| Age | Commit message (Collapse) | Author |
|---|
|
* anders/diameter/watchdog_leak/OTP-11019:
Minor doc fix
Add testcase to exercise reconnect behaviour
Fix watchdog table leak
|
|
Use the default address address (as selected by gen_tcp) if none is
configured, passing it in the new 'connected' message introduced by the
previous commit.
The corresponding update to diameter_sctp has to wait until problems
with inet:sockname/1 are resolved: the function currently only returns
one address, and sometimes {0,0,0,0}. See OTP-11018.
|
|
A transport module can return a local address list from its start/3
function in order to specify addresses to be used as Host-IP-Address
during capabilities exchange. Now allow addresses to be communicated in
a 'connected' message in the case of a connecting transport, so that
diameter_tcp (in particular) can make local address configuration
optional, communicating the gen_tcp default after connection
establishment instead.
|
|
A service process maintains a table keyed on watchdog process pids. When
a watchdog process dies the corresponding entry should be removed but
this was broken in commit f115a9f7, causing entries with watchdog state
DOWN to accumulate.
Watchdog processes die as a result of diameter:remove_transport/2, or
when a peer reestablishes a connection in the listening case. Neither is
typically a frequent occurrence.
The fault manifests itself in the return value of
diameter:service_info(SvcName, transport), which displays entries for
watchdog processes that are no longer alive.
|
|
{error, Reason} is now returned, instead of the options being ignored.
Note that diameter:add_transport/2 purposely ignores unknown options and
that the behaviour is documented. This is historic: some users depend on
it in order to store their own options for identifying transport config,
instead of using the reference returned by add_transport.
|
|
The validation of {sequence, {H,N}} incorrectly checked that H was an
N-bit integer, instead of the intended 32-N.
|
|
Crashing watchdog and peer_fsm processes was somewhat unseemly. Emit an
error report and die silently instead.
|
|
Faulty configuration was previously passed directly on to watchdog and
peer_fsm processes, diameter:add_transport/2 happily returning ok and
the error resulting on failure of watchdog and/or peer_fsm processes.
Now check for errors before getting this far, returning {error, Reason}
from diameter:add_transport/2 when one is detected. There are still
some errors that can only be detected after transport start (eg. a
misbehaving callback) but most will be caught early.
|
|
'infinity' is a valid transport_config timeout.
|
|
Remove unused functions, add dialyzer specs, make wait/1 less fallible.
|
|
|
|
|
|
RFC 6733 has changed the arity of Vendor-Id in this Grouped AVP, from 1*
in RFC 3588 to 1 in RFC 6773. This impacts the generated dictionary
modules: Vendor-Id is expected to be list-valued in the 3588 dictionary,
integer-valued in the 6733 dictionary. This, in turn, breaks the
independence of capabilities configuration on a service or transport
from the dictionary that will be used to encode an outgoing CER or CEA.
This commit fixes this by massaging any Vendor-Specific-Application-Id
config as appropriate when constructing CER or CEA for a given
dictionary.
|
|
|
|
Allow both share_peers and use_shared_peers to be a list of nodes, or a
function that returns a list of nodes.
|
|
This is the functionality that allows transports to be shared between
identically-named services on different nodes, which has been neither
documented nor tested (until now).
|
|
Make it just a number of timeouts, without a new DWR being sent.
|
|
Counters read by diameter:service_info(SvcName, transport) can be
selected at the same time as the diameter_stats server is folding them
into another key, possibly resulting in inaccurate values. Have
diameter_stats select from the server process to avoid this and add
diameter_stats:sum/1 to sum values from all contributors on a given
term.
|
|
|
|
To make the number of watchdogs sent before the transitions REOPEN ->
OKAY and OKAY -> SUSPECT configurable. Using anything other then the
default config is non-standard and should only be used for test.
|
|
* anders/diameter/SoC/OTP-10688:
Add RFC 6733 commentary to SoC chapter
Add RFC 6737 Capabilities Update dictionary
Remove cruft from doc Makefile
|
|
|
|
Although it's neither built nor supported in code yet.
|
|
Also add comments to identify the corresponding OTP releases.
Purposely don't use regexp version numbers to be explicit about
previously released versions. (Could use them in one direction but then
app suite needs to be adapted.)
|
|
Add a 'strict' target to compile with -Werror and fix a dependency.
Reorder test suites alphabetically. The two suites that currently take
the longest to run conveniently come last.
|
|
The workaround (commit 57d5564f) was to dialyzer only understanding
nowarn_unused_function on individual functions. This is no longer the
case as of R15B01 (commit 477fd95a).
|
|
RFC 3588 allowed only 3xxx result codes in an answer-message (that is,
an answer that sets the E-bit) while RFC 6733 also allows 5xxx result
codes. Setting request_errors = answer tells diameter to answer 5xxx
errors itself. Returning {answer_message, integer()} from a
handle_request callback allows both 3xxx and 5xxx result codes to be
set. {protocol_error, integer()} is retained for 3xxx result codes.
|
|
In particular, don't put an error tuple in the errors field of
a #diameter_packet{} when Result-Code and the E-bit are in conflict, put
{integer(), #diameter_avp{}}.
|
|
When receiving a request for which errors have been detected during
decode, diameter previously used the errors list in the decoded
diameter_packet record to unconditionally set Result-Code and Failed-AVP
in the outgoing answer. It wasn't particularly delicate in doing so
however and would happily set a 5xxx Result-Code even if a
handle_request callback returned an answer-message, leading to an encode
error. This behaviour became even less endearing as of commit ac452e28,
which made it possible to handle_request to take place even for protocol
errors. (ie. When a callback typically should return an answer-message.)
This commit fixes the behaviour by only setting a value that's
appropriate for the answer in question, either a 3xxx or a 5xxx,
depending on if the answer's an answer-message or not. It also allows
handle_request to prevent diameter from setting anything by setting
errors = false in a returned diameter_packet. Ideally it should have
been errors = [] but the empty list is the default value for the errors
field and changing the default (ideally there shouldn't have been one)
would require recompilation of all modules including diameter.hrl:
choose the less attractive 'false' to avoid such backwards
incompatibility.
The request reception is also refactored somewhat to shorten some call
chains.
|
|
Configuring the value 'callback' all errors detected in incoming
requests to result in a handle_request callback. The default value
'answer_3xxx' is the previous behaviour in which diameter answers
protocol errors without a callback.
|
|
The value determines whether or not an unexpected message length in the
header of an incoming messages causes the peer process to exit, the
message to be discarded or handled as usual. The latter may only be
appropriate for message-oriented transport (eg. SCTP) since
stream-oriented transport (eg. TCP) may not be able to recover the
message boundary once a length error has occurred.
|
|
|
|
Don't start a new timer with each incoming message. Instead, start a
timer at timeout and flush after two successive timeouts with no message
reception.
|
|
|
|
Which will be the case in R16B.
|
|
|
|
|
|
Traffic handling is connected to the service implementation through the
pick_peer callback and failover but diameter_service was getting
unwieldy as home to both the service process and traffic handling.
|
|
In particular, remove fields containing values that are known (as of the
preceding commit) to the request process.
|
|
In order to be able to remove fields from the request process that don't
need to be there and do less in the service process. The pick_peer
callback now takes place in the request process in the case of immutable
state, just as in the case of the initial send.
|
|
As an inverse to '#get-'/1 in the preceding commit.
|
|
The generated '#get-'/1 has one clause for each exported record r, whose
definition is equivalent to the following.
'#get-'(#r{} = Rec) ->
[r | lists:zip(record_info(r, fields), tl(tuple_to_list(Rec)))];
The record name at the head of the list is the same format that diameter
accepts for outgoing message.
|
|
Instead, use whatever dictionary a transport has configured as
supporting application id 0. This is to support the updated RFC 6733
dictionaries (which bring with them updated records) and also to be able
to transparently support any changed semantics (eg. 5xxx in
answer-message).
|
|
The RFC 6733 accounting dictionary includes rfc6733 in its name. The RFC
3588 dictionary doesn't and is left as-is for backwards compatibility.
|
|
There is no such transition in RFC 3539, the state remains in INITIAL.
|
|
This was the result of the watchdog process exiting as a consequence of
peer death in some casesi, causing a restarted transport to enter
INITIAL when it should enter REOPEN. The watchdog now remains alive as
long as peer shutdown isn't requested and a 'close' message to the
service process (instead of watchdog death) generates 'closed' events
from the service.
|
|
|
|
That is, make the naming match that of the corresponding modules.
This has long been fairly confusing.
|
|
In particular, use watchdog messages as input and do away with the older
connection_up/down (and other) messages. Also, only maintain the
watchdog state, not the older up/down op state.
|
|
Service process informs the watchdog process which informs the peer
process. (Instead of going directly to the latter in one case.)
|
