res_pjsip_outbound_registration: registration stops due to fatal 4xx response

During outbound registration it is possible to receive a fatal (any permanent/
non-temporary 4xx, 5xx, 6xx) response from the registrar that is simply due
to a problem with the registrar itself. Upon receiving the failure response
Asterisk terminates outbound registration for the given endpoint.

This patch adds an option, 'fatal_retry_interval', that when set continues
outbound registration at the given interval up to 'max_retries' upon receiving
a fatal response.

ASTERISK-25485 #close

Change-Id: Ibc2c7b47164ac89cc803433c0bbe7063bfa143a2

CHANGES

@@ -162,6 +162,13 @@ res_pjsip
    will be used instead. The new SIP resolver provides NAPTR support, improved
    SRV support, and AAAA record support.
 
+res_pjsip_outbound_registration
+-------------------------------
+* A new 'fatal_retry_interval' option has been added to outbound registration.
+  When set (default is zero), and upon receiving a failure response to an
+  outbound registration, registration is retried at the given interval up to
+  'max_retries'.
+
 CEL Backends
 ------------------
 

configs/samples/pjsip.conf.sample

@@ -931,6 +931,14 @@
                         ; registration is unsuccessful (default: "60")
 ;forbidden_retry_interval=0     ; Interval used when receiving a 403 Forbidden
                                 ; response (default: "0")
+;fatal_retry_interval=0 ; Interval used when receiving a fatal response.
+                        ; (default: "0") A fatal response is any permanent
+                        ; failure (non-temporary 4xx, 5xx, 6xx) response
+                        ; received from the registrar. NOTE - if also set
+                        ; the 'forbidden_retry_interval' takes precedence
+                        ; over this one when a 403 is received. Also, if
+                        ; 'auth_rejection_permanent' equals 'yes' a 401 and
+                        ; 407 become subject to this retry interval.
 ;server_uri=    ; SIP URI of the server to register against (default: "")
 ;transport=     ; Transport used for outbound authentication (default: "")
 ;type=  ; Must be of type registration (default: "")

contrib/ast-db-manage/config/versions/28ce1e718f05_add_fatal_response_interval.py

+"""add fatal_response_interval
+
+Revision ID: 28ce1e718f05
+Revises: 154177371065
+Create Date: 2015-10-20 17:57:45.560585
+
+"""
+
+# revision identifiers, used by Alembic.
+revision = '28ce1e718f05'
+down_revision = '154177371065'
+
+from alembic import op
+import sqlalchemy as sa
+
+
+def upgrade():
+    op.add_column('ps_registrations', sa.Column('fatal_retry_interval', sa.Integer))
+
+
+def downgrade():
+    op.drop_column('ps_registrations', 'fatal_retry_interval')

res/res_pjsip_outbound_registration.c

@@ -100,6 +100,21 @@
 						buggy registrars.
 					</para></description>
 				</configOption>
+				<configOption name="fatal_retry_interval" default="0">
+					<synopsis>Interval used when receiving a Fatal response.</synopsis>
+					<description><para>
+						If a fatal response is received, chan_pjsip will wait
+						<replaceable>fatal_retry_interval</replaceable> seconds before
+						attempting registration again. If 0 is specified, chan_pjsip will not
+						retry after receiving a fatal (non-temporary 4xx, 5xx, 6xx) response.
+						Setting this to a non-zero value may go against a "SHOULD NOT" in RFC3261,
+						but can be used to work around buggy registrars.</para>
+						<note><para>if also set the <replaceable>forbidden_retry_interval</replaceable>
+						takes precedence over this one when a 403 is received.
+						Also, if <replaceable>auth_rejection_permanent</replaceable> equals 'yes' then
+						a 401 and 407 become subject to this retry interval.</para></note>
+					</description>
+				</configOption>
 				<configOption name="server_uri">
 					<synopsis>SIP URI of the server to register against</synopsis>
 					<description><para>
@@ -277,6 +292,8 @@ struct sip_outbound_registration {
 	unsigned int retry_interval;
 	/*! \brief Interval at which retries should occur for permanent responses */
 	unsigned int forbidden_retry_interval;
+	/*! \brief Interval at which retries should occur for all permanent responses */
+	unsigned int fatal_retry_interval;
 	/*! \brief Treat authentication challenges that we cannot handle as permanent failures */
 	unsigned int auth_rejection_permanent;
 	/*! \brief Maximum number of retries permitted */
@@ -312,6 +329,8 @@ struct sip_outbound_registration_client_state {
 	unsigned int retry_interval;
 	/*! \brief Interval at which retries should occur for permanent responses */
 	unsigned int forbidden_retry_interval;
+	/*! \brief Interval at which retries should occur for all permanent responses */
+	unsigned int fatal_retry_interval;
 	/*! \brief Treat authentication challenges that we cannot handle as permanent failures */
 	unsigned int auth_rejection_permanent;
 	/*! \brief Determines whether SIP Path support should be advertised */
@@ -799,6 +818,14 @@ static int handle_registration_response(void *data)
 			schedule_registration(response->client_state, response->client_state->forbidden_retry_interval);
 			ast_log(LOG_WARNING, "403 Forbidden fatal response received from '%s' on registration attempt to '%s', retrying in '%u' seconds\n",
 				server_uri, client_uri, response->client_state->forbidden_retry_interval);
+		} else if (response->client_state->fatal_retry_interval
+			   && response->client_state->retries < response->client_state->max_retries) {
+			/* Some kind of fatal failure response received, so retry according to configured interval */
+			response->client_state->status = SIP_REGISTRATION_REJECTED_TEMPORARY;
+			response->client_state->retries++;
+			schedule_registration(response->client_state, response->client_state->fatal_retry_interval);
+			ast_log(LOG_WARNING, "'%d' fatal response received from '%s' on registration attempt to '%s', retrying in '%u' seconds\n",
+				response->code, server_uri, client_uri, response->client_state->fatal_retry_interval);
 		} else {
 			/* Finally if there's no hope of registering give up */
 			response->client_state->status = SIP_REGISTRATION_REJECTED_PERMANENT;
@@ -1186,6 +1213,7 @@ static int sip_outbound_registration_perform(void *data)
 	}
 	state->client_state->retry_interval = registration->retry_interval;
 	state->client_state->forbidden_retry_interval = registration->forbidden_retry_interval;
+	state->client_state->fatal_retry_interval = registration->fatal_retry_interval;
 	state->client_state->max_retries = registration->max_retries;
 	state->client_state->retries = 0;
 	state->client_state->support_path = registration->support_path;
@@ -1909,6 +1937,7 @@ static int load_module(void)
 	ast_sorcery_object_field_register(ast_sip_get_sorcery(), "registration", "expiration", "3600", OPT_UINT_T, 0, FLDSET(struct sip_outbound_registration, expiration));
 	ast_sorcery_object_field_register(ast_sip_get_sorcery(), "registration", "retry_interval", "60", OPT_UINT_T, 0, FLDSET(struct sip_outbound_registration, retry_interval));
 	ast_sorcery_object_field_register(ast_sip_get_sorcery(), "registration", "forbidden_retry_interval", "0", OPT_UINT_T, 0, FLDSET(struct sip_outbound_registration, forbidden_retry_interval));
+	ast_sorcery_object_field_register(ast_sip_get_sorcery(), "registration", "fatal_retry_interval", "0", OPT_UINT_T, 0, FLDSET(struct sip_outbound_registration, fatal_retry_interval));
 	ast_sorcery_object_field_register(ast_sip_get_sorcery(), "registration", "max_retries", "10", OPT_UINT_T, 0, FLDSET(struct sip_outbound_registration, max_retries));
 	ast_sorcery_object_field_register(ast_sip_get_sorcery(), "registration", "auth_rejection_permanent", "yes", OPT_BOOL_T, 1, FLDSET(struct sip_outbound_registration, auth_rejection_permanent));
 	ast_sorcery_object_field_register_custom(ast_sip_get_sorcery(), "registration", "outbound_auth", "", outbound_auth_handler, outbound_auths_to_str, outbound_auths_to_var_list, 0, 0);