PRODUCT_LINE: BCM ============================================ TITLE ===== BCM370.248-FEPS ======================== Release: 3.7 Issue: N/A Category: GEN Superceded By: NONE Release Date: 20071128 Patch Conflict(s): N/A Special Instructions: NO Patch Version: N/A ======================== Software Update Name: BCM370.248-FEPS Applicable H/W Platforms: BCM1000, BCM200, BCM400, SRG200 1.0, SRG400 1.0 Applicable S/W Platforms: BCM 3.7, SRG 1.0 based on BCM 3.7 Category: GEN Installation Recommendation: This patch should be applied to all new installs of BCM 3.7 and SRG 1.0. Existing installations should have this update applied if they are experiencing any of the issues that are corrected. There is no need to schedule specific maintenance to apply this upgrade if there are no reported issues. In those cases, systems should be updated at the next regular service opportunity. Component & Version: FEPS 37.260.0.33 rv32h323.dll (ver 4.0.0.48g) Dependencies: Required patches BCM370.183-CORE or later The core does not need to be applied before this patch, but the core must be applied for this patch to work correctly. Product Dependencies: None Size: ~1.5 MB System Impact: Time to apply - approximately 20 minutes Does patch application force reboot: Yes, 2 reboots Other: 1. VoIP Trunks will be unusable during the application of the patch. 2 . If it is suspected that a UDP attack has already occurred on the BCM, then the BCM should be rebooted prior to applying this patch in order for the patch to apply properly. 3. During the BCM reboots, IP sets will deregister from the BCM. Limitations: None Patch Removable: No Description: ----------- The following issues are addressed: 1. FEPS UDP thread corruption upon receiving non-MCDN packets on port 15000 The thread that processes UDP packets received by FEPS made the assumption that any packets received on port 15000 contained MCDN data. Upon receiving non-MCDN packets on this port, memory corruption would then occur causing a core file to be generated by that thread. Q01753820 This patch includes the content of the following superseded patches: BCM370.241-FEPS --------------- 1. VoIP calls fail intermittently - MGS Transport mismatch. When attempting to resume media path before the far end had closed its own media channels, the BCM would end up using the wrong media transport address resulting in the VoIP call to fail and an MGS transport alarm to be generated. Q01663827 2. Unnecessary codec renegotiation at beginning of VoIP Call. Unnecessary codec renegotiation after the initial capability exchange, could result in temporary interruption of ringback, or delay in speech path establishment. Q01677775-04 3. VoIP Gateway crash upon receiving GatekeeperReject. The VoIP Gateway was generating an exception as a result of not correctly handling certain GatekeeperReject reasons. Q01592829-04 BCM370.229-FEPS --------------- 1. VoIP Gateway crash VoIP Gateway crash upon receiving gatekeeper admission confirmation after call release. Q01524349 2. VoIP lines locking up/not releasing As a result of media path not being disconnected correctly, VoIP trunks would lockup. Q01512048 3. No speech path upon retrieving parked call, or taking call off hold. The VoIP Gateway was not correctly triggering renegotiation in the event of a payload size change during the call. Q01540598 BCM370.206-FEPS --------------- 1. "Force G711 for 3.1kHz Audio" feature does not function correctly The "Force G.711 for 3.1KHz Audio" feature would only be activated if T.38 Fax Support were also activated. Instead, these two features are actually mutually exclusive in that if faxes are to be sent over G.711 using the "Force G.711 for 3.1 KHz Audio" feature, then T.38 Fax Support feature should be disabled. If both features are enabled, T.38 will take precedence. Q01491995 BCM370.175-FEPS --------------- 1. Gatekeeper Deregistration Problem Upon losing connectivity to the primary gatekeeper, the BCM attempts to switch to the backup gatekeeper. In the case in which the BCM network configuration is not valid for the BCM to be able to register with the backup gatekeeper, the VoIP Gateway enters a state in which it continually attempts to discover the backup gatekeeper every 10 seconds, but with each attempt failing. At this point, two problems are exposed: 1) the VoIP gateway fails to switch back to the primary after 3 of these type of failures, and 2) the VoIP Gateway fails to close the RAS transactions in each of these failure scenarios. As a result, after the Radvision configured max RAS out transactions (350), the VoIP Gateway runs out of RAS transaction resources, and must be restarted in order to recover. CR Q01404705 2. BCM VoIP Gateway memory leak GKACallback structure not freed when running in Direct call signaling mode resulting in memory leak. Q01406115-01 3. Third Party H.323 Gateway Interop results in inability to send DTMF tones to IVR In this scenario a user on the BCM makes an outbound call to a third party H323 VoIP Gateway which is the interface to an IVR, and because this gateway sends back to the BCM a Progress H.225 message instead of an Alerting and does not connect the call, the BCM will not be able to send DTMF tones to the IVR. The Acme Packet Session Border Controller is an example of such a third party gateway with which this problem has been seen. Q01379114 BCM370.149-FEPS --------------- 1. Incorrect Release Reasons and Tones Calls to CS1000 sets that are busy can result in the caller receiving an undecipherable release reason displayed on their set, for example "Unknown 17", which should be "user busy". There will be no busy tone. The BCM Monitor UIP trace for MCDN over IP will show a NOTIFY message being received from the CS1000 side immediately before the DISCONNECT. CR Q01316141 2. Intermittent Fax Failure on calls with CS1000 and BCM 4.0 Fax calls can fail intermittently due to T38 channels not being opened. The T38 channels are not opened because the BCM does not always advertise the T38 capability in its capability set. One scenario in which this can commonly occur is if the first VoIP call after a BCM reboot is a fax attempt. The issue has been resolved by ensuring that the T38 capability is not erroneously omitted from the capability set. This issue has been fixed by ensuring T38 channels are opened. CR Q01379032 BCM_370.134_FEPS.33.372000 -------------------------- 1. Issue Fixed: Intermittent VoIP calls will drop soon after initiation. Intermittent VoIP calls will drop from the user's perspective for no reason. These calls will drop soon after initiation. This is an extremely rare problem. If however a BCM suffers from this problem the symptoms can occur regularly. CR Q01286946 2. Issue Fixed: Intermittent Inbound Trunk Lockup On rare occasions when an inbound VoIP trunk call is answered, a trunk can lockup. This can occur on sites with a small number of VoIP trunks, i.e. on trunks that will be used for both inbound and outbound calls. When the trunk locks up the first inbound call will fail and so will the next outbound call. No further calls can be made on this trunk. CR Q01232655, Q01282054 3. Issue Fixed: Overloaded GateKeeper Current BCM timeout for GateKeeper requests is 2 seconds, the H225 standard (a part of the H323 standard family) recommends 3 seconds so the BCM timeout has been extended. This helps alleviate congestion at the GateKeeper when many BCMs send registration requests that may timeout. CR Q01291957 4. Issue Fixed: Intermittent VoIP Trunk Lockup VoIP trunks can lockup. MCDN must be enabled for problem to occur, problem has been found where outbound calls of the following type exist: BCM---VOIP--->Meridian/Succession---PRI--->PSTN. No further calls can be made on the locked up trunks. CR Q01282054, Q01281447, Q01120093, Q01290752 BCM_370.126_FEPS.33.371900 -------------------------- 1. Issue Fixed: Intermittent VoIP calls will drop soon after initiation. Intermittent VoIP calls will drop from the user's perspective for no reason. These calls will drop soon after initiation. This is an extremely rare problem. If however a BCM suffers from this problem it can occur regularly. CR Q01286946 2. Issue Fixed: Intermittent Inbound Trunk Lockup On rare occasions when an inbound VoIP trunk call is answered, a trunk can lockup. Can occur on sites with a small number of VoIP trunks, i.e. on trunks that will be used for both inbound and outbound calls. When trunk locks up the first inbound call will fail and so will the next outbound call. No further calls can be made on this trunk. CR Q01232655, Q01282054 3. Issue Fixed: Overloaded GateKeeper Current BCM timeout for GateKeeper requests is 2 seconds, the H225 standard (a part of the H323 standard family) recommends 3 seconds so the BCM timeout has been extended. This helps alleviate congestion at the GateKeeper when many BCMs send registration requests that may timeout. CR Q01291957 BCM_370.103_FEPS.33.371700 -------------------------- 1. Issue Fixed: Intermittent VoIP Trunk Lockup VoIP trunks can lockup. MCDN must be enabled for problem to occur, problem has been found where outbound calls of the following type exist: BCM---VOIP--->Meridian/Succession---PRI--->PSTN. No further calls can be made on this trunk. CR Q01282054, Q01281447, Q01120093, Q01290752 BCM_370.073_FEPS.33.371600 --------------------------- 1. Issue Fixed: One Way Voice Path on VoIP trunks. One way voice path can occur when BCM acts as tandem node and a blind call transfer occurs during which the transferring party transfers soon after or before ringback is received. Can occur with regular non-transfer calls as well. In either case, this will only occur if FEPS version 130 or greater was applied. CR Q01145202 2. Issue Fixed: Dropped Calls With Reason 'Codec Mismatch' Calls to CS1000 can drop with reason codec mismatch; however it is not a codec mismatch. Call will drop immediately after initiation. CR Q01211403 3. Issue Fixed: VoIP Gateway crash on loss of connection to GateKeeper When connection to GateKeeper is lost VoIP Gateway can crash. The VoIP Gateway will restart automatically, but active calls will be dropped. This problem is extremely rare, since a user must make a call within the 25s window during which the VoIP Gateway has not determined that connection to GateKeeper was lost, even then, a crash is very rare. Q01212372 - (formerly tracked by Q01210132) BCM_370.051_FEPS.33.371500 -------------------------- Issue Fixed: VoIP Call Drop when a switch from Primary to Backup GateKeeper occurs Q01095627 Issue Fixed: SN07Plus: H.323 GateKeeper to GateKeeper T38 Interoperability: BCM T38 FAX calls fail Q01011670 Issue fixed: Load test: VoIP trunk lockups resolved through upgrade of FEPS to a version 150 or greater and core to version 6.15 or greater. Q01121345 Issue Fixed: CS2000-BCM Interoperability: Incoming Private/Subscriber VoIP calls are not properly handled on BCM. Q01086112 Issue Fixed: On an incoming call that has CLID the BCM still displays CLI on set. Q01105644 Issue Fixed: Failure to initialize H.323 protocol stack Q01088391 Issue Fixed: BCM erroneously switches to backup GateKeeper '0.0.0.0' and then never switches back to primary GateKeeper. Q01099819 Issue Fixed: CS1000 Interoperability: No stutter tone heard on BCM->Succession call when using Succession Authentication Code feature. Q01080871 Issue Fixed: No User Busy reason displayed or tone heard when BCM calls a busy set. Q01084701 Issue Fixed: No voice path on external call forward. Q01042953 Issue Fixed: Tenor GW Interoperability: One way speech path. Q01038212 Issue Fixed: SN07 Interoperability: BCM T.38 Calls fail with CS2000. Q01011670 Issue Fixed: One way speech path, and calls dropping. Q01006425