[Home]

Summary:ASTERISK-10158: [patch] Dundi support for true dynamic peering (the '*' mapping)
Reporter:David Smith (dds)Labels:
Date Opened:2007-08-24 03:37:48Date Closed:2007-09-18 11:15:18
Priority:MajorRegression?No
Status:Closed/CompleteComponents:PBX/pbx_dundi
Versions:Frequency of
Occurrence
Related
Issues:
Environment:Attachments:( 0) dundi.conf
( 1) dyndundi.patch
( 2) output.txt
Description:The sample configuration file for DUNDi claims that you can specify a '*' peer as a special unspecified entity to match any peer. This is useful in situations where you have a dynamic server pool, such as when doing speech recognition processing on a grid, and you want to make use of DUNDi's ability to find a server to forward the call to based on the current load.

The implementation of this was non-existent; apparently the person who wrote the sample configuration file was under a different impression than the author(s) of pbx_dundi.c. I've implemented this support and included a patch in this bug. I would very much like this support to be included in the next release, as I believe others would do.

Thanks,

****** ADDITIONAL INFORMATION ******

--- asterisk-1.4.10~dfsg.orig/pbx/pbx_dundi.c
+++ asterisk-1.4.10~dfsg/pbx/pbx_dundi.c
@@ -267,6 +267,7 @@
static AST_LIST_HEAD_NOLOCK_STATIC(alltrans, dundi_transaction);

static int dundi_xmit(struct dundi_packet *pack);
+static struct dundi_peer *any_peer;

static void dundi_debug_output(const char *data)
{
@@ -486,7 +487,8 @@
if (!dundi_eid_cmp(&cur->eid,eid))
break;
}
-
+        if (!cur && any_peer)
+          cur = any_peer;
return cur;
}

@@ -1505,7 +1507,7 @@
unsigned char *bufcpy;
struct dundi_ie_data ied;
struct dundi_ies ies;
- struct dundi_peer *peer;
+ struct dundi_peer *peer = NULL;
char eid_str[20];
char eid_str2[20];
memset(&ied, 0, sizeof(ied));
@@ -1589,6 +1591,21 @@
case DUNDI_COMMAND_REGREQ:
/* A register request -- should only have one entity */
peer = find_peer(ies.eids[0]);
+                /** Dynamic Peering **/
+                /* if the peer is not found and we have a valid 'any_peer' setting */
+                if (peer == any_peer && any_peer != NULL)
+                  {
+                    /* copy any_peer into a new peer object */
+                    peer = ast_calloc(1, sizeof(*peer));
+                    memcpy(peer, any_peer, sizeof(*peer));
+                    /* set EID to remote EID */
+                    peer->eid = *ies.eids[0];
+                    /* add the peer the peer list */
+                    AST_LIST_LOCK(&peers);
+                    AST_LIST_INSERT_HEAD(&peers, peer, list);
+                    AST_LIST_UNLOCK(&peers);
+                  }
+                /** End Dynamic Peering **/
if (!peer || !peer->dynamic) {
dundi_ie_append_cause(&ied, DUNDI_IE_CAUSE, DUNDI_CAUSE_NOAUTH, NULL);
dundi_send(trans, DUNDI_COMMAND_REGRESPONSE, 0, 1, &ied);
@@ -1801,7 +1818,7 @@
dundi_send(trans, DUNDI_COMMAND_CANCEL, 0, 1, NULL);
break;
case DUNDI_COMMAND_ENCREJ:
- if ((ast_test_flag(trans, FLAG_SENDFULLKEY)) || AST_LIST_EMPTY(&trans->lasttrans) || !(peer = find_peer(&trans->them_eid))) {
+          if ((ast_test_flag(trans, FLAG_SENDFULLKEY)) || AST_LIST_EMPTY(&trans->lasttrans) || !(peer = find_peer(&trans->them_eid))) {
/* No really, it's over at this point */
if (!final)
dundi_send(trans, DUNDI_COMMAND_CANCEL, 0, 1, NULL);
@@ -4330,6 +4347,7 @@
dundi_ttl = DUNDI_DEFAULT_TTL;
dundi_cache_time = DUNDI_DEFAULT_CACHE_TIME;
cfg = ast_config_load(config_file);
+        any_peer = NULL;


if (!cfg) {
@@ -4453,8 +4471,14 @@
/* Entries */
if (!dundi_str_to_eid(&testeid, cat))
build_peer(&testeid, ast_variable_browse(cfg, cat), &globalpcmodel);
- else
- ast_log(LOG_NOTICE, "Ignoring invalid EID entry '%s'\n", cat);
+ else if (!strcasecmp(cat, "*"))
+                          /* any peer */
+                          {
+                            build_peer(&empty_eid, ast_variable_browse(cfg, cat), &globalpcmodel);
+                            any_peer = find_peer(NULL);
+                          }
+                        else
+                         ast_log(LOG_NOTICE, "Ignoring invalid EID entry '%s'\n", cat);
}
cat = ast_category_browse(cfg, cat);
}
Comments:By: Ronald Chan (loloski) 2007-09-07 21:58:01

Any news here please?
By: Russell Bryant (russell) 2007-09-11 10:43:45

In the future, please use tabs instead of spaces for indentation.  Also, take a look at doc/CODING-GUIDELINES for other coding standards.
By: Digium Subversion (svnbot) 2007-09-11 10:45:06

Repository: asterisk
Revision: 82250

------------------------------------------------------------------------
r82250 | russell | 2007-09-11 10:45:06 -0500 (Tue, 11 Sep 2007) | 4 lines

The sample dundi.conf claims support for a wildcard peer entry - [*], but the
code did not support it.  This patch makes it work.
(closes issue ASTERISK-10158, patch by dds, with some changes by me)

------------------------------------------------------------------------
By: Digium Subversion (svnbot) 2007-09-11 10:46:28

Repository: asterisk
Revision: 82251

------------------------------------------------------------------------
r82251 | russell | 2007-09-11 10:46:28 -0500 (Tue, 11 Sep 2007) | 12 lines

Merged revisions 82250 via svnmerge from
https://origsvn.digium.com/svn/asterisk/branches/1.4

........
r82250 | russell | 2007-09-11 11:03:42 -0500 (Tue, 11 Sep 2007) | 4 lines

The sample dundi.conf claims support for a wildcard peer entry - [*], but the
code did not support it.  This patch makes it work.
(closes issue ASTERISK-10158, patch by dds, with some changes by me)

........

------------------------------------------------------------------------
By: Michiel van Baak (mvanbaak) 2007-09-12 04:25:58

as tbsky noted in 0010703:
dundi.conf can not do "[*]" config, dds make a patch for it. i try it and it works well. but when using this setup under freePBX, asterisk often crash and core dump when press freePBX "reload" bar. so i think there maybe something wrong with it.
sorry i can not comment on that patch, since that bug is closed now.
By: Michiel van Baak (mvanbaak) 2007-09-12 04:26:52

Can you provide your dundi.conf and a backtrace ?
By: tbsky (tbsky) 2007-09-18 03:07:55

hi:
  sorry. i just learn how to backtrace. i upload my dundi.conf and backtrace output.txt.
  thanks for help!
By: David Smith (dds) 2007-09-18 03:26:31

Would a decent enough solution to make it so that in destroy_peer, the peer->permit and peer->include are only free()d if they are different then the any_peer->permit and any_peer->include, respectively?

Thanks,
David
By: Digium Subversion (svnbot) 2007-09-18 11:12:14

Repository: asterisk
Revision: 82802

------------------------------------------------------------------------
r82802 | russell | 2007-09-18 11:12:14 -0500 (Tue, 18 Sep 2007) | 4 lines

When copying the contents from the wildcard peer, do a deep copy instead of
shallow copy so that it doesn't crash when beging destroyed.
(closes issue ASTERISK-10158, patch by me)

------------------------------------------------------------------------
By: Digium Subversion (svnbot) 2007-09-18 11:15:18

Repository: asterisk
Revision: 82803

------------------------------------------------------------------------
r82803 | russell | 2007-09-18 11:15:18 -0500 (Tue, 18 Sep 2007) | 12 lines

Merged revisions 82802 via svnmerge from
https://origsvn.digium.com/svn/asterisk/branches/1.4

........
r82802 | russell | 2007-09-18 11:31:01 -0500 (Tue, 18 Sep 2007) | 4 lines

When copying the contents from the wildcard peer, do a deep copy instead of
shallow copy so that it doesn't crash when beging destroyed.
(closes issue ASTERISK-10158, patch by me)

........

------------------------------------------------------------------------