1. TM Module Jiri Kuthan FhG FOKUS Copyright � 2003 FhG FOKUS Revision History Revision $Revision$ $Date$ __________________________________________________________________ 1.1. Overview 1.2. Known Issues 1.3. Parameters 1.3.1. fr_timer (integer) 1.3.2. fr_inv_timer (integer) 1.3.3. wt_timer (integer) 1.3.4. delete_timer (integer) 1.3.5. retr_timer1 (integer) 1.3.6. retr_timer2 (integer) 1.3.7. noisy_ctimer (integer) 1.3.8. restart_fr_on_each_reply (integer) 1.3.9. unix_tx_timeout (integer) 1.3.10. aggregate_challenges (integer) 1.3.11. blst_methods_add (unsigned integer) 1.3.12. blst_methods_lookup (unsigned integer) 1.4. Functions 1.4.1. t_relay_to_udp(ip, port), t_relay_to_tcp(ip, port) 1.4.2. t_relay() 1.4.3. t_on_failure(failure_route) 1.4.4. t_on_reply(onreply_route) 1.4.5. t_on_branch(branch_route) 1.4.6. append_branch() 1.4.7. t_newtran() 1.4.8. t_reply(code, reason_phrase) 1.4.9. t_lookup_request() 1.4.10. t_retransmit_reply() 1.4.11. t_release() 1.4.12. t_forward_nonack(ip, port) 1.4.13. t_set_fr(fr_inv_timeout [, fr_timeout]) 1.4.14. t_branch_timeout() 1.4.15. t_branch_replied() 1.4.16. t_any_timeout() 1.4.17. t_any_replied() 1.4.18. t_is_canceled() 1.5. TM Module API 1.5.1. Defines 1.5.2. Functions 1.5.2.1. register_tmcb(cb_type, cb_func) 1.5.2.2. load_tm(*import_structure) 1.1. Overview TM module enables stateful processing of SIP transactions. The main use of stateful logic, which is costly in terms of memory and CPU, is some services inherently need state. For example, transaction-based accounting (module acc) needs to process transaction state as opposed to individual messages, and any kinds of forking must be implemented statefully. Other use of stateful processing is it trading CPU caused by retransmission processing for memory. That makes however only sense if CPU consumption per request is huge. For example, if you want to avoid costly DNS resolution for every retransmission of a request to an unresolvable destination, use stateful mode. Then, only the initial message burdens server by DNS queries, subsequent retransmissions will be dropped and will not result in more processes blocked by DNS resolution. The price is more memory consumption and higher processing latency. From user's perspective, there are these major functions : t_relay, t_relay_to_udp and t_relay_to_tcp. All of them setup transaction state, absorb retransmissions from upstream, generate downstream retransmissions and correlate replies to requests. t_relay forwards to current URI (be it original request's URI or a URI changed by some of URI-modifying functions, such as sethost). t_relay_to_udp and t_relay_to_tcp forward to a specific address over UDP or TCP respectively. In general, if TM is used, it copies clones of received SIP messages in shared memory. That costs the memory and also CPU time (memcpys, lookups, shmem locks, etc.) Note that non-TM functions operate over the received message in private memory, that means that any core operations will have no effect on statefully processed messages after creating the transactional state. For example, calling record_route after t_relay is pretty useless, as the RR is added to privately held message whereas its TM clone is being forwarded. TM is quite big and uneasy to program--lot of mutexes, shared memory access, malloc and free, timers--you really need to be careful when you do anything. To simplify TM programming, there is the instrument of callbacks. The callback mechanisms allow programmers to register their functions to specific event. See t_hooks.h for a list of possible events. Other things programmers may want to know is UAC--it is a very simplistic code which allows you to generate your own transactions. Particularly useful for things like NOTIFYs or IM gateways. The UAC takes care of all the transaction machinery: retransmissions , FR timeouts, forking, etc. See t_uac prototype in uac.h for more details. Who wants to see the transaction result may register for a callback. 1.2. Known Issues * Local ACK/CANCELs copy'n'pastes Route and ignores deleted Routes (solved in ser 2.1). * Possibly, performance could be improved by not parsing non-INVITEs, as they do not be replied with 100, and do not result in ACK/CANCELs, and other things which take parsing. However, we need to rethink whether we don't need parsed headers later for something else. Remember, when we now conserver a request in sh_mem, we can't apply any pkg_mem operations to it any more. (that might be redesigned too). * Another performance improvement may be achieved by not parsing CSeq in replies until reply branch matches branch of an INVITE/CANCEL in transaction table. * t_replicate should be done more cleanly--Vias, Routes, etc. should be removed from a message prior to replicating it (well, does not matter any longer so much as there is a new replication module). 1.3. Parameters Revision History Revision $Revision$ $Date$ 1.3.1. fr_timer (integer) Timer which hits if no final reply for a request or ACK for a negative INVITE reply arrives (in milliseconds). Default value is 30000 ms (30 seconds). See also: t_set_fr(). Example 1. Set fr_timer parameter ... modparam("tm", "fr_timer", 10000) ... 1.3.2. fr_inv_timer (integer) Timer which hits if no final reply for an INVITE arrives after a provisional message was received (in milliseconds). Note: this timer can be restarted when a provisional response is received. For more details see restart_fr_on_each_reply. Default value is 120000 ms (120 seconds). See also: t_set_fr(). Example 2. Set fr_inv_timer parameter ... modparam("tm", "fr_inv_timer", 180000) ... 1.3.3. wt_timer (integer) Time for which a transaction stays in memory to absorb delayed messages after it completed (in milliseconds); also, when this timer hits, retransmission of local cancels is stopped (a puristic but complex behavior would be not to enter wait state until local branches are finished by a final reply or FR timer--we simplified). Default value is 5000 ms (5 seconds). Example 3. Set wt_timer parameter ... modparam("tm", "wt_timer", 1000) ... 1.3.4. delete_timer (integer) Time after which a to-be-deleted transaction currently ref-ed by a process will be tried to be deleted again (in milliseconds). Default value is 200 milliseconds. Example 4. Set delete_timer parameter ... modparam("tm", "delete_timer", 100) ... 1.3.5. retr_timer1 (integer) Initial retransmission period (in milliseconds). Default value is 500 milliseconds. Example 5. Set retr_timer1 parameter ... modparam("tm", "retr_timer1", 1000) ... 1.3.6. retr_timer2 (integer) Maximum retransmission period (in milliseconds). The retransmission interval starts with retr_timer1 and increases until it reaches this value. After this it stays constant at retr_timer2. Default value is 4000 milliseconds. Example 6. Set retr_timer2 parameter ... modparam("tm", "retr_timer2", 2000) ... 1.3.7. noisy_ctimer (integer) If set, INVITE transactions that time-out (FR INV timer) will be always replied. If it's not set, the transaction has only one branch and no response was ever received on this branch, it will be silently dropped (no 408 reply will be generated) This behavior is overridden if a request is forked, the transaction has a failure route or callback, or some functionality explicitly turned it on for a transaction (like acc does to avoid unaccounted transactions due to expired timer). Turn this off only if you know the client UACs will timeout and their timeout interval for INVITEs is lower or equal than tm's fr_inv_timer. Default value is 0 (false). Example 7. Set noisy_ctimer parameter ... modparam("tm", "noisy_ctimer", 1) ... 1.3.8. restart_fr_on_each_reply (integer) If set (default), the fr_inv_timer for an INVITE transaction will be restarted for each provisional reply received (rfc3261 mandated behaviour). If not set, the fr_inv_timer will be restarted only for the first provisional replies and for increasing replies greater or equal 180 (e.g. 180, 181, 182, 185, ...). Setting it to 0 is especially useful when dealing with bad UAs that continuously retransmit 180s, not allowing the transaction to timeout (and thus making impossible the implementation of certain services, like automatic voicemail after x seconds). Default value is 1 (on). See also: fr_inv_timer. Example 8. Set restart_fr_on_each_reply parameter ... modparam("tm", "restart_fr_on_each_reply", 0) ... 1.3.9. unix_tx_timeout (integer) Unix socket transmission timeout, in milliseconds. If unix sockets are used (e.g.: to communicate with sems) and sending a message on a unix socket takes longer then unix_tx_timeout, the send will fail. The default value is 500 milliseconds. Example 9. Set unix_tx_timeout parameter ... modparam("tm", "unix_tx_timeout", 250) ... 1.3.10. aggregate_challenges (integer) If set (default), the final reply is a 401 or a 407 and more then one branch received a 401 or 407, then all the WWW-Authenticate and Proxy-Authenticate headers from all the 401 and 407 replies will be aggregated in a new final reply. If only one branch received the winning 401 or 407 then this reply will be forwarded (no new one will be built). If 0 only the first 401, or if no 401 was received the first 407, will be forwarded (no header aggregation). Default value is 1 (required by rfc3261). Example 10. Set aggregate_challenges parameter ... modparam("tm", "aggregate_challenges", 0) ... 1.3.11. blst_methods_add (unsigned integer) Bitmap of method types that trigger blacklisting on transaction timeouts. (This setting has no effect on blacklisting because of send failures.) The following values are associated to the request methods: INVITE=1, CANCEL=2, ACK=4 (not retransmitted, thus, never times-out), BYE=8, INFO=16, REGISTER=32, SUBSCRIBE=64, NOTIFY=126, OTHER=256 (all the unknown types). Check parser/msg_parser.h for farther details. Change the value carefully, because requests not having provisional response (everything but INVITE) can easily cause the next hop to be inserted into the blacklist by mistake. For exmaple the next hop is a proxy, it is alive, but waiting for the response of the UAS, and has higher fr_timer value. The default value is 1, only INVITEs trigger blacklisting Example 11. Set blst_methods_add parameter ... # INVITEs and REGISTERs trigger blacklisting modparam("tm", "blst_methods_add", 33) ... 1.3.12. blst_methods_lookup (unsigned integer) Bitmap of method types that are looked-up in the blacklist before statefull forwarding. See also blst_methods_add The default value is 4294967287, every method type except BYE. (We try to deliver BYEs no matter what) Example 12. Set blst_methods_lookup parameter ... # lookup only INVITEs modparam("tm", "blst_methods_lookup", 1) ... 1.4. Functions Revision History Revision $Revision$ $Date$ 1.4.1. t_relay_to_udp(ip, port), t_relay_to_tcp(ip, port) Relay a message statefully to a fixed destination. This along with t_relay is the function most users want to use--all other are mostly for programming. Programmers interested in writing TM logic should review how t_relay is implemented in tm.c and how TM callbacks work. Meaning of the parameters is as follows: * ip - IP address where the message should be sent. * port - Port number. Example 13. t_relay_to_udp usage ... t_relay_to_udp("1.2.3.4", "5060"); ... 1.4.2. t_relay() Relay a message statefully to destination indicated in current URI. (If the original URI was rewritten by UsrLoc, RR, strip/prefix, etc., the new URI will be taken). Returns a negative value on failure--you may still want to send a negative reply upstream statelessly not to leave upstream UAC in lurch. Example 14. t_relay usage ... if (!t_relay()) { sl_reply_error(); break; }; ... 1.4.3. t_on_failure(failure_route) Sets failure routing block, to which control is passed after a transaction completed with a negative result but before sending a final reply. In the referred block, you can either start a new branch (good for services such as forward_on_no_reply) or send a final reply on your own (good for example for message silo, which received a negative reply from upstream and wants to tell upstream "202 I will take care of it"). Note that the set of commands which are usable within failure_routes is strictly limited to rewriting URI, initiating new branches, logging, and sending stateful replies (t_reply). Any other commands may result in unpredictable behavior and possible server failure. Note that whenever failure_route is entered, uri is reset to value which it had on relaying. If it temporarily changed during a reply_route processing, subsequent reply_route will ignore the changed value and use again the original one. Meaning of the parameters is as follows: * failure_route - Failure route block to be called. Example 15. t_on_failure usage ... route { t_on_failure("1"); t_relay(); } failure_route[1] { revert_uri(); setuser("voicemail"); append_branch(); } ... See test/onr.cfg for a more complex example of combination of serial with parallel forking. 1.4.4. t_on_reply(onreply_route) Sets the reply routing block, to which control is passed when a reply for the current transaction is received. Note that the set of commands which are usable within onreply_routes is limited. Meaning of the parameters is as follows: * onreply_route - Onreply route block to be called. Example 16. t_on_reply usage ... loadmodule "/usr/local/lib/ser/modules/nathelper.so" ... route { /* if natted */ t_on_reply("1"); t_relay(); } onreply_route[1] { if (status=~ "(183)|2[0-9][0-9]"){ force_rtp_proxy(); search_append('^(Contact|m)[ \t]*:.*sip:[^>[:cntrl:]]*', ';nat=y es'); } if (nat_uac_test("1")){ fix_nated_contact(); } } 1.4.5. t_on_branch(branch_route) Sets the branch routing block, to which control is passed after forking (when a new branch is created). For now branch routes are intended only for last minute changes of the SIP messages (like adding new headers). Note that the set of commands which are usable within branch_routes is very limited. It is not possible to drop a message or generate a reply. Meaning of the parameters is as follows: * branch_route - branch route block to be called. Example 17. t_on_branch usage ... route { t_on_branch("1"); t_relay(); } branch_route[1] { if (uri=~"sip:[0-9]+"){ append_hf("P-Warn: numeric uri\r\n"); } } 1.4.6. append_branch() Similarly to t_fork_to, it extends destination set by a new entry. The difference is that current URI is taken as new entry. Example 18. append_branch usage ... set_user("john"); t_fork(); set_user("alice"); t_fork(); t_relay(); ... 1.4.7. t_newtran() Creates a new transaction, returns a negative value on error. This is the only way a script can add a new transaction in an atomic way. Typically, it is used to deploy a UAS. Example 19. t_newtran usage ... if (t_newtran()) { log("UAS logic"); t_reply("999","hello"); } else sl_reply_error(); ... See test/uas.cfg for more examples. 1.4.8. t_reply(code, reason_phrase) Sends a stateful reply after a transaction has been established. See t_newtran for usage. Meaning of the parameters is as follows: * code - Reply code number. * reason_phrase - Reason string. Example 20. t_reply usage ... t_reply("404", "Not found"); ... 1.4.9. t_lookup_request() Checks if a transaction exists. Returns a positive value if so, negative otherwise. Most likely you will not want to use it, as a typical application of a looku-up is to introduce a new transaction if none was found. However this is safely (atomically) done using t_newtran. Example 21. t_lookup_request usage ... if (t_lookup_request()) { ... }; ... 1.4.10. t_retransmit_reply() Retransmits a reply sent previously by UAS transaction. Example 22. t_retransmit_reply usage ... t_retransmit_reply(); ... 1.4.11. t_release() Remove transaction from memory (it will be first put on a wait timer to absorb delayed messages). Example 23. t_release usage ... t_release(); ... 1.4.12. t_forward_nonack(ip, port) mainly for internal usage--forward a non-ACK request statefully. Meaning of the parameters is as follows: * ip - IP address where the message should be sent. * port - Port number. Example 24. t_forward_nonack usage ... t_forward_nonack("1.2.3.4", "5060"); ... 1.4.13. t_set_fr(fr_inv_timeout [, fr_timeout]) Sets the fr_inv_timeout and optionally fr_timeout for the current transaction. If the transaction is already created (e.g called after t_relay() or in an onreply_route) all the branches will have their final response timeout updated on-the-fly. If one of the parameters is 0, its value won't be changed. Meaning of the parameters is as follows: * fr_inv_timeout - new final response timeout (in milliseconds) for INVITEs. See also fr_inv_timer. fr_timeout - new final response timeout (in milliseconds) for non-INVITE transaction, or INVITEs which haven't received yet a provisional response. See also fr_timer. See also: fr_timer, fr_inv_timer. Example 25. t_set_fr usage ... route { t_set_fr(10000); # set only fr invite timeout to 10s t_on_branch("1"); t_relay(); } branch_route[1] { # if we are calling the pstn, extend the invite timeout to 50s # for all the branches, and set the no-reply-received timeout to 2s if (uri=~"sip:[0-9]+"){ t_set_fr(50000, 2000); } } 1.4.14. t_branch_timeout() Returns true if the failure route is executed for a branch that did timeout. It can be used only from the failure_route. Example 26. t_branch_timeout usage ... failure_route[0]{ if (t_branch_timeout()){ log("timeout\n"); # ... } } 1.4.15. t_branch_replied() Returns true if the failure route is executed for a branch that did receive at least one reply in the past (the "current" reply is not taken into account). It can be used only from the failure_route. Example 27. t_branch_replied usage ... failure_route[0]{ if (t_branch_timeout()){ if (t_branch_replied()) log("timeout after receiving a reply (no answer?)\n"); else log("timeout, remote side seems to be down\n"); # ... } } 1.4.16. t_any_timeout() Returns true if at least one of the current transactions branches did timeout. Example 28. t_any_timeout usage ... failure_route[0]{ if (!t_branch_timeout()){ if (t_any_timeout()){ log("one branch did timeout\n"); sl_send_reply("408", "Timeout"); } } } 1.4.17. t_any_replied() Returns true if at least one of the current transactions branches did receive some reply in the past. If called from a failure or onreply route, the "current" reply is not taken into account. Example 29. t_any_replied usage ... onreply_route[0]{ if (!t_any_replied()){ log("first reply received\n"); # ... } } 1.4.18. t_is_canceled() Returns true if the current transaction was canceled. Example 30. t_is_canceled usage ... failure_route[0]{ if (t_is_canceled()){ log("transaction canceled\n"); # ... } } 1.5. TM Module API Revision History Revision $Revision$ $Date$ There are applications which would like to generate SIP transactions without too big involvement in SIP stack, transaction management, etc. An example of such an application is sending instant messages from a website. To address needs of such apps, SER accepts requests for new transactions via fifo pipes too. If you want to enable this feature, start FIFO server with configuration option. fifo="/tmp/ser_fifo" Then, an application can easily launch a new transaction by writing a transaction request to this named pipe. The request must follow very simple format, which is :t_uac_from:[<file_name>]\n <method>\n <sender's uri>\n <dst uri>\n <CR_separated_headers>\n <body>\n .\n \n (Filename is to where a report will be dumped. ser assumes /tmp as file's directory.) Note the request write must be atomic, otherwise it might get intermixed with writes from other writers. You can easily use it via Unix command-line tools, see the following example: [jiri@bat jiri]$ cat > /tmp/ser_fifo :t_uac_from:xxx MESSAGE sip:sender@iptel.org sip:mrx@iptel.org header:value foo:bar bznk:hjhjk p_header: p_value body body body yet body end of body . or cat test/transaction.fifo > /tmp/ser_fifo 1.5.1. Defines * ACK_TAG enables stricter matching of acknowledgments including to-tags. Without it, to-tags are ignored. It is disabled by default for two reasons: + It eliminates an unlikely race condition in which transaction's to-tag is being rewritten by a 200 OK whereas an ACK is being looked up by to-tag. + It makes UACs happy who set wrong to-tags. It should not make a difference, as there may be only one negative reply sent upstream and 200/ACKs are not matched as they constitute another transaction. It will make no difference at all when the new magic cookie matching is enabled anyway. * CANCEL_TAG similarly enables strict matching of CANCELs including to-tags--act of mercy to UACs, who screw up the to-tags (however, it still depends on how forgiving the downstream UAS is). Like with ACK_TAG, all this complex transactions matching goes with RFC3261's magic cookie away anyway. 1.5.2. Functions 1.5.2.1. register_tmcb(cb_type, cb_func) For programmatic use only--register a function to be called back on an event. See t_hooks.h for more details. Meaning of the parameters is as follows: * cb_type - Callback type. * cb_func - Callback function. 1.5.2.2. load_tm(*import_structure) For programmatic use only--import exported TM functions. See the acc module for an example of use. Meaning of the parameters is as follows: * import_structure - Pointer to the import structure.