[asterisk-users] Dozens of SIP NOTIFY messages with unique call ID's, and the same mailbox repeated multiple times on 1.6.2.6

[Sean Brady]

(sorry this is so long)

I could really use a helping hand.  I have a 1.6.2.6 installation using 
LDAP as the realtime engine for voicemail users, SIP users, queues, and 
some custom hotdesking families.  I'm also using ODBC voicemail storage.

The issue that I am having is that the UA's (Polycom 501's, 301's and 
430's) are receiving multiple SIP NOTIFY messages when the user has a 
new voicemail, to the point where it will crash the UA.  I have also 
noticed that the same mailbox is repeated in the CLI output several 
times (see below).

CLI output on "sip show peer Desk004" as of right now:
Mailbox      : 
2404 at gtf,2404 at gtf,2404 at gtf,2404 at gtf,2404 at gtf,2404 at gtf,2404 at gtf,2404 at gtf,2404 at gtf,2404 at gtf

The actual LDAP attribute on the peer is: 2404 at gtf

It appears- and I am trying to confirm- that the number of SIP NOTIFY 
messages is related to the number of extra mailbox entries.  I am using 
realtime cache on SIP peers, and I have qualify enabled.

When the peer first registers, there is only one mailbox entry, which 
matches the LDAP attribute.  I have noticed that if I check the mailbox 
using VoicemailMain, passing in the username and mailbox as arguments, 
the number of mailboxes in the Mailbox field on the "sip show peer 
Desk004" output increases from 1 to 3.  I just did a test where I 
restarted Asterisk, rebooted the phone, and it registered showing 1 
mailbox.  I checked the voicemail on the phone, then did a "sip show 
peer Desk004".  There are now 3 entries.  Every time that I check the 
voicemail, the number of entries increments by 2.

I also counted 28 retransmits of 102 NOTIFY messages with a MWI payload 
sent to the same peer.  It appears to send 28 NOTIFY messages, and 
retransmits each of them 4 times (I need to get a pcap of this, I've 
just been looking at the SIP debug on the peer).

I have tested this with several extensions, all exhibit the same 
behavior.  It appears that the UA is acting properly.  The same UA's 
with the same firmware and virtually the same configuration files work 
perfectly on 1.4.23.1 (the reg username and display name are different 
on 1.6.2.6).  I do not see this behavior on the same system with a peer 
specified in sip.conf.

I believe that this may be a bug with res_ldap in 1.6.2.6, however I 
don't think that I can rule out configuration issues until I pass it by 
the list first.  Is there something that I am doing wrong here, or is 
this a bug?


Thanks in advance for your help, it's greatly appreciated.  Feel free to 
contact off-list as well.

System:
RHEL 5.5 kernel 2.6.18-194.el5xen
Asterisk 1.6.2.6 built from source
DAHDI 2.3 built from source

LDAP packages managed by YUM:

mozldap.x86_64                                           6.0.5-1.el5
nss_ldap.i386                                               253-25.el5
nss_ldap.x86_64                                           253-25.el5
openldap.i386                                              2.3.43-12.el5
openldap.x86_64                                          2.3.43-12.el5
python-ldap.x86_64                                      2.2.0-2.1

LDAP Server:
389 Directory Server                                   1.2.5-1.el5