?
Notably we realised that all of those requests were coming from a single client/IP address (I don't remember which).
I forget whether other (still-in-rotation) webservers were still receiving requests from this client, but I think they were?
If we rebooted the webserver, no further requests would come in after restarting.
Web stack was Windows, IIS, ASP.NET; pretty old school even at the time. All servers individually owned and configured.
What was happening?
We vaguely waved our hands and asserted that the client's integration with us was "holding an HTTP tunnel open and sending multiple requests through it", rather than sending each request separately, and thus was maintaining that tunnel even after the LB stopped sending new requests to that server. But that was BS-waffle, and since we never needed to actually understand what was going on, we ignored it and moved on with our lives :)
But I'd still like to know what we were seeing, if anyone can diagnose it from that description.
We vaguely waved our hands and asserted that the client's integration with us was "holding an HTTP tunnel open and sending multiple requests through it", rather than sending each request separately, and thus was maintaining that tunnel even after the LB stopped sending new requests to that server.
That sounds like a good explanation.
Normally, a LB will refuse new connections to a removed server, but will allow open connections to live on until they naturally close. This is known as "connection draining" or "graceful shutdown".
If one of your clients had HTTP keepalive on, and was holding a TCP connection open and sending HTTP requests through it for a long time, it would give the symptoms you describe.
Most LBs will have a configuration knob for how long to wait for connections to close before force-closing them during this "connection draining" time. You can set a timeout here to avoid this scenario if it is a problem for you.
The HTTP connection handling behaviour of clients will vary at the client's discretion, to a large extent. Perhaps most of your clients were of one type (say, web browsers) and weren't holding open a single connection for 10 mins, but perhaps one client was different (say, a programmatic HTTP API client)?
Further reading about "connection draining" on AWS Load Balancers here (the exact details will vary by LB vendor): https://docs.aws.amazon.com/elasticloadbalancing/latest/classic/config-conn-drain.html
Further reading about HTTP keep alive here: https://en.wikipedia.org/wiki/HTTP_persistent_connection
I have an ASP.NET Web API application running behind a load balancer. Some clients keep an HTTP busy connection alive for too much time, creating unnecessary affinity and causing high load on some server instances. In order to fix that, I wish to gracefully close a connection that is doing too much requests in a short period of time (thus forcing the client to reconnect and pick a different server instance) while at same time keeping low traffic connections alive indefinitely. Hence I cannot use a static configuration.
Is there some API that I can call to flag a request to "answer this then close the connection" ? Or can I simply add the Connection: close HTTP header that ASP.NET will see and close the connection for me?
It looks like the good solution for your situation will be the built-in IIS functionality called Dynamic IP restriction. "To provide this protection, the module temporarily blocks IP addresses of HTTP clients that make an unusually high number of concurrent requests or that make a large number of requests over small period of time."
It is supported by Azure Web Apps:
https://azure.microsoft.com/en-us/blog/confirming-dynamic-ip-address-restrictions-in-windows-azure-web-sites/
If that is the helpful answer, please mark it as a helpful or mark it as the answer. Thanks!
I am not 100% sure this would work in your situation, but in the past I have had to block people coming from specific IP addresses geographically and people coming from common proxies. I created an Authorized Attribute class following:
http://www.asp.net/web-api/overview/security/authentication-filters
In would dump the person out based on their IP address by returning a HttpStatusCode.BadRequest. On every request you would have to check a list of bad ips in the database and go from there. Maybe you can handle the rest client side, because they are going to get a ton of errors.
Write an action filter that returns a 302 Found response for the 'blocked' IP address. I would hope, the client would close the current connection and try again on the new location (which could just be the same URL as the original request).
I am trying to create a Web Server of my own and there are several questions about working of Web servers we are using today. Questions are:
After receiving a HTTP request from a client through port 80, does server respond using same port 80?
If yes then while sending a large file say a pic in MB's, webserver will be unable to receive requests from other clients?
Is a computer port duplex or simplex? (Can it send and receive at the same time)?
If another port on server side is used to send response to client, then (if TCP is used, which is generally used), again 3-way handshaking will be done which will be overhead...
http://beej.us/guide/bgnet/output/html/singlepage/bgnet.html here is a good guide on what's going on with webservers, although it's in c but the concepts are all there. This will explain the whole client server relationship as well as some implementation details.
I'll just give a high level on what's going on:
Usually what happens is when your server gets a new request that comes in it creates a fork that will process it, that way you are not bogged down by each request, when the request comes in the child process is handed a new file to write to(again this is all implementation details).
So really you have one server waiting for requests and for each request it received it spawns a child to process to deal with this request. I'm sure there are much easier languages to implement this stuff than c(I had to do both a c and java server serving to either one in my past) but c really gets you to understand the things that are going on and I'm betting that is what you are looking for here
Now there are a couple of things to think about:
how you want the webserver to work. The example explains the parent child process.
Do you want to use tcp/UDP there are differences in the way to payload gets delivered.
You don't have to connect on port 80. that's just the default for web.
Hopefully the guide will help you.
Yes. The server sends the response using the TCP connection established by the client, so it also responds using the same port. The server can handle connections from multiple clients using the same port because TCP connections are identified by (local-ip, local-port, remote-ip, remote-port), so the server can even handle multiple connections from same client provided that the source ports are different.
There are different techniques you can use to be able to serve multiple clients at the same time. These include
using multiple processes or threads: when one is busy serving a client the others can serve other clients.
using events: the server listens for events from the OS: when it can write a block of data to a connection it writes it, when a new client connects it accepts the connection, ...
Frequently both approaches are be combined.
A TCP connection is duplex: you can send and receive at the same time. The HTTP protocol is based on a simple request-response model though: at any given time only one party is "talking."
I'm making a multiplayer game and often i want to test out if it perfectly works on global network, because sometime it's just works locally, so how could i do that without sending my client to friend to test it out.
If you want to test it for the "global network" - you have to test it that way. There are multiple things that can go wrong which are not an issue on a local network. Just off the top of my head
latency (important for a game)
NAT (common cause of problems depending on your game architecture - more so if P2P)
security
connection errors (Wifi/3G intermittent loss)
There are many aspects of networking that are often subtly different when you're talking over the internet rather than running on either localhost or a local network.
All manner of delays can occur, and this can throw out some poor assumptions in your code.
The TCP flow can be affected by TCP's flow control (when the TCP Window fills up the sender will stop sending and report the fact to you (or maybe not report it, if you're using async APIs)).
TCP reads that return 'complete messages' on localhost and your own network may start to return pieces of a message.
UDP datagrams may go missing and never arrive or may arrive multiple times or in any sequence.
So you're right to think that you need to test your code for these edge cases which rarely show up on your own network. You're also right to think that simply sending a client to a friend, or running it on a remote machine, is not enough.
One approach is to build a dedicated test client which sends known game play and checks that it gets expected responses (how hard this is depends on your protocol and your game). Once you have that working you then have the test client deliberately send data in such a way that the items above are tested. So, if you're using UDP, you might put some code in your test client so that it sometimes doesn't bother to send a UDP datagram at all. The client should think it sent it. The networking layer simply ditches it. This tests your UDP protocol for missing datagrams. Then send some datagrams multiple times, then send some out of sequence, etc. For TCP add delays, break logical "messages" into separate network sends with large delays between them; ideally send each distinct message type as a sequence of single bytes to check that the server's 'message accumulation code' works correctly.
Once you have done this you need to do the same for your client code, perhaps by adding a "fuzzing" option to your server's network code to do the same kind of thing...
Personally I tend to try and take a step back and do as much of this as possible in dedicated 'unit tests' (I know that some people will say that these aren't unit tests, call them what you like, just write them!). These tests exercise your networking layer using real networking (talking to a dummy server/client that the test creates) and validate the horrible edge cases.
I have developed serve-client model based on UDP. Client are connected to server on random basis. I mean number of clients alive at a time is not fixed.
Any new client can communicate any time. It means, there could be 1 live client or 100 clients or any number of clients.
Now in such model, I need to add HTTP requests. Browser could send request to server and then server will forward that to any of client based on some identification.
Is there any method or readymade server(like nginix or lighttpd), which I can use for this requirement.
My big worry is that, destination client are not fixed, they keep changing. Most of server (nginix or lighttd) have static entries for destination address.
I visualize your scenario as multiple sensors that connect to the servers when they have something to say, and then they send a request and wait for the answer.
I visualize you also want to somehow administer such modules so that you want to access to them via HTTP.
You could leave the new configuration items on the regular server so that upon any update connection the response would include (in a piggy-backed fashion) the changes to the node.
Or the server could mark somehow your interest in accessing a certain node, and then, when this connects, the server could notify the interested client. The sensor should pay attention to clients wanting to connect to them during a window time.
Certainly, more information would help us help you.