blob: c67077cbeb800a1265f1e2ba801fb293fa0d43f8 [file] [log] [blame]
Andy Grover0c5f9b82009-02-24 15:30:38 +00001
2Overview
3========
4
5This readme tries to provide some background on the hows and whys of RDS,
6and will hopefully help you find your way around the code.
7
8In addition, please see this email about RDS origins:
9http://oss.oracle.com/pipermail/rds-devel/2007-November/000228.html
10
11RDS Architecture
12================
13
14RDS provides reliable, ordered datagram delivery by using a single
15reliable connection between any two nodes in the cluster. This allows
16applications to use a single socket to talk to any other process in the
17cluster - so in a cluster with N processes you need N sockets, in contrast
18to N*N if you use a connection-oriented socket transport like TCP.
19
20RDS is not Infiniband-specific; it was designed to support different
21transports. The current implementation used to support RDS over TCP as well
22as IB. Work is in progress to support RDS over iWARP, and using DCE to
23guarantee no dropped packets on Ethernet, it may be possible to use RDS over
24UDP in the future.
25
26The high-level semantics of RDS from the application's point of view are
27
28 * Addressing
29 RDS uses IPv4 addresses and 16bit port numbers to identify
30 the end point of a connection. All socket operations that involve
31 passing addresses between kernel and user space generally
32 use a struct sockaddr_in.
33
34 The fact that IPv4 addresses are used does not mean the underlying
35 transport has to be IP-based. In fact, RDS over IB uses a
36 reliable IB connection; the IP address is used exclusively to
37 locate the remote node's GID (by ARPing for the given IP).
38
39 The port space is entirely independent of UDP, TCP or any other
40 protocol.
41
42 * Socket interface
43 RDS sockets work *mostly* as you would expect from a BSD
44 socket. The next section will cover the details. At any rate,
45 all I/O is performed through the standard BSD socket API.
46 Some additions like zerocopy support are implemented through
47 control messages, while other extensions use the getsockopt/
48 setsockopt calls.
49
50 Sockets must be bound before you can send or receive data.
51 This is needed because binding also selects a transport and
52 attaches it to the socket. Once bound, the transport assignment
53 does not change. RDS will tolerate IPs moving around (eg in
54 a active-active HA scenario), but only as long as the address
55 doesn't move to a different transport.
56
57 * sysctls
58 RDS supports a number of sysctls in /proc/sys/net/rds
59
60
61Socket Interface
62================
63
64 AF_RDS, PF_RDS, SOL_RDS
65 These constants haven't been assigned yet, because RDS isn't in
66 mainline yet. Currently, the kernel module assigns some constant
67 and publishes it to user space through two sysctl files
68 /proc/sys/net/rds/pf_rds
69 /proc/sys/net/rds/sol_rds
70
71 fd = socket(PF_RDS, SOCK_SEQPACKET, 0);
72 This creates a new, unbound RDS socket.
73
74 setsockopt(SOL_SOCKET): send and receive buffer size
75 RDS honors the send and receive buffer size socket options.
76 You are not allowed to queue more than SO_SNDSIZE bytes to
77 a socket. A message is queued when sendmsg is called, and
78 it leaves the queue when the remote system acknowledges
79 its arrival.
80
81 The SO_RCVSIZE option controls the maximum receive queue length.
82 This is a soft limit rather than a hard limit - RDS will
83 continue to accept and queue incoming messages, even if that
84 takes the queue length over the limit. However, it will also
85 mark the port as "congested" and send a congestion update to
86 the source node. The source node is supposed to throttle any
87 processes sending to this congested port.
88
89 bind(fd, &sockaddr_in, ...)
90 This binds the socket to a local IP address and port, and a
91 transport.
92
93 sendmsg(fd, ...)
94 Sends a message to the indicated recipient. The kernel will
95 transparently establish the underlying reliable connection
96 if it isn't up yet.
97
98 An attempt to send a message that exceeds SO_SNDSIZE will
99 return with -EMSGSIZE
100
101 An attempt to send a message that would take the total number
102 of queued bytes over the SO_SNDSIZE threshold will return
103 EAGAIN.
104
105 An attempt to send a message to a destination that is marked
106 as "congested" will return ENOBUFS.
107
108 recvmsg(fd, ...)
109 Receives a message that was queued to this socket. The sockets
110 recv queue accounting is adjusted, and if the queue length
111 drops below SO_SNDSIZE, the port is marked uncongested, and
112 a congestion update is sent to all peers.
113
114 Applications can ask the RDS kernel module to receive
115 notifications via control messages (for instance, there is a
116 notification when a congestion update arrived, or when a RDMA
117 operation completes). These notifications are received through
118 the msg.msg_control buffer of struct msghdr. The format of the
119 messages is described in manpages.
120
121 poll(fd)
122 RDS supports the poll interface to allow the application
123 to implement async I/O.
124
125 POLLIN handling is pretty straightforward. When there's an
126 incoming message queued to the socket, or a pending notification,
127 we signal POLLIN.
128
129 POLLOUT is a little harder. Since you can essentially send
130 to any destination, RDS will always signal POLLOUT as long as
131 there's room on the send queue (ie the number of bytes queued
132 is less than the sendbuf size).
133
134 However, the kernel will refuse to accept messages to
135 a destination marked congested - in this case you will loop
136 forever if you rely on poll to tell you what to do.
137 This isn't a trivial problem, but applications can deal with
138 this - by using congestion notifications, and by checking for
139 ENOBUFS errors returned by sendmsg.
140
141 setsockopt(SOL_RDS, RDS_CANCEL_SENT_TO, &sockaddr_in)
142 This allows the application to discard all messages queued to a
143 specific destination on this particular socket.
144
145 This allows the application to cancel outstanding messages if
146 it detects a timeout. For instance, if it tried to send a message,
147 and the remote host is unreachable, RDS will keep trying forever.
148 The application may decide it's not worth it, and cancel the
149 operation. In this case, it would use RDS_CANCEL_SENT_TO to
150 nuke any pending messages.
151
152
153RDMA for RDS
154============
155
156 see rds-rdma(7) manpage (available in rds-tools)
157
158
159Congestion Notifications
160========================
161
162 see rds(7) manpage
163
164
165RDS Protocol
166============
167
168 Message header
169
170 The message header is a 'struct rds_header' (see rds.h):
171 Fields:
172 h_sequence:
173 per-packet sequence number
174 h_ack:
175 piggybacked acknowledgment of last packet received
176 h_len:
177 length of data, not including header
178 h_sport:
179 source port
180 h_dport:
181 destination port
182 h_flags:
183 CONG_BITMAP - this is a congestion update bitmap
184 ACK_REQUIRED - receiver must ack this packet
185 RETRANSMITTED - packet has previously been sent
186 h_credit:
187 indicate to other end of connection that
188 it has more credits available (i.e. there is
189 more send room)
190 h_padding[4]:
191 unused, for future use
192 h_csum:
193 header checksum
194 h_exthdr:
195 optional data can be passed here. This is currently used for
196 passing RDMA-related information.
197
198 ACK and retransmit handling
199
200 One might think that with reliable IB connections you wouldn't need
201 to ack messages that have been received. The problem is that IB
202 hardware generates an ack message before it has DMAed the message
203 into memory. This creates a potential message loss if the HCA is
204 disabled for any reason between when it sends the ack and before
205 the message is DMAed and processed. This is only a potential issue
206 if another HCA is available for fail-over.
207
208 Sending an ack immediately would allow the sender to free the sent
209 message from their send queue quickly, but could cause excessive
210 traffic to be used for acks. RDS piggybacks acks on sent data
211 packets. Ack-only packets are reduced by only allowing one to be
212 in flight at a time, and by the sender only asking for acks when
213 its send buffers start to fill up. All retransmissions are also
214 acked.
215
216 Flow Control
217
218 RDS's IB transport uses a credit-based mechanism to verify that
219 there is space in the peer's receive buffers for more data. This
220 eliminates the need for hardware retries on the connection.
221
222 Congestion
223
224 Messages waiting in the receive queue on the receiving socket
225 are accounted against the sockets SO_RCVBUF option value. Only
226 the payload bytes in the message are accounted for. If the
227 number of bytes queued equals or exceeds rcvbuf then the socket
228 is congested. All sends attempted to this socket's address
229 should return block or return -EWOULDBLOCK.
230
231 Applications are expected to be reasonably tuned such that this
232 situation very rarely occurs. An application encountering this
233 "back-pressure" is considered a bug.
234
235 This is implemented by having each node maintain bitmaps which
236 indicate which ports on bound addresses are congested. As the
237 bitmap changes it is sent through all the connections which
238 terminate in the local address of the bitmap which changed.
239
240 The bitmaps are allocated as connections are brought up. This
241 avoids allocation in the interrupt handling path which queues
242 sages on sockets. The dense bitmaps let transports send the
243 entire bitmap on any bitmap change reasonably efficiently. This
244 is much easier to implement than some finer-grained
245 communication of per-port congestion. The sender does a very
246 inexpensive bit test to test if the port it's about to send to
247 is congested or not.
248
249
250RDS Transport Layer
251==================
252
253 As mentioned above, RDS is not IB-specific. Its code is divided
254 into a general RDS layer and a transport layer.
255
256 The general layer handles the socket API, congestion handling,
257 loopback, stats, usermem pinning, and the connection state machine.
258
259 The transport layer handles the details of the transport. The IB
260 transport, for example, handles all the queue pairs, work requests,
261 CM event handlers, and other Infiniband details.
262
263
264RDS Kernel Structures
265=====================
266
267 struct rds_message
268 aka possibly "rds_outgoing", the generic RDS layer copies data to
269 be sent and sets header fields as needed, based on the socket API.
270 This is then queued for the individual connection and sent by the
271 connection's transport.
272 struct rds_incoming
273 a generic struct referring to incoming data that can be handed from
274 the transport to the general code and queued by the general code
275 while the socket is awoken. It is then passed back to the transport
276 code to handle the actual copy-to-user.
277 struct rds_socket
278 per-socket information
279 struct rds_connection
280 per-connection information
281 struct rds_transport
282 pointers to transport-specific functions
283 struct rds_statistics
284 non-transport-specific statistics
285 struct rds_cong_map
286 wraps the raw congestion bitmap, contains rbnode, waitq, etc.
287
288Connection management
289=====================
290
291 Connections may be in UP, DOWN, CONNECTING, DISCONNECTING, and
292 ERROR states.
293
294 The first time an attempt is made by an RDS socket to send data to
295 a node, a connection is allocated and connected. That connection is
296 then maintained forever -- if there are transport errors, the
297 connection will be dropped and re-established.
298
299 Dropping a connection while packets are queued will cause queued or
300 partially-sent datagrams to be retransmitted when the connection is
301 re-established.
302
303
304The send path
305=============
306
307 rds_sendmsg()
308 struct rds_message built from incoming data
309 CMSGs parsed (e.g. RDMA ops)
310 transport connection alloced and connected if not already
311 rds_message placed on send queue
312 send worker awoken
313 rds_send_worker()
314 calls rds_send_xmit() until queue is empty
315 rds_send_xmit()
316 transmits congestion map if one is pending
317 may set ACK_REQUIRED
318 calls transport to send either non-RDMA or RDMA message
319 (RDMA ops never retransmitted)
320 rds_ib_xmit()
321 allocs work requests from send ring
322 adds any new send credits available to peer (h_credits)
323 maps the rds_message's sg list
324 piggybacks ack
325 populates work requests
326 post send to connection's queue pair
327
328The recv path
329=============
330
331 rds_ib_recv_cq_comp_handler()
332 looks at write completions
333 unmaps recv buffer from device
334 no errors, call rds_ib_process_recv()
335 refill recv ring
336 rds_ib_process_recv()
337 validate header checksum
338 copy header to rds_ib_incoming struct if start of a new datagram
339 add to ibinc's fraglist
340 if competed datagram:
341 update cong map if datagram was cong update
342 call rds_recv_incoming() otherwise
343 note if ack is required
344 rds_recv_incoming()
345 drop duplicate packets
346 respond to pings
347 find the sock associated with this datagram
348 add to sock queue
349 wake up sock
350 do some congestion calculations
351 rds_recvmsg
352 copy data into user iovec
353 handle CMSGs
354 return to application
355
356