93 Commits

Author SHA1 Message Date
Ian Barber
3ae68d67a4 Revert "Use the hiccup mechanism to notify the socket end of the pair of the change in state, and have it shutdown that end, and shutdown the local end normally. This seems to resolve the shutdown and race condition issues."
This reverts commit 67497a26439239fdb95652b46d7e4e41fe672b31.
2012-06-12 14:46:23 +01:00
Ian Barber
c9926f6f24 Revert "As Martin pointed out, there is a race condition in the old code where a pipe could start shutting down after disconnection, but the new one could connect first. This connection would not get a pipe created for it, so the messages could never flow. The simplest way round this would be a flag, but it is possibly for a very bouncy but fast connection to go up and down twice I imagine, so instead I have added a counter. This starts at zero, and will null out the pipe if terminate is called while it is zero. On a disconnect situation the counter is incremented, and the pipe is the not nulled if the value is non zero. In the terminated function it is decremented for each pipe that is shut down, and the assertion that the terminated pipe == the current pipe is skipped while it is non-zero. This should deal with the race condition and not allow any extra terminated() calls without hitting the assertion."
This reverts commit a5f7300da6aa2224638fc932fe5ca3624189b1c1.
2012-06-12 14:45:14 +01:00
Ian Barber
19da88be67 Revert "Filter read and write activated calls from the pipe to the session, and delay shutdown in terminated until the final pipe is shutdown."
This reverts commit a90c1db7d2c0d32f9a0177d46502505efac654a5.
2012-06-12 14:44:41 +01:00
Ian Barber
eb14890d23 Revert "Revert "Merge branch 'master' of github.com:ianbarber/libzmq""
This reverts commit 029d3dfae2c2bf9e10c7f05d78593f481569bbd7.
2012-06-12 14:43:18 +01:00
Ian Barber
029d3dfae2 Revert "Merge branch 'master' of github.com:ianbarber/libzmq"
This reverts commit 33459029793474d14cdb48f85ad2c58dc7ed156f, reversing
changes made to 889b0e6f29f0cce566339b13b8a44f3b97836b6e.
2012-06-12 14:13:17 +01:00
Ian Barber
6117a2b099 Revert "Replace incomplete count with a std::set"
This reverts commit 4aa5ba3d119b98c8f1e4e682a6ee7c57d460cc55.
Unintentional merge
2012-06-12 14:07:54 +01:00
Ian Barber
4aa5ba3d11 Replace incomplete count with a std::set
This commit removes the countdown flag and adds a set to store the pipes
that are currently being disconnected.
2012-06-12 12:13:21 +01:00
Ian Barber
a90c1db7d2 Filter read and write activated calls from the pipe to the session, and delay shutdown in terminated until the final pipe is shutdown. 2012-06-11 07:58:59 +01:00
Ian Barber
a5f7300da6 As Martin pointed out, there is a race condition in the old code where a pipe could start shutting down after disconnection, but the new one could connect first. This connection would not get a pipe created for it, so the messages could never flow. The simplest way round this would be a flag, but it is possibly for a very bouncy but fast connection to go up and down twice I imagine, so instead I have added a counter. This starts at zero, and will null out the pipe if terminate is called while it is zero. On a disconnect situation the counter is incremented, and the pipe is the not nulled if the value is non zero. In the terminated function it is decremented for each pipe that is shut down, and the assertion that the terminated pipe == the current pipe is skipped while it is non-zero. This should deal with the race condition and not allow any extra terminated() calls without hitting the assertion. 2012-06-10 19:57:02 +01:00
Ian Barber
67497a2643 Use the hiccup mechanism to notify the socket end of the pair of the change in state, and have it shutdown that end, and shutdown the local end normally. This seems to resolve the shutdown and race condition issues. 2012-06-08 23:55:42 +01:00
Ian Barber
b84b007981 Reverted to a simpler shutdown. This seems to disconnect and reconnect the pipe properly, but there is a problem in overall shutdown when the pipe has blocked and reconnected - the session seems to get terminated() called on it only in shutdown for the original pipe, by which point it has been replaced. I am not sure at the moment why this only happens then, but this does mean this patch is broken at the moment 2012-06-06 23:12:56 +01:00
Ian Barber
336f72720a The previous code to block the socket from receiving on that pipe during a disconnect was vulnerable to a race condition. This code calls with the terminate functions of both ends of the pipe - I believe this should be safer. This required storing a pointer to the socket end of the pipe 2012-06-05 21:44:23 +01:00
Ian Barber
5008f385ba When detaching a pipe, as well as checking the delay on connect sockopt is set, also ensure that the protocol is not pgm or epgm as we are not implementing the functionality for multicase types 2012-06-05 18:41:38 +01:00
Ian Barber
09956dee93 Restoring comment for clarity 2012-06-04 11:41:20 +01:00
Ian Barber
ace7c99b91 Removing unnecessary outpipe values that had been used for reconnecting existing pipes - no longer needed when using a pipe term for the delay_attach situation. 2012-06-04 11:40:14 +01:00
Ian Barber
55cbdfcf26 Fix incorrect whitespace in if statement 2012-06-04 10:31:30 +01:00
Ian Barber
c13f1d52ff On the advice of Martin Hurton, removed the new command type and just terminated the pipe in a reconnect situation, and notified the socket of the same. This handles the blocking properly, but at the cost of potentially losing in flight messages. However, this is a reasonable trade off given how much simpler it makes the patch. 2012-06-04 10:27:16 +01:00
Ian Barber
6f6466f088 Fix a number of whitespace issues in various parts of the code, add validation to most calls on the test and take a first stab at implementing the reconnection pipe blocking.
It didn't seem straightforward to use any of the existing process calls, so I have added a new command to command_t and friends called detach. This instructs the socket_base to remove the pipe from it's pipe list. The session base stores a copy of the outpipe, and will resend the bind command on reconnection. This should allow balancing again.
2012-06-03 22:57:47 +01:00
Ian Barber
06485d9200 Remove the extra outpipe handling as the session is quite capable of delaying the creation of the pipe until the connection has happened. Simply don't build the pipe, and let it do that automatically. 2012-06-03 22:05:36 +01:00
Ian Barber
297af95451 And another typo on the same comment 2012-06-03 21:38:41 +01:00
Ian Barber
b79aaaf473 Remove extra brackets as suggested by Martin H, and fix up a comment which was missing a word 2012-06-03 21:34:41 +01:00
Ian Barber
fe3fb419fe After speaking with Ben Gray and the discussion on the mailing list, this is an attempt to create a sockopt to allow connecting pipes to not immediately be available for traffic. The problem is in a PUSH to many PULL situation, where there is a connect to a PULL which is not there. This connect will immediately create a pipe (unlike bind), and traffic will be load balanced to that pipe. This means if there is a persistently unavailable end point then the traffic will queue until HWM is hit, and older messages will be lost.
This patch adds a sockopt ZMQ_DELAY_ATTACH_ON_CONNECT, which if set to 1 will attempt to preempt this behavior. It does this by extending the use of the session_base to include in the outbound as well as the inbound pipe, and only associates the pipe with the socket once it receives the connected callback via a process_attach message. This works, and a test has been added to show so, but may introduce unexpected complications. The shutdown logic in this class has become marginally more awkward because of this, requiring the session to serve as the sink for both pipes if shutdown occurs with a still-connecting pipe in place. It is also possible there could be issues around flushing the messages, but as I could not directly think how to create such an issue I have not written any code with regards to that.

The documentation has been updated to reflect the change, but please do check over the code and test and review.
2012-06-01 17:58:19 +01:00
Martin Hurton
7fe45af30f Fix issue #370
The patch extends the internal session's API with the reset method.
This method is used to reset a session's state so that it can
handle a new connection.
2012-05-29 22:10:51 +02:00
Martin Hurton
24b79c7e0b Prefer errno_assert/alloc_assert to zmq_assert 2012-05-28 23:17:03 +02:00
Martin Hurton
130dfc5b5f Fix issue #369
The bug was that after reconnect, the session did not
handle identity messages properly.
2012-05-28 17:08:27 +02:00
Martin Hurton
52ed4cdccf Remove dead code 2012-05-27 15:18:38 +02:00
KennyTM~
c995de6584 Allow the ZMQ_MONITOR code compilable on gcc 4.7 on Linux.
The current ZMQ_MONITOR code does not compile in gcc 4.7, as -pedantic
and -Werror are enabled, and ISO C++ doesn't allow casting between
normal pointers (void*) and function pointers, as pedantically their
size could be different. This caused the library not compilable. This
commit workaround the problem by introducing one more indirection, i.e.
instead of calling

    (void *)listener

which is an error, we have to use

    *(void **)&listener

which is an undefined behavior :) but works on most platforms

Also, `optval_ = monitor` will not set the parameter in getsockopt(),
and the extra casting caused the LHS to be an rvalue which again makes
the code not compilable. The proper way is to pass a pointer of function
pointer and assign with indirection, i.e. `*optval_ = monitor`.

Also, fixed an asciidoc error in zmq_getsockopt.txt because the `~~~~`
is too long.
2012-05-13 20:49:05 +08:00
Lourens Naudé
5c6f72c17c ZMQ_MONITOR socket option registers a callback / event sink for changes in socket state 2012-05-04 02:32:46 +01:00
Sergey KHripchenko
74ae19ac1f spaces deleted 2012-04-21 18:36:20 +04:00
Martin Hurton
00b4571bf1 Fixed issue #334 2012-03-25 20:06:06 +02:00
Pieter Hintjens
c12fedc70a Completed internal renaming of XREP/XREQ to ROUTER/DEALER 2012-03-22 11:36:19 -05:00
Emmanuel Taurel
107581213c Disable reconnection option
Add value -1 to the ZMQ_RECONNECT_IVL to disable the reconnection algorithm
2012-03-20 09:22:27 +01:00
Staffan Gimåker
b9fb48f47b Resolve addresses in the calling thread on connect.
This allows us to actually report an error to the caller on resolve
failure, rather than asserting later on in the io thread.

Signed-off-by: Staffan Gimåker <staffan@spotify.com>
2012-02-16 14:42:55 +01:00
Pieter Hintjens
afe8cd503f Revert "fix warn unused"
- anonymous contributor

This reverts commit 7b7d404269cc9bb28e1954ed38dbeb8f63079fd1.
2012-02-14 18:43:33 -06:00
niXman
7b7d404269 fix warn unused 2012-02-10 13:39:34 +04:00
Martin Sustrik
a4843b65d2 Identities re-introduced
However, the "durable socket" behaviour wasn't re-added.
Identities are used solely for routing in REQ/REP pattern.

Signed-off-by: Martin Sustrik <sustrik@250bpm.com>
2011-11-04 08:00:47 +01:00
Martin Sustrik
8e21d64c97 Copyright dates adjusted to reflect reality
Signed-off-by: Martin Sustrik <sustrik@250bpm.com>
2011-11-01 18:06:11 +01:00
Martin Sustrik
7842c71073 LABELS and COMMANDs removed
Signed-off-by: Martin Sustrik <sustrik@250bpm.com>
2011-11-01 13:39:54 +01:00
Martin Sustrik
626099aa2a VTCP transport removed
Signed-off-by: Martin Sustrik <sustrik@250bpm.com>
2011-10-31 16:37:20 +01:00
Martin Sustrik
ac7717b7b3 250bpm copyrights added
Signed-off-by: Martin Sustrik <sustrik@250bpm.com>
2011-10-31 16:20:30 +01:00
Martin Sustrik
dee8b2360a New style ROUTER socket removed.
Signed-off-by: Martin Sustrik <sustrik@turist.(none)>
2011-10-31 15:44:42 +01:00
Martin Sustrik
06bdf2c4f9 Check message syntax in REQ asynchronously
This patch adds support for checking messages as they arrive
(as opposed to when they are recv'd by the user) and drop
the connection if they are malformed.

It also uses this new feature to check for validity of inbound
messages in REQ socket.

Signed-off-by: Martin Sustrik <sustrik@250bpm.com>
2011-09-16 09:29:43 +02:00
Martin Sustrik
f78d9b6bfc Session class separated into socket-type-specific sessions
This is a preliminary patch allowing for socket-type-specific
functionality in the I/O thread. For example, message format
can be checked asynchronously and misbehaved connections dropped
straight away.

Signed-off-by: Martin Sustrik <sustrik@250bpm.com>
2011-09-15 10:00:23 +02:00