Check for an act on failures to update contacts during registration.
authorMark Michelson <mmichelson@digium.com>
Tue, 6 May 2014 17:47:20 +0000 (17:47 +0000)
committerMark Michelson <mmichelson@digium.com>
Tue, 6 May 2014 17:47:20 +0000 (17:47 +0000)
There was an underlying issue in a realtime backend where database updates
would fail. Since we were not checking for failure, we would end up in a
strange state where the old database entry was still present but Asterisk
thought that it had been updated. Now when an entry fails to update, we
print a warning and delete the old contact from sorcery so there is no
mismatch between foreground and backend state.

Patches:
res_pjsip_registrar.patch by John Hardin (License #6512)
........

Merged revisions 413358 from http://svn.asterisk.org/svn/asterisk/branches/12

git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@413359 65c4cc65-6c06-0410-ace0-fbb531ad65f3

res/res_pjsip_registrar.c

index 58f4b32..adfa408 100644 (file)
@@ -537,7 +537,12 @@ static int rx_task(void *data)
                                ast_string_field_set(contact_update, user_agent, user_agent);
                        }
 
-                       ast_sip_location_update_contact(contact_update);
+                       if (ast_sip_location_update_contact(contact_update)) {
+                               ast_log(LOG_ERROR, "Failed to update contact '%s' expiration time to %d seconds.\n",
+                                       contact->uri, expiration);
+                               ast_sorcery_delete(ast_sip_get_sorcery(), contact);
+                               continue;
+                       }
                        ast_debug(3, "Refreshed contact '%s' on AOR '%s' with new expiration of %d seconds\n",
                                contact_uri, aor_name, expiration);
                        ast_test_suite_event_notify("AOR_CONTACT_REFRESHED",