| Re: Issue 30: Inconsistent state tracking on AC prior toDTLSEstablishment | <– Date –> <– Thread –> |
From: Charles Clancy (clancy cs.umd.edu)
|
|
| Date: Sat, 15 Dec 2007 14:53:03 -0800 (PST) | |
I'm trying to compare Pat and Scott's suggested changes. Scott is
suggesting removing (f) and adding a new transition from discovery to
idle (at least those would be necessary for your proposed text to make
sense). Pat is also suggesting doing that, but is also suggesting some
additional changes that deal with further separation of a logical global
state machine in the AC and per-WTP state machines (by adding the "Dead"
state and a rearranging a few more state transitions).
I think there are problems in the state machine if we only patch the idle/discovery transitions, as suggested by Scott. In particular, there's no way to throttle failed DTLS connections when no discovery is performed (the only error exit from DTLS Setup is to Idle -- we need a way to get from DTLS Setup to sulking to prevent repeated crypto failures from causing Idle->DTLSSetup->Idle loops).
I'm not seeing Scott's issues with the new state machine (i.e. the "new problems and confusion"). The discovery-idle-sulking changes seem logically identical between the two proposals. If there are text issues, I think those can be sorted out.
I think Scott's suggestions could work, but would need some additional transitions (i.e. DTLS Setup to Sulking for the WTP, among others) to make everything work right.
Whether Pat's changes are too drastic -- I guess that's for the WG to decide. I do think Pat has a complete set of changes that solve all the issues raised in issue 30.
Pat Calhoun (pacalhou) wrote:
I think there are problems in the state machine if we only patch the idle/discovery transitions, as suggested by Scott. In particular, there's no way to throttle failed DTLS connections when no discovery is performed (the only error exit from DTLS Setup is to Idle -- we need a way to get from DTLS Setup to sulking to prevent repeated crypto failures from causing Idle->DTLSSetup->Idle loops).
I'm not seeing Scott's issues with the new state machine (i.e. the "new problems and confusion"). The discovery-idle-sulking changes seem logically identical between the two proposals. If there are text issues, I think those can be sorted out.
I think Scott's suggestions could work, but would need some additional transitions (i.e. DTLS Setup to Sulking for the WTP, among others) to make everything work right.
Whether Pat's changes are too drastic -- I guess that's for the WG to decide. I do think Pat has a complete set of changes that solve all the issues raised in issue 30.
-- t. charles clancy, ph.d. eng.umd.edu/~tcc electrical & computer engineering, university of maryland
Pat Calhoun (pacalhou) wrote:
The reason why I like the new state machine is that it is very clear on how the AC deals with the fact that it keeps multiple copies of state machine context. The previous version really didn't make that clear at all. As an implementer, I know that I just have to follow the guidelines, which clearly state when a copy is made, and managed independently, and when it is deleted.
I recognize this provides some implementation guidance, and some folks don't like that. But it certainly is clear.
If the issue has to do with the Sulking state, then we can address that. I would, however, be interested in Charles' take on that issue.
PatC
-----Original Message-----
From: Scott Kelly [mailto:skelly [at] arubanetworks.com] Sent: Wednesday, December 12, 2007 9:17 AM
To: Pat Calhoun (pacalhou); capwap
Subject: RE: [Capwap] Issue 30: Inconsistent state tracking on AC prior
toDTLSEstablishment
We've had a private discussion on this topic off-list, but have not been able to come to agreement, so I'll make my comments here. I don't think we should (or need to) introduce these dramatic changes to the document just prior to last call.
The "problem" we are trying to resolve resulted when someone added
language saying the AC MAY maintain discovery state. This change was
apparently introduced last January. There is no record of working group
or mailing list discussion -- the change just appeared in a document
update.
On Oct 12, Pat posted a comment saying the state machine language surrounding discovery state is inconsistent. I agree. It was made inconsistent by the introduction of discovery state to the AC.
In the issue tracker, there is a comment from Charles which must have been copied from a private email, as I cannot find it in the list archives. Here is that comment:
----------------------------------------------
It seems to me that to be in any state at all assumes state is being
maintained. Describing a transition from one state to another with
behavior caveatted with "if state is being maintained" doesn't make any
sense to me. If we have no state, then we have no state transitions.
IIRC, don't we allow connections that don't do discovery first (it's been a while, correct me if I'm wrong)? In which case, DTLS on the AC should always be listening, and transition "f" can be simply a WTP transition. The AC
transition would be back to Idle rather than DTLS setup. You could run discovery as its own subgraph of the AC state machine. If you wanted to share state information between the two, you could have an API to do that. ------------------------------------------------
I agree with what Charles says above.
The changes Pat is proposing do not fix the problem - they actually introduce new problems and confusion. Since we have agreed that the AC SHOULD NOT maintain discovery state, the discovery-idle-sulking transitions make no sense in the context of the AC. However, fixing this does not require separate state machines, or dramatic changes to the existing state machine.
In draft-08, section 2.3, just below Figure 3, it says "The CAPWAP
protocol state machine, depicted above, is used by both the AC and the
WTP. In cases where states are not shared (i.e. not implemented in one
or the other of the AC or WTP), this is explicitly called out in the
transition descriptions below."
This implies that we just need to make sure the state machine language around AC discovery processing is clear. We can do this by adopting three simple changes recommended in Pat's text below (with minor edits):
For Idle-to-Discovery it should say:
AC: This state transition is executed by the AC when a Discovery Request message is received. The AC SHOULD respond with a Discovery Response message (see Section 5.2). The AC SHOULD NOT maintain WTP state at this point (see Section 12 for more information).
For Discovery to Idle it should say:
AC: This state transition is executed by the AC after processing a Discovery Request.
For Discovery to Sulking it says: AC: This is an invalid state transition for the AC.
With these simple changes, the problem is solved.
--Scott
-----Original Message-----
From: Pat Calhoun (pacalhou) [mailto:pcalhoun [at] cisco.com]
Sent: Tuesday, December 04, 2007 3:28 PM
To: capwap
Subject: Re: [Capwap] Issue 30: Inconsistent state tracking on AC prior toDTLSEstablishment
All,
I wanted to send the new state machine in order to get as many eyes on
this one as possible. I have done what I believe is a very thorough review, but this is a complex state machine. I will also include all of the state machine text in order to make it simpler to review.
Note that
other than the teardown, no changes were made from the join/config states on (meaning, only the lower part of the state machine figure was modified).
Since there is significant text here, I want to provide a little bit of information on how this now works:
- The WTP behaves pretty much the same as before, meaning that it uses
a single state context. When a connection attempt fails, the state machine goes back to the Idle state, unless some counters have hit a threshold, in which case the state transitions to the Sulking state. As before the WTP can do the Discovery phase, in which case either state
(5) or (c) is
used. - The AC now defines two separate threads; Listener and Service. The Listener is the main thread that handles the discovery messages. Once a connection request is received, it spawns a Service thread to deal with the WTP. You will notice that the new state machine uses numerals
when it applies either only to the WTP, or to the Listener thread. All state transitions that use alphabetic, or symbols, is for the Service thread
or the WTP. The biggest difference in how the state machines work is that on the AC, when a connection is being established, and fails, the
Service thread's state moves back to the DTLS Setup state, not Idle.
When the error thresholds are hit, the state moves to Sulking - the quiet period. You will also notice a new Dead state, which is ONLY valid for the Service thread, which is where the thread is terminated,
and resources are freed.
Comments welcomed,
PatC
<text> 2.3. CAPWAP State Machine Definition
The following state diagram represents the lifecycle of a WTP-AC
session. Use of DTLS by the CAPWAP protocol results in the
juxtaposition of two nominally separate yet tightly bound state
machines. The DTLS and CAPWAP state machines are coupled through an
API consisting of commands (see Section 2.3.2.1) and notifications
(see Section 2.3.2.2). Certain transitions in the DTLS state machine
are triggered by commands from the CAPWAP state machine, while
certain transitions in the CAPWAP state machine are triggered by
notifications from the DTLS state machine.
/-------------------------------------\ | /-------------------------\| | w| || | x+----------+ y +------------+ || | | Run |-->| Reset |-\|| | +----------+ +------------+ ||| u| V ^ ^ ^ z||| +------------+--------/ | | ||| | Data Check | /-------/ | ||| +------------+<-------\ | | ||| | | | ||| /------------------+--------\ | ||| m| t| o| q v r| ||| +--------+ +-----------+ +--------------+||| | Join |---->| Configure | | Image Data |||| +--------+ n +-----------+ +--------------+||| ^ |l p| s| ||| | | \-------------------\ | ||| | \--------------------------------------\| | ||| \------------------------\ || | ||| /--------------<----------------+---------------\ || | ||| | /------------<----------------+-------------\ | || | ||| | | i |k 8| | vv v vvv | | d+----------------+<--+--------------+ +-----------+ /-|-|---| DTLS Setup | | DTLS Connect |-->| DTLS TD | | | | +----------------+e +--------------+ 7 +-----------+ | | | 6| ^ ^ |f ^ j| ^ |~ v v v | | | | | | | | | | | | | | \-------\ | /----+------/ | | | | | | | | | | \---\ | | | | v c| 1 |5 2 v |g |h v v | | \->+------+-->+------+ +-----------+ +--------+ | | | Idle | | Disc | | Authorize | | Dead | | | +------+<--+------+ +-----------+ +--------+ | | ^ 0 |3 ^ | | | | |b | |9 |4 | | | \->+---------+<------/ | \--->| Sulking |----------------------------------/ +---------+ a
Figure 3: CAPWAP Integrated State Machine
The CAPWAP protocol state machine, depicted above, is used by both
the AC and the WTP. In cases where states are not shared (i.e. not
implemented in one or the other of the AC or WTP), this is explicitly
called out in the transition descriptions below. For every state
defined, only certain messages are permitted to be sent and received.
The CAPWAP control messages definitions specify the
state(s) in which
each message is valid.
Since the WTP only communicates with a single AC, it only has a
single instance of the CAPWAP state machine. The state machine works
differently on the AC since it communicates with many WTPs. The AC
uses the concept of two threads. Note that the term thread used here
does not necessarily imply that implementers must use threads, but it
is one possible way of implementing the AC's state machine.
Listener Thread - The AC's Listener thread handles the shared
services, which includes receiving and responding to Discovery
Requests. The Listener thread handles the common tasks, up to the
DTLS Setup state. The state machine transitions in the above
figure are represented by numerals.
Service Thread - The AC's Service thread handles the per WTP
states, and one such thread exists per WTP connection. This
thread starts during the DTLS Setup state, which is when the
DTLSListen command is invoked. When created, the Service thread
inherits a copy of the state machine context from the Listener
thread. When communication with the WTP is complete, the Service
thread is terminated. The state machine transitions in the above
figure are represented by alphabetic characters (including
symbols).
2.3.1. CAPWAP Protocol State Transitions
This section describes the various state transitions, and the events
that cause them. This section does not discuss interactions between
DTLS- and CAPWAP-specific states. Those interactions, and DTLS-
specific states and transitions, are discussed in Section 2.3.2.
Idle to Discovery (1): This transition occurs once device initialization is complete.
WTP: The WTP enters the Discovery state prior to transmitting the
first Discovery Request message (see Section 5.1). Upon
entering this state, the WTP sets the DiscoveryInterval timer
(see Section 4.7). The WTP resets the DiscoveryCount counter
to zero (0) (see Section 4.8). The WTP also clears all
information from ACs it may have received during a previous
Discovery phase.
AC: This state transition is executed by the AC's Listener thread, and occurs when a Discovery Request message is received. The AC SHOULD respond with a Discovery Response message (see Section 5.2). The AC SHOULD NOT maintain WTP state at this point (see Section 12 for more information).
Discovery to Discovery (2): In the Discovery state, the WTP determines which AC to connect to.
WTP: This transition occurs when the DiscoveryInterval timer
expires. If the WTP is configured with a list of ACs, it
transmits a Discovery Request message to every AC from which it
has not received a Discovery Response message. For every
transition to this event, the WTP increments the DiscoveryCount
counter. See Section 5.1 for more information on how the WTP
knows the ACs to which it should transmit the Discovery Request
messages. The WTP restarts the DiscoveryInterval timer
whenever it transmits Discovery Request messages.
AC: This is an invalid state transition for the AC.
Discovery to Idle (0): This transition occurs on the AC's Listener thread when the Discovery processing is complete.
WTP: This is an invalid state transition for the WTP.
AC: This state transition is executed by the AC's Listener thread
when it has transmitted the Discovery Response, in response to
a Discovery Request.
Discovery to Sulking (3): This transition occurs on a WTP when AC Discovery fails.
WTP: The WTP enters this state when the DiscoveryInterval timer
expires and the DiscoveryCount variable is equal to the
MaxDiscoveries variable (see Section 4.8). Upon entering this
state, the WTP MUST start the SilentInterval timer. While in
the Sulking state, all received CAPWAP protocol messages
received MUST be ignored.
AC: This is an invalid state transition for the AC.
Sulking to Idle (4): This transition occurs on a WTP when it must restart the discovery phase.
WTP: The WTP enters this state when the SilentInterval timer (see
Section 4.7) expires. The FailedDTLSSessionCount,
DiscoveryCount and FailedDTLSAuthFailCount counters are reset
to zero.
AC: This is an invalid state transition for the AC.
Sulking to Sulking (a): The Sulking state provides the silent period, minimizing the possibility for Denial of Service (DoS) attacks.
WTP: All packets received from the AC while in the sulking state
are ignored.
AC: All packets receive from the WTP on the AC's Service thread,
while in the sulking state, are ignored.
Sulking to Dead (b): This transition occurs on the AC when the quiet
period expires.
WTP: This is an invalid state transition for the WTP.
AC: The AC enters this state when the SilentInterval timer (see
Section 4.7) expires. The AC must clean up all resources
associated with the control plane DTLS session. The data plane
DTLS session is also shutdown, and all resources freed, if a
DTLS session was established for the data plane. Any timers
set for the current instance of the state machine are also
cleared. The AC's Service thread is terminated.
Idle to DTLS Setup (c): This transition occurs to establish a secure
DTLS session with the peer.
WTP: The WTP initiates this transition by invoking the DTLSStart
command, which starts the DTLS session establishment with the
chosen AC. When the discovery phase is bypassed, it is assumed
the WTP has locally configured ACs.
AC: The AC initiates this transition by invoking the DTLSListen
command (see Section 2.3.2.1), which informs the DTLS stack
that it is willing to listen for an incoming session. The AC's
Listener thread forks an instance of the Service thread, along
with a copy of the state context. If the AC had maintained WTP
state information during the Discovery exchange, or through
some other means that may include static configuration of WTPs,
the AC MAY provide optional qualifiers in the DTLSListen
command to only accept session requests a specific WTP. Note
that the AC SHOULD NOT maintain state information based on an
unsecured Discovery Request message, as this can lead to a
Denial of Service attack (see Section 12). In such instances,
the AC MUST ensure that this state information is freed after
a period, which is implementation specific.
Discovery to DTLS Setup (5): This transition occurs to establish a secure DTLS session with the peer.
WTP: The WTP initiates this transition by invoking the DTLSStart
command (see Section 2.3.2.1), which starts the DTLS session
establishment with the chosen AC. The decision of which AC to
connect to is the result of the discovery phase, which is
described in Section 3.3.
AC: This is an invalid state transition for the AC.
DTLS Setup to Idle (6): This transition occurs when the DTLS connection setup has failed.
WTP: The WTP initiates this state transition when it receives a
DTLSEstablishFail notification from DTLS (see Section 2.3.2.2),
and the FailedDTLSSessionCount or the FailedDTLSAuthFailCount
counter have not reached the value of the
MaxFailedDTLSSessionRetry variable (see Section 4.8). This
error notification aborts the secure DTLS session
establishment. When this notification is received, the
FailedDTLSSessionCount counter is incremented.
AC: This is an invalid state transition for the AC.
DTLS Setup to Sulking (d): This transition occurs when repeated attempts to setup the DTLS connection have failed.
WTP: The WTP enters this state when the FailedDTLSSessionCount or
the FailedDTLSAuthFailCount counter reaches the value of the
MaxFailedDTLSSessionRetry variable (see Section 4.8). Upon
entering this state, the WTP MUST start the SilentInterval
timer. While in the Sulking state, all received CAPWAP and
DTLS protocol messages received MUST be ignored.
AC: The AC enters this state with the specific WTP when the FailedDTLSSessionCount or the FailedDTLSAuthFailCount counter reaches MaxFailedDTLSSessionRetry variable (see Section 4.8). Upon entering this state, the AC's Service thread MUST start the SilentInterval timer, and ignore all CAPWAP and DTLS protocol messages received from the WTP. The AC immediately transitions the state to Idle.
DTLS Setup to DTLS Setup (e): This transition occurs when the DTLS Session failed to be established.
WTP: This is an invalid state transition for the WTP.
AC: The AC initiates this state transition by the Service thread
when it receives a DTLSEstablishFail notification from DTLS
(see Section 2.3.2.2). This error notification aborts the
secure DTLS session establishment. When this notification is
received, the FailedDTLSSessionCount counter is incremented.
DTLS Setup to Authorize (f): This transition occurs when an incoming
DTLS session is being established, and the DTLS stack needs
authorization to proceed with the session establishment.
WTP: This state transition occurs when the WTP receives the DTLSPeerAuthorize notification (see Section 2.3.2.2). Upon entering this state, the WTP performs an authorization check against the AC credentials. See Section 2.4.4 for more information on AC authorization.
AC: This state transition occurs when the AC receives the DTLSPeerAuthorize notification (see Section 2.3.2.2). Upon entering this state, the AC performs an authorization check against the WTP credentials. See Section 2.4.4 for more information on WTP authorization.
Authorize to DTLS Connect (g): This transition occurs to notify the
DTLS stack that the session should be established.
WTP: This state transition occurs when the WTP has successfully authorized the AC's credentials (see Section 2.4.4). This is done by invoking the DTLSAccept DTLS command (see Section 2.3.2.1).
AC: This state transition occurs when the AC has successfully
authorized the WTP's credentials (see Section 2.4.4). This is
done by invoking the DTLSAccept DTLS command (see
Section 2.3.2.1).
Authorize to DTLS Teardown (h): This transition occurs to notify the
DTLS stack that the session should be aborted.
WTP: This state transition occurs when the WTP was unable to authorize the AC, using the AC credentials. The WTP then aborts the DTLS session by invoking the DTLSAbortSession command (see Section 2.3.2.1).
AC: This state transition occurs when the AC was unable to authorize the WTP, using the WTP credentials. The AC then aborts the DTLS session by invoking the DTLSAbortSession command (see Section 2.3.2.1).
DTLS Connect to DTLS Teardown (7): This transition occurs when the DTLS Session failed to be established.
WTP: This state transition occurs when the WTP receives either a
DTLSAborted or DTLSAuthenticateFail notification (see
Section 2.3.2.2), indicating that the DTLS session was not
successfully established. When this transition occurs due to
the DTLSAuthenticateFail notification, the
FailedDTLSAuthFailCount is incremented, otherwise the
FailedDTLSSessionCount counter is incremented.
AC: This is an invalid state transition for the AC.
DTLS Connect to DTLS Setup (i): This transition occurs when the DTLS
Session failed to be established.
WTP: This is an invalid state transition for the WTP.
AC: This state transition occurs when the AC receives either a DTLSAborted or DTLSAuthenticateFail notification (see Section 2.3.2.2), indicating that the DTLS session was not successfully established, and both of the FailedDTLSAuthFailCount and FailedDTLSSessionCount counters have not reached the value of the MaxFailedDTLSSessionRetry variable (see Section 4.8).
DTLS Connect to Dead (j): This transition occurs when the DTLS Session failed to be established.
WTP: This is an invalid state transition for the WTP.
AC: This state transition occurs when the AC receives either a DTLSAborted or DTLSAuthenticateFail notification (see Section 2.3.2.2), indicating that the DTLS session was not successfully established, and either the FailedDTLSAuthFailCount and FailedDTLSSessionCount counters have reached the value of the MaxFailedDTLSSessionRetry variable (see Section 4.8).
DTLS Connect to Join (k): This transition occurs when the DTLS Session is successfully established.
WTP: This state transition occurs when the WTP receives the
DTLSEstablished notification (see Section 2.3.2.2), indicating
that the DTLS session was successfully established. When this
notification is received, the FailedDTLSSessionCount counter is
set to zero.
AC: This state transition occurs when the AC receives the
DTLSEstablished notification (see Section 2.3.2.2), indicating
that the DTLS session was successfully established. When this
notification is received, the FailedDTLSSessionCount counter is
set to zero, and the WaitJoin timer is started (see
Section 4.7).
Join to DTLS Teardown (l): This transition occurs when the join process failed.
WTP: This state transition occurs when the WTP receives a Join
Response message with a Result Code message element containing
an error, if the Image Identifier provided by the AC in the
Join Response message differs from the WTP's currently running
firmware version and the WTP has the requested image in its
non-volatile memory, or if the WaitDTLS timer expires. This
causes the WTP to initiate the DTLSShutdown command (see
Section 2.3.2.1). This transition also occurs if the WTP
receives one of the following DTLS notifications: DTLSAborted,
DTLSReassemblyFailure or DTLSPeerDisconnect.
AC: This state transition occurs either if the WaitJoin timer
expires or if the AC transmits a Join Response message with a
Result Code message element containing an error. This causes
the AC to initiate the DTLSShutdown command (see
Section 2.3.2.1). This transition also occurs if the AC
receives one of the following DTLS notifications: DTLSAborted,
DTLSReassemblyFailure or DTLSPeerDisconnect.
Join to Image Data (m): This state transition is used by the WTP and
the AC to download executable firmware.
WTP: The WTP enters the Image Data state when it receives a
successful Join Response message and determines and the
included Image Identifier message element is not the same as
its currently running image. The WTP also detects that the
requested image version is not currently available in the WTP's
non-volatile storage (see Section 9.1 for a full description of
the firmware download process). The WTP initializes the
EchoInterval timer (see Section 4.7), and transmits the Image
Data Request message (see Section 9.1.1) requesting the start
of the firmware download.
AC: This state transition occurs when the AC receives the Image Data Request message from the WTP. The AC MUST transmit an Image Data Response message (see Section 9.1.2) to the WTP, which includes a portion of the firmware. The AC MUST start the ImageDataStartTimer timer (see Section 4.7).
Join to Configure (n): This state transition is used by the WTP and
the AC to exchange configuration information.
WTP: The WTP enters the Configure state when it receives a
successful Join Response, and determines that the included
Image Identifier message element is the same as its currently
running image. The WTP transmits the Configuration Status
message (see Section 8.2) to the AC with message elements
describing its current configuration. The WTP also starts the
ResponseTimeout timer (see Section 4.7).
AC: This state transition occurs immediately after the AC
transmits the Join Response message to the WTP. If the AC
receives the Configuration Status message from the WTP, the AC
MUST transmit a Configuration Status Response message (see
Section 8.3) to the WTP, and MAY include specific message
elements to override the WTP's configuration. The AC also
starts the ChangeStatePendingTimer timer (see Section 4.7).
Configure to Reset (o): This state transition is used to reset the
connection either due to an error during the configuration phase,
or when the WTP determines it needs to reset in order for the new
configuration to take effect.
WTP: The WTP enters the Reset state when it receives a Configuration Status Response indicating an error or when it determines that a reset of the WTP is required, due to the characteristics of a new configuration.
AC: The AC transitions to the Reset state when it receives a
Change State Event message from the WTP that contains an error
for which AC policy does not permit the WTP to provide service.
This state transition also occurs when the AC
ChangeStatePendingTimer timer expires.
Configure to DTLS Teardown (p): This transition occurs when the configuration process aborts due to a DTLS error.
WTP: The WTP enters this state when it receives one of the
following DTLS notifications: DTLSAborted,
DTLSReassemblyFailure or DTLSPeerDisconnect (see
Section 2.3.2.2). The WTP MAY tear down the DTLS session if it
receives frequent DTLSDecapFailure notifications.
AC: The AC enters this state when it receives one of the
following DTLS notifications: DTLSAborted,
DTLSReassemblyFailure or DTLSPeerDisconnect (see
Section 2.3.2.2). The WTP MAY tear down the DTLS session if it
receives frequent DTLSDecapFailure notifications.
Image Data to Image Data (q): The Image Data state is used by the WTP and the AC during the firmware download phase.
WTP: The WTP enters the Image Data state when it receives an Image Data Response message indicating that the AC has more data to send.
AC: This state transition occurs when the AC receives the Image Data Request message from the WTP while already in the Image Data state. The AC resets the ImageDataStartTimer timer.
Image Data to Reset (r): This state transition is used to reset the
DTLS connection prior to restarting the WTP after an image
download.
WTP: When an image download completes, the WTP enters the Reset state. The WTP MAY also transition to this state upon receiving an Image Data Response message from the AC (see Section 9.1.2) indicating a failure.
AC: The AC enters the Reset state when an error occurs during the
image download process or if the ImageDataStartTimer timer
expires.
Image Data to DTLS Teardown (s): This transition occurs when the firmware download process aborts due to a DTLS error.
WTP: The WTP enters this state when it receives one of the
following DTLS notifications: DTLSAborted,
DTLSReassemblyFailure or DTLSPeerDisconnect (see
Section 2.3.2.2). The WTP MAY tear down the DTLS session if it
receives frequent DTLSDecapFailure notifications.
AC: The AC enters this state when it receives one of the
following DTLS notifications: DTLSAborted,
DTLSReassemblyFailure or DTLSPeerDisconnect (see
Section 2.3.2.2). The WTP MAY tear down the DTLS session if it
receives frequent DTLSDecapFailure notifications.
Configure to Data Check (t): This state transition occurs when the WTP and AC confirm the configuration.
WTP: The WTP enters this state when it receives a successful Configuration Status Response message from the AC. The WTP initializes the EchoInterval timer (see Section 4.7), and transmits the Change State Event Request message (see Section 8.6).
AC: This state transition occurs when the AC receives the Change
State Event Request message (see Section 8.6) from the WTP.
The AC responds with a Change State Event Response message (see
Section 8.7). The AC MUST start the DataCheckTimer timer (see
Section 4.7).
Data Check to DTLS Teardown (u): This transition occurs when the WTP
does not complete the Data Check exchange.
WTP: This state transition occurs if the WTP does not receive the
Change State Event Response before a CAPWAP transmission
timeout occurs.
AC: The AC enters this state when the DataCheckTimer timer expires (see Section 4.7).
Data Check to Run (V): This state transition occurs when the linkage
between the control and data channels has occured, causing the WTP
and AC to enter their normal state of operation.
WTP: The WTP enters this state when it receives a successful
Change State Event Response message from the AC. The WTP
initiates the data channel, which MAY require the establishment
of a DTLS session, starts the DataChannelKeepAlive timer (see
Section 4.7) and transmits a Data Channel Keep Alive packet
(see Section 4.4.1). The WTP then starts the
DataChannelDeadInterval timer (see Section 4.7).
AC: This state transition occurs when the AC receives the Data Channel Keep Alive packet (see Section 4.4.1), with a Session ID message element matching that included by the WTP in the Join Request message. The AC disables the DataCheckTimer timer. Note that if AC policy is to require the data channel to be encrypted, this process would also require the establishment of a data channel DTLS session. Upon receiving the Data Channel Keep Alive packet, the AC transmits its own Data Channel Keep Alive packet.
Run to DTLS Teardown (w): This state transition occurs when an error
has occured in the DTLS stack, causing the DTLS session to be
torndown.
WTP: The WTP enters this state when it receives one of the
following DTLS notifications: DTLSAborted,
DTLSReassemblyFailure or DTLSPeerDisconnect (see
Section 2.3.2.2). The WTP MAY tear down the DTLS session if it
receives frequent DTLSDecapFailure notifications. The WTP also
transitions to this state if the underlying reliable
transport's RetransmitCount counter has reached the
MaxRetransmit variable (see Section 4.7).
AC: The AC enters this state when it receives one of the
following DTLS notifications: DTLSAborted,
DTLSReassemblyFailure or DTLSPeerDisconnect (see
Section 2.3.2.2). The WTP MAY tear down the DTLS session if it
receives frequent DTLSDecapFailure notifications. The AC
transitions to this state if the underlying reliable
transport's RetransmitCount counter has reached the
MaxRetransmit variable (see Section 4.7).
Run to Run (x): This is the normal state of operation.
WTP: This is the WTP's normal state of operation. There are many
events that result this state transition:
Configuration Update: The WTP receives a Configuration Update
Request message(see Section 8.4). The WTP MUST respond with
a Configuration Update Response message (see Section 8.5).
Change State Event: The WTP receives a Change State Event
Response message, or determines that it must initiate a
Change State Event Request message, as a result of a failure
or change in the state of a radio.
Echo Request: The WTP sends an Echo Request message (Section 7.1) or receives the corresponding Echo Response message, (see Section 7.2) from the AC.
Clear Config Request: The WTP receives a Clear Configuration Request message (see Section 8.8). The WTP MUST reset its configuration back to manufacturer defaults.
WTP Event: The WTP sends a WTP Event Request message,
delivering information to the AC (see Section 9.4). The WTP
receives a WTP Event Response message from the AC (see
Section 9.5).
Data Transfer: The WTP sends a Data Transfer Request message to the AC (see Section 9.6). The WTP receives a Data Transfer Response message from the AC (see Section 9.7).
Station Configuration Request: The WTP receives a Station Configuration Request message (see Section 10.1), to which it MUST respond with a Station Configuration Response message (see Section 10.2).
AC: This is the AC's normal state of operation:
Configuration Update: The AC sends a Configuration Update Request message (see Section 8.4) to the WTP to update its configuration. The AC receives a Configuration Update Response message (see Section 8.5) from the WTP.
Change State Event: The AC receives a Change State Event
Request message (see Section 8.6), to which it MUST respond
with the Change State Event Response message (see
Section 8.7).
Echo Request: The AC receives an Echo Request message (see
Section 7.1), to which it MUST respond with an Echo Response
message(see Section 7.2).
Clear Config Response: The AC receives a Clear Configuration Response message from the WTP (see Section 8.9).
WTP Event: The AC receives a WTP Event Request message from
the WTP (see Section 9.4) and MUST generate a corresponding
WTP Event Response message (see Section 9.5).
Data Transfer: The AC receives a Data Transfer Request message
from the WTP (see Section 9.6) and MUST generate a
corresponding Data Transfer Response message (see
Section 9.7).
Station Configuration Request: The AC sends a Station
Configuration Request message (see Section 10.1) or receives
the corresponding Station Configuration Response message
(see Section 10.2) from the WTP.
Run to Reset (y): This state transition is used when either the AC
or WTP tear down the connection. This may occur as part of normal
operation, or due to error conditions.
WTP: The WTP enters the Reset state when it receives a Reset Request message from the AC.
AC: The AC enters the Reset state when it transmits a Reset Request message to the WTP.
Reset to DTLS Teardown (z): This transition occurs when the CAPWAP reset is complete, to terminate the DTLS session.
WTP: This state transition occurs when the WTP receives a Reset Response message. This causes the WTP to initiate the DTLSShutdown command (see Section 2.3.2.1).
AC: This state transition occurs when the AC transmits a Reset Response message. The AC does not invoke the DTLSShutdown command (see Section 2.3.2.1).
DTLS Teardown to Idle (8): This transition occurs when the DTLS session has been shutdown.
WTP: This state transition occurs when the WTP has successfully
cleaned up all resources associated with the control plane DTLS
session. The data plane DTLS session is also shutdown, and all
resources freed, if a DTLS session was established for the data
plane. Any timers set for the current instance of the state
machine are also cleared.
AC: This is an invalid state transition for the AC.
DTLS Teardown to Sulking (9): This transition occurs when repeated attempts to setup the DTLS connection have failed.
WTP: The WTP enters this state when the FailedDTLSSessionCount or
the FailedDTLSAuthFailCount counter reaches the value of the
MaxFailedDTLSSessionRetry variable (see Section 4.8). Upon
entering this state, the WTP MUST start the SilentInterval
timer. While in the Sulking state, all received CAPWAP and
DTLS protocol messages received MUST be ignored.
AC: This is an invalid state transition for the AC.
DTLS Teardown to Dead (~): This transition occurs when the DTLS session has been shutdown.
WTP: This is an invalid state transition for the WTP.
AC: This state transition occurs when the AC has successfully
cleaned up all resources associated with the control plane DTLS
session. The data plane DTLS session is also shutdown, and all
resources freed, if a DTLS session was established for the data
plane. Any timers set for the current instance of the state
machine are also cleared. The AC's Service thread is
terminated.
12.3. Discovery Attacks
[...]
Some implementations may wish to pass information about clients who
have passed the discovery phase to the DTLS layer, authorizing only
those clients to initiate a DTLS handshake. Note that the impact of
this on mitigating denial of service attacks against the DTLS layer
is minimal, because DTLS already uses client-side cookies to minimize
processor consumption attacks. As a result, implementations SHOULD
NOT maintain state between the discovery and DTLS handshake phases of
the CAPWAP protocol initialization.
</text>
PatC -----Original Message----- From: Pat Calhoun (pacalhou) Sent: Tuesday, December 04, 2007 11:17 AM To: capwap Subject: [Capwap] Issue 30: Inconsistent state tracking on AC prior to
DTLSEstablishment
I wanted to provide a follow-up based on the conversation during the meeting yesterday. The issue that is being raised in this issue is that the state machine doesn't really work well on the AC. There is text that states that the AC has a different "state context" per WTP, but right now this would require that the AC maintains state during the discovery, etc. However, we have all agreed that the AC should not
have to do so.
So we agreed that we needed to provide more clarity on the state machine that allows for the AC to provide services such as responding to Discovery Requests, without explicitely maintaining per WTP state.
I am working on modifications to the state machine, and expect to have
this sent to the list later today.
PatC
_________________________________________________________________
To unsubscribe or modify your subscription options, please visit:
http://lists.frascone.com/mailman/listinfo/capwap
Archives: http://lists.frascone.com/pipermail/capwap _________________________________________________________________ To unsubscribe or modify your subscription options, please visit: http://lists.frascone.com/mailman/listinfo/capwap
Archives: http://lists.frascone.com/pipermail/capwap
_________________________________________________________________ To unsubscribe or modify your subscription options, please visit: http://lists.frascone.com/mailman/listinfo/capwap
Archives: http://lists.frascone.com/pipermail/capwap
-
Issue 30: Inconsistent state tracking on AC prior to DTLS Establishment Pat Calhoun (pacalhou), December 4 2007
-
Re: Issue 30: Inconsistent state tracking on AC prior to DTLSEstablishment Pat Calhoun (pacalhou), December 4 2007
-
Re: Issue 30: Inconsistent state tracking on AC prior toDTLSEstablishment Scott Kelly, December 12 2007
- Re: Issue 30: Inconsistent state tracking on AC prior toDTLSEstablishment Pat Calhoun (pacalhou), December 13 2007
- Re: Issue 30: Inconsistent state tracking on AC prior toDTLSEstablishment Charles Clancy, December 15 2007
-
Re: Issue 30: Inconsistent state tracking on AC prior toDTLSEstablishment Scott Kelly, December 12 2007
- Re: Issue 30: Inconsistent state tracking on AC prior toDTLSEstablishment Margaret Wasserman, December 15 2007
-
Re: Issue 30: Inconsistent state tracking on AC prior to DTLSEstablishment Pat Calhoun (pacalhou), December 4 2007
Results generated by Tiger Technologies using MHonArc.