/
relp.html
347 lines (347 loc) · 19.3 KB
/
relp.html
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN">
<html><head><title>RELP - The Reliable Event Logging Protocol (Specification)</title></head>
<body>
<h1>RELP - The Reliable Event Logging Protocol</h1>
<p>This is the specification for the reliable event logging
protocol, called "RELP".</p>
<p>Version: 0.0.1<br>
Date: 2008-03-19<br>
Author: <a href="http://www.gerhards.net/rainer">Rainer
Gerhards</a></p>
<p>Copyright (C) 2008 by Rainer Gerhards and Adiscon GmbH.
Released under the terms of the GNU FDL.</p>
<h2>DESCRIPTION OF THE RELP PROTOCOL</h2>
<p>Relp uses a client-server model with (mostly fixed roles. The
initiating part of the
connection is called the client, the listening part the server. In the
state
diagrams below, C stand for client and S for server.
</p>
<p>Relp employs a command-response model, that is the client
issues commands to
which the server responds. To facilitate full-duplex communication,
multiple
commands can be issued at the same time, thus multiple responses may be
outstanding at a given time. The server may reply in any order. To
conserve
ressources, the number of outstanding commands is limited by a window.
Each
command is assigned a (relative) unique, monotonically increasing ID
(called the
"transaction number" or txnr for short. Each
response must include that ID. Responses must be sent by the server in
the exact same order as commands where received.</p>
<p>There is one potential issue with this communications model:
In it, the server is able to respond to the client only in form or a
response package. It is not capable of issuing a command itself. During
normal operations, this does not pose any problem. However, if the
server is shut down, there is no way for it to tell the client of its
shutdown intension, especially if the client does not send any data at
that moment. The same situation arises when the server timeouts a
sessions where the client has not sent any data for an extended period
of time (this probably is a good thing to conserve server ressources).
Of course, an obvious answer to that need is that the server
simply closes the connection after having sent all responses (or all
that it is still capable of). The client will notice the closed
connection at latest when it tries to send new commands. However, it
may be worth considering an alternate mode where the server can issue
commands to the client. One potential way to do this would be to
reserve a specific txnr (0, for example) for this use. That txnr is
only to be used by the server and only to send "out-of-band" commands.
These commands must not be acknowledged by the client, because that
would complicate the protocol more than useful for this case. So
out-of-band-commands have a "hint" status - they hint the client about
something, but if they do not arrive at the client, nothing fatal
should happen. The ultimate session closure indication would still be
the tcp session close. [Please see the "High Latency Environment" use
case below for some implications this hinting process may have].</p>
<p>For the time being, it has been decided that such out of band
hints be not supported.</p>
<p>A command and its response is called a relp transaction. A
response must always carry the same txnr as the original command.
</p>
<p>If something goes really wrong, both the client and the server
may terminate
the TCP connection at any time. Any outstanding commands are considered
to
have been unsuccessful in this case.</p>
<h3>ERROR HANDLING</h3>
<p>
If something goes wrong (e.g. invalid framing), the peer that detects
the problem
closes the underlying (TCP) connection. It may send an error/abort
hint, but
MUST NOT wait for its response. The reasoning for this is that today's
Internet is
not a friendly place. Trying to gracefully resolve communications
problems may
lead to an attack vector. On the other hand, if the problem was of
temporary nature,
it may be cleared up by the connection reset and a new connect. So
aborting errored
connections is considered both secure as well as reasonable efficient.</p>
<p>Note: the ability to send abort/error messages depends on the
decision of out-of-band-hint-commands (see above).
</p>
<h3>SENDING MESSAGES</h3>
Because it is so important, I'd like to point it specifically out:
sending a
message is "just" another RELP command. The reply to that command is
the ACK/NAK
for the message. So every message *is* acknowledged. RELP options
indicate how
"deep" this acknowledge is (once we have implemented that), in the most
extreme
case a RELP client may ask a RELP server to ack only after the message
has been
completely acted upon (e.g. successfully written to database) - but
this is far
away in the future. For now, keep in mind that message loss will always
be detected
because we have app-level acknowledgement.
<h3>RELP FRAME</h3>
All relp transaction are carried out over a consistent framing.
<pre>RELP-FRAME = HEADER DATA TRAILER<br>DATA = [SP 1*OCTET] ; command-defined data, if DATALEN is 0, no data is present<br>HEADER = TXNR SP COMMAND SP DATALEN<br>TXNR = NUMBER ; relp transaction number, monotonically increases, starts at 1<br>DATALEN = NUMBER<br>#old:COMMAND = "open" / "syslog" / "close" / "rsp" / "abort" ; max length = 32<br>COMMAND = 1*32ALPHA<br>TRAILER = LF ; to detect framing errors and enhance human readibility<br>ALPHA = letter ; ('a'..'z', 'A'..'Z')<br>NUMBER = 1*9DIGIT<br>DIGIT = %d48-57<br>LF = %d10<br>SP = %d32<br><br>RSP DATA CONTENT:<br>RSP-HEADER = TXNR SP RSP-CODE [SP HUMANMSG] LF [CMDDATA]<br>RSP-CODE = 200 / 500 ; 200 is ok, all the rest currently erros<br>HUAMANMSG = *OCTET ; a human-readble message without LF in it<br>CMDDATA = *OCTET ; semantics depend on original command<br>TXNR is as in the relp frame, it is the TXNR of the frame being responded to.<br></pre>
<p>
DATALEN is the number of octets in DATA (so the frame length excluding
the length
of HEADER and TRAILER). Please note that the theoretical maximum
(999,999,999 octets)
is in (almost?) all cases unsuitable for actual message transfer. Thus,
the actual
maximum data length is negotiated during session setup. In relp version
1, it is
always 128K. In later versions, it shall be operator-configurable.
</p>
<p>TXNR 0 is reserved for hint commands. A RELP
connection start with TXNR 1. Note that TXNR montonically increases but
is latched at 999,999,999. After that TXNR, the next TXNR is
1. </p>
<h3>COMMAND SEMANTICS</h3>
Command can be either regular commands or hints. Regular commands are
responded to by a rsp package. They are transmitted on txnrs of 1 or
above. Hints are commands that are not being responded. They are always
transmitted with a txnr of zero. Regular commands can only be issued by
the client. Hints can be issued both by the client and the server.<br>
<h4>"Command" "rsp"</h4>
This is not a real command but a response to a client-issued command.
The TXNR MUST match the client's
command
TXNR. The data part contains RSP-HEADER as defined above. It is a
response code,
optionally followed by a space and additional data (depending on the
client's command). Return state values are: 200 - OK, 500 - error.<br>
<br>
The ABNF for a "rsp"'s data portion is as follows:<br>
<pre>RSPDATA = STATUS SP HUMAN-TEXT [LF DATA]<br>STATUS = 3DIGIT<br>HUMAN-TEXT = *80ALPHA<br>DATA = 1*OCTET<br></pre>
<h4>Command "open"</h4>
Opens a connection to the server. Must include offers inside the
data, at a minimum the "relp_version" offer. Offers
provide information about services supported by the client.
<p>When the server receives an open, it parses the offers, checks
what it itself supports
and provides those offers that it accepts in the "rsp". The server may
send a "rsp" with an empty offer part in case it doesn't like any of
the offers. To establish a session, a "relp_version" offer MUST be
included in the response. If it is not, the client MUST close the
connection. </p>
<p>When the client receives the "rsp", it checks the servers
offers and finally selects
those that should be used during the session. Please note that this
doesn't imply the
client selects e.g. security strength. To require a specific security
strength, the
server must be configured to offer only those options back to the
client that it is
happy to accept. So the client can only select from those. As such,
even though the
client makes the final feature selection, the server is dictating what
needs to be used.
</p>
<p>Note that the connection is only ready AFTER the client has
received the "open" response.</p>
<h4>Command "close"</h4>
Closes a connection to the server. Once the client has sent a "close"
command, it must not transmit any other command over the session. When
the server's rsp answer is received, the client must close the
connection.<br><h4>Hint Command "serverclose"</h4>This
hint SHOULD be sent by the server before it closes a RELP connection.
It advised the client that the server has closed the connection. The
client SHOULD use this knowledge to close the connection on the
client-side, too. Please note that this is a hint, and as such
transferred with a TXNR of 0 and without any response from the client.
Note that the client can not ask the server to defer closing of the
connection. A client MUST correctly process a server-closed connection
even if the server does not send a "serverclose" hint. In thus case, it
must rely on the transport layer connection status. The primary utility
of "serverclose" is to enable the client to quickly detect a
server-closed connection, clean up ressources and go into termination
(or recovery code).<h4>Command "starttls"</h4>
Does the same thing as ESMTP STARTTLS: switches the connection
to TLS mode.
<h4>Command "syslog"</h4>
This command is used to transmit a syslog message, which (in syslog
message format) is contained within the commands data portion.
<h3>OFFERS</h3>
During session setup, "offers" are exchange between client and server.
An "offer" describes
a specific feature or operation mode. Always present must be the
"relp_version" offer which
tells the other side which version of relp is in use.
<p>ABNF for offer strings
</p>
<pre>OFFER = LF FEATURENAME ["=" VALUE *("," VALUE)]<br>FEATURENAME = *32OCTET<br>VALUE = *255OCTET<br><br>Currently defined values:<br>FEATURENAME VALUE<br>relp_version 1 (this specification)<br></pre>
<h4>relp_version</h4>
<p>This offer describes the protocol version. It MUST be provided
in the
open command. It has one parameter, the numerical version id. The
intial version is 1. There also exists an experimental version 0, which
shall not be used in practice.
</p>
<h4>commands</h4>
<p>This offer describes which commands are supported by the
remote peer. Command that must be specified to drive the RELP protocol
itself (e.g. "open", "close", "rsp") MUST NOT be specified. Optional
commands must be specified. There may be multiple commands specified.</p>
<p>Sample: commands=syslog,eventlog (eventlog is a hypothetical,
not yet known command).</p>
<p>Please note that new commands may be specified outside of the
relp specification. Also note that if the server response does not
include the command the client is interested in, this command MUST NOT
be used. The client may then either resort to another, less featured
supported command or may terminate the relp connection.
</p>
<h4>relp_software</h4>
<p>This offer describes the software (library) that is processing
the relp
protocol. It is an optional offer. The idea is that when each peer
knows which software on the other peer is handling the protocol, it may
be able to work-around some known software issues. The "relp_software"
has two parameters, the first one being the name of the software as it
calls itself, the second one the version as it calls itself and the
third one being an URL where information about this software may be
found. It is suggested that all three parameters are given, but only
the first one MUST actually be provided.
</p>
<h3>STATE DIAGRAMS</h3>
<p>... detailling some communications scenarios:
</p>
<pre>Session Startup:<br>C S<br>cmd: "open", data: offer -----> (selects supported offers)<br>(selects offers to use) <----- cmd: "rsp", data "accepted offers"<br><br> ... transmission channel is ready to use ....<br><br>Message Transmission<br>C S<br>cmd: "syslog", data: syslogmsg -----> (processes message)<br>(indicates syslog as processed) <----- cmd: "rsp", data OK/Error<br><br>Session Termination<br>C S<br>cmd: "close", data: none? -----> (processes termination request)<br>(terminates session) <----- cmd: "rsp", data OK/Error<br> (terminates session)<br></pre>
<h3>SECURITY CONSIDERATIONS</h3>
<h4>Window Size</h4>
A very large window can be abused for denial of service attacks.
<h4>Maximum DATALEN</h4>
A too-large DATALEN can be absued for denial of service attacks.
<h2>Use Cases</h2>
These uses cases are not really a part of the
specification. I have added them to provide some insight into
implementation specifics. These use cases may be removed (or moved to
another part of the doc set) some time in the future. In the mean time,
the serve a valuable purpose when thinking about protocol features.
<h3>Server Session Closure with a Hibernated Client</h3>
I define this use case in support of <a href="http://www.rsyslog.com">rsyslog</a> and other
possible clients which try to reduce ressource consumption.<br>
<br>
Rsyslog utilizes highly threaded design, among others to take advantage
of multicore-machines. On
the other hand, it conserves resources by only running as many worker
threads as actually needed. So if there is no traffic, it may even be
running no workers at all. The RELP client will be implemented as an
output
plugin and will be executed on a worker thread. So in low-traffic
cases, the RELP client kind of hibernates, maybe even for hours (if no
new message is scheduled for transmission).
<p>In this use case, I examine a typical scenario in a
low-message volume case. Here, the client is asked to transmit a
message every now and than, at a pace of at most one message every few
seconds. It may also happen that no message is to be sent for several
minutes or even hours. Such a scenario is not as extreme as it sounds -
it may, for example, to be used for error notifications, which
hopefully happen very infrequently.</p>
<p>In this scenario, the client sends a command, but will usually
not receive an immediate response, because client processing is already
terminated when the server sends the response packet. Thus, the tcp
stacks receive buffer buffers the server's response until the client
has another message to send. Then, the client first checks for any
outstanding responses, processes them and then send the new message. It
is expected that in this scenario, each client invokation will process
the past response and send a new packet. The response to that packet
will not be processed by the invocation that sent the command but by
the next one. So we always have one outstanding response inside the OS
buffers. Now consider the server is shutting down the connection due to
timeout or due to the fact that it needs to shutdown itself (maybe even
on an urgent case, like power failure - the point is it can not wait).
On next client invokation, the client finds an outstanding response and
processes it. Then, it tries to send a new command. That will fail,
because the connection has been terminated. The usual session recovery
logic is used to restablish the connection when the server is back
online. The command in question is transferred after session
re-establishment.</p>
<h3>High Latency Environment</h3>
<p>We have a high latency environment that is otherwise quite
reliable
(satellite link). We have a high traffic load. To cover this, a large<br>
transmission
window has been selected (let's say 1000 messages). Now the client
sends a burst of messages but then has nothing to do and
falls
asleep. Then, the server starts sending acks. These acks are not taken
off the wire by the client (as it is inactive - rsyslog case).
Eventually, the tcp window fills. Thus the server can no longer send
acks. This does not immediately pose any problems, as the server adds
the ack frames to its internal send queue.</p>
<p>If the client comes
out of hibernation, it receives all previous sent frames, freeing the
server's tcp send window. Thus, the server can continue to send data
from its send queue (and drain it). The client, having received the
server's acks, will not stall because its own RELP window has been
cleared by the acks.</p>
<p>The situation is more complicated if
the server intends to shut down while the client is in hibernation. To
do so, the server usually sends a "adviseclose" command followed by a
"close" command. However, neither of these commands can be sent because
the client does not take them off the wire. The close operation is now
stalled. The only option left to the server is an unconditional
termination of the session. While everything that has already been sent
to the client will be picked up by it when it comes out of hibernation,
anything left in the server's send queue will be lost in this case. As
such, acks are potentially lost. This will lead to message duplication
as the client assumes that the frames unacked at time of force-close
where not processed (there is no other safe assumption). Consequently,
the client will re-send these frames in the next session.</p>
<p>A cure
to this situation is to have the client concurrently listen to server
requests, e.g. by running a receiver on a separate thread. The RELP
protocol does not demand this behaviour. However, it can be used to
solve the above server stall. Let's assume this is being done. Do we
now have a sufficient guarantee that the server does not stall? Under
normal conditions, it is a safe assumption that the client will be able
to receive all frames sent by the server. So a buildup of server queues
should, if at all, happen only for a short instant. However, if
something goes wrong on the transmission line (especially in
high-latency cases), there may be a somewhat extended period of time in
which the server can not send acks (but only in a magnitude of a few
seconds at most). If the server intends to shutdown during such a
period, a short timeout may enable it to avoid a fore-shutdown.
However, there are still cases thinkable where a force shutdown may be
required. These are deemed to be highly unlikely.</p>
<p>It is the
protocol implementer's choice if the slight less chance of a server
force-shutdown justifies the addition of a background thread to a
program that otherwise doesn't need one. From a protocol point of view,
a force-shutdown is a valid operation. Also, while it causes potential
message duplication, it can not cause message loss.</p>
<h2>Feedback and Discussion</h2>
<p>Feedback on this document and the RELP protocol is
appreciated. The <a href="http://lists.adiscon.net/mailman/listinfo/relp">RELP
mailing list</a> is probably a good place to provide it.
Questions are also happily accepted.</p>
<h2>Copyright</h2>
<p>Copyright (C) 2008, 2009 Rainer Gerhards and Adiscon.
<p>
Permission is granted to copy, distribute and/or modify this document
under the terms of the GNU Free Documentation License, Version 1.3
or any later version published by the Free Software Foundation;
with no Invariant Sections, no Front-Cover Texts, and no Back-Cover Texts.
A copy of the license can be obtained from <a href="http://www.gnu.org/copyleft/fdl.html">http://www.gnu.org/copyleft/fdl.html</a> and is also included in the <a href="gfdl.html">license file</a> (if there is any difference in those two, the one contained in the license file is the one that this work is licensed under).
</body></html>