I have a websocket server. It accepts thousands of connections from clients. Read data from and write data to clients. It will work normally for weeks. But something wrong will happen occasionally, maybe once two weeks. In a very short time, the new clients will establish connections to server and send a protocol immediately. The server side websocket.onOpen() will be invoked, but it fails to read the protocol data from client. And later the client may close the connection. But on the server side, the connections will keep in the state of CLOSE_WAIT, but never successfully closed. Via netstat I can see that the CLOSE_WAIT connections' read buffer is not empty and keep that value(never be read). So I guess that the server's failing to read data and the close FIN package leads to the connection to keep in CLOSE_WAIT state.
So I want to know under what circumstance may the websocket fail to read data from reading buffers.
Related
I was reading this post and it was saying there could be an issue with deadlocks if you send too much data without receiving. Is it bad to send the whole file over in a single send call? If so then what is the correct way of doing it?
I have tried sending large files using single send calls and wait until i receive it on the other end also. Sometimes the connection hangs. Maybe it could be a deadlock or improper use?
TL;DR: there are no problems doing large send with TCP itself, but there are problems in the specific example you cite.
As for TCP in general:
Using a large sent is not a problem. The network layer of your OS will take care of everything. All you need to do is make sure is that the data gets actually transmitted to the OS, i.e. check the result of sent and retry with everything not already covered by the previous sent. sent will just block your application if it currently cannot send (write buffer full). But this requires that the server will actually receive the data. If not then the server side read buffer will fill up which causes the TCP window to decrease and ultimately the send to stop until the server is actually reading the previously sent data.
As for your specific linked example:
In your specific linked example there is an application protocol on top of TCP which changes the semantics. It is not plain TCP anymore where the client could send without receiving, but it actually requires the client to also receive data. To cite the relevant part:
The server sends one byte for every 3 bytes received.
Thus, if you send a large amount of data, then the server will send a matching amount of data back - size being one third of what you have sent. This sender emitted data will be put in the read buffer of your socket. If you don't recv then this read buffer will get full. This will cause the client network stack to signal to the server a TCP window of 0 and the server will stop sending data.
If the TCP window is 0 then the server cannot send anymore data on this socket. This means that the server will be stuck in send. If the server cannot handle recv and send on the same socket in parallel, then the server will be stuck in send and not call recv anymore - which fill fill up the server side read buffer and ultimately cause the TCP window for data from client to server to be 0 too.
In this situation both client and server will be stuck in send since nobody is receiving the data sent by the other and thus the TCP window stays 0 in both directions - deadlock.
I am reading the documentation of Alexa Voice Service capabilities and came across the part on managing HTTP2 connection. I don't really understand how this down channel works behind the scenes. Is it using server push? Well, could server push be used to keep a long connection? Or is it just using some tricks to keep the connection alive for a very long time?
As stated on the documentation, the client needs to establish a down channel stream with the server.
Based on what I read here https://www.rfc-editor.org/rfc/rfc7540, From this state diagram:
once the stream sends a HEADER frame, followed by an END STREAM flag, the state will be half-closed(local) on the point of view of the client. So, this is how half-closed state for the device happened, as stated in above image. Correct me that if I am wrong.
For managing the HTTP connection, this is what it says.
Based on my understanding: the client sets a timeout of 60minutes for the GET request. After the request is sent, the server will not send any response. Then the connection will remain open for 60minutes. But once a response is sent from the server, the connection should be closed. Isn't that supposed to happen? Or, is it because when the server sends response through the down channel stream, it did not send an END STREAM flag so the stream will not be closed?
But once a response is sent from the server, the connection should be closed.
HTTP/1.1 and HTTP/2 use persistent connections, which means that a single connection can be used not just for one request/response, but for several request/response cycles.
Only HTTP/1.0 was closing the connection after the response, and so for HTTP/2 this is not the case, the connection will remain open until either peer decides to explicitly close it.
The recommendations about the idle timeouts are exactly to prevent the client to explicitly close the connection too early when it sees no network traffic, independently from requests or responses.
I have a client/server setup in which clients send a single request message to the server and gets a bunch of data messages back.
The server is implemented using a ROUTER socket and the clients using a DEALER. The communication is asynchronous.
The clients are typically iPads/iPhones and they connect over wifi so the connection is not 100% reliable.
The issue I’m concern about is if the client connects to the server and sends a request for data but before the response messages are delivered back the communication goes down (e.g. out of wifi coverage).
In this case the messages will be queued up on the server side waiting for the client to reconnect. That is fine for a short time but eventually I would like to drop the messages and the connection to release resources.
By checking activity/timeouts it would be possible in the server and the client applications to identify that the connection is gone. The client can shutdown the socket and in this way free resources but how can it be done in the server?
Per the ZMQ FAQ:
How can I flush all messages that are in the ZeroMQ socket queue?
There is no explicit command for flushing a specific message or all messages from the message queue. You may set ZMQ_LINGER to 0 and close the socket to discard any unsent messages.
Per this mailing list discussion from 2013:
There is no option to drop old messages [from an outgoing message queue].
Your best bet is to implement heartbeating and, when one client stops responding without explicitly disconnecting, restart your ROUTER socket. Messy, I know, this is really something that should have a companion option to HWM. Pieter Hintjens is clearly on board (he created ZMQ) - but that was from 2011, so it looks like nothing ever came of it.
This is a bit late but setting tcp keepalive to a reasonable value will cause dead sockets to close after the timeouts have expired.
Heartbeating is necessary for either side to determine the other side is still responding.
The only thing I'm not sure about is how to go about heartbeating many thousands of clients without spending all available cpu just on dealing with the heartbeats.
I'm using uwsgi's websockets support and so far it's looking great, the server detects when the client disconnects and the client as well when the server goes down. But i'm concerned this will not work in every case/browser.
In other frameworks, namely sockjs, the connection is monitored by sending regular messages that work as heartbeats/pings. But uwsgi sends PING/PONG frames (ie. not regular messages/control frames) according to the websockets spec and so from the client side i have no way to know when the last ping was received from the server. So my question is this:
If the connection is dropped or blocked by some proxy will browsers reliably (ie. Chrome, IE, Firefox, Opera) detect no PING was received from the server and signal the connection as down or should i implement some additional ping/pong system so that the connection is detected as closed from the client side?
Thanks
You are totally right. There is no way from client side to track or send ping/pongs. So if the connection drops, the server is able of detecting this condition through the ping/pong, but the client is let hung... until it tries to send something and the underlying TCP mechanism detect that the other side is not ACKnowledging its packets.
Therefore, if the client application expects to be "listening" most of the time, it may be convenient to implement a keep alive system that works "both ways" as Stephen Clearly explains in the link you posted. But, this keep alive system would be part of your application layer, rather than part of the transport layer as ping/pongs.
For example you can have a message "{token:'whatever'}" that the server and client just echoes with a 5 seconds delay. The client should have a timer with a 10 seconds timeout that stops every time that messages is received and starts every time the message is echoed, if the timer triggers, the connection can be consider dropped.
Although browsers that implement the same RFC as uWSGI should detect reliably when the server closes the connection cleanly they won't detect when the connection is interrupted midway (half open connections)t. So from what i understand we should employ an extra mechanism like application level pings.
I have an EventMachine server sending TCP data down to a Mac client (via GCDAsyncSocket). It always works flawlessly for a while, but inevitably the server suddenly stops sending data on a connection-by-connection basis. The connection is still maintained, and the server still receives data from the client, but it doesn't go the other way.
When this happens, I've discovered via connection#get_outbound_data_size that the connection send buffer is filling up infinitely (via #send_data) and not being sent to the client.
Are there specific (and hopefully fixable) reasons why this might occur? The reactor keeps humming along, and other active connections to the server continue working fine (though they sometimes fall into buffer hell as well).
I see one reason at least: when the remote client no longer read data from its side of the TCP connection (with a recv() call or whatever).
Then, the scenario is: the receiving TCP buffer on the client side becomes full. And the OS can no longer accepts TCP pacquets from its peer, since it cannot store them queue them. As a consequence, the sending TCP buffer on the server side becomes full too as your application continue to send paquets on the socket! Soon your server is no longer able to write into the socket since the send() system call will :
blocks undefinitively. (waiting for buffer to empty enough for the new paquet)
ot returns with an EWOULDBLOCK error. (if you configured your socket as a non-blocking one)
I usually met that kind of use case in TEST environment when I put a breakpoint in my code on the client side.
There was a patch was applied to GCDAsyncSocket on March 23 that prevents the reads from stopping. Did this patch solve your problem?