Release Notes for P103 Patch Releases

Tenor Gateways and Call Relays - AS/AF/AX/DX/BX/CMS/CR60/CRSP

A software patch is a release of software that has been created by Quintum to resolve an inconsistency or 'bug' that has been identified within a previous release of Generally Available (GA) software or patch. They may also add features. Patch software does not go through any testing from Quintum, but has been verified by the customer that the identified the bug is resolved with this patch.

Patches (for Second Generation Tenors) can be identified by the 3rd set of digits in the release name as PX-Y-Z. Where the Z is the patch number. All patches will be built on top of the latest GA release of software. For example, P102-11-00 is the latest GA software, then the first patch for this software will be P102-11-01.

If additional patches are required for a specific GA release, each new patch will contain all previous patch fixes. For example, the patch P100-11-02 will include all patches from P100-11-01.

On a scheduled basis, Quintum will release a full GA version that will include all previous patches and will have been put through our full system tests.

We hope that this process will aid our customers by providing fixes in a more rapid manner.

CMS COMPATIBILITY NOTE:

P101 and LATER CODE ON CMS ONLY SUPPORTS SERIES 2 CONTROLLER AND PERIPHERAL CARDS!

See this document for more details:

http://www.quintum.com/support/products/2G/cms/sysdoc/CMS_series2_notice.pdf

The notes below are cumulative since the last GA Release, P103-08-00. The most recent changes are in the top section, the history is similarly in most recent first order.

P103-08-14

2450 Analog MFG Noise Test - NOT RUN AXM1600

Analog "noise tests" were not being run on AXM1600. This has been resolved.

2465 Root Process crash

Under certain circumstances the "root" process was crashing. It has been resolved.

2856 & 2954 new DAA chip support

Changes and tweeks for new DAA chip.

2662 AF/AS/AX/BX/DX: Slow Start Call with codec mismatch may cause crash

On H.245 "slow start" calls, when the incoming SETUP contained an H245Control list that has a TCS with no matching codecs it may crash the system. This has been resolved.

2881 CMS crashes when it runs out of memory buffers

In some cases there were not enough memory buffers assigned on the CMS. A crash could result. This has been resolved.

2950 AF; show -v

Various fixes and changes to the show -v output for the Tenor AF.

3006 new framer driver for Infineon version 2.2

Support for new Infineon framer chip

3036 AFT200 & AFT400 - channels offline alarms

AFT200 and AFT400 were showing "channel offline" alarms in error. This has been resolved.

3071 New serial number prefixes for RoHS boards

Serial number prefixes were incremented for RoHS compliant systems.

3099 RoHS support for CMS

Support for new RoHS boards for the CMS.

3186 P103 Only: SIP Payload size is not checked for notify messages

In very rare circumstances it is possible for a improperly encoded NOTIFY message to crash the system. A check was added to prevent this.

3193 Tenor watchdogs when radius receives corrupted access accept

Tenor could crash on certain corrupted access-accept messages.

P103-08-12

This release is for Call Relay 60 only. All other platforms should use P103-08-11.

2805 Call Relay 60 SIP configuration not saved

CR60 only. Configuration for the SIP User Agent was not being saved properly. This problem has been resolved.

P103-08-11

2773 CMS,CR-SP; Some config changes cause eventual crash

CMS/CRSP only. There were cases in P103 code where making config changes to Static routes, IPRG, Codec Profiles and some other objects could cause the system to crash upon submit. Common symptoms include: CH, Telnet and "watchdog without logging" crashes.

This has been resolved.

2398 Remote call id is not parsed properly

When the remote callID (callID on the remote Quintum GW) had a value greater than the max positive integer value (0x7FFFFFFF) , it was not parsed properly from Non Standard data. As a result, the same remote call ID was reported for all the calls. This has been resolved.

2588 CH task disappears in CRSP

In cases where the Tenor was receiving two calls with the same SIP callid, the CH process was dying. The crash occurred while the later call was failing and Tenor was freeing up the earlier call. The SIP call id should be globally unique, but tenor should not crash. The change adds a check for CH callid along with SIP callid to find the correct call object.

2589 SIP ACK message failure

During internal testing it was found the Tenor sent a malformed ACK message when the far end responds with 401 Unauthorized and includes a Record-Route header with the first route containing an LR parameter, and no Contact header.

The fix is to be a little more flexible and if the contact uri is not present use instead the first entry in the route list in the ACK request.

2792/2658 OutboundCallDetection feature problems

Fixes to the OutboundCallDetection feature set. Those using OCD should use P103-08-11 rather than earlier versions of code.

2695 SrcCallSigAddress has Port value of 0 instead of 1720

There were situations where the outgoing H323 SETUP messages show the Source Call Signal Address having its Q931 Port=0 instead of 1720. These situations have been eliminated and the problem resolved.

2770 Transferred call was being aborted before Connected

There were situations where an Unattended transfer was aborted prematurely. This has been resolved.

2779 CMS;CR-SP; The IrDA receiver on the CMS should be turned off

CMS/CRSP only. It was found that random conditions (such as flourescent lighting) could trigger the IRDA port to generate random characters as if there were entered from the console port. The IRDA port has been disabled in this release.

2803 StaticChannelConnection disappeared after CMS reboot

The StaticChannelConnection parameter was not being saved into flash. This only affects those deploying the SS7 system.

2812 ALL; Being able to reset unit from CLI and FTP reliably

There were certain situations where a system in an unstable state may not be able to reliably reboot via the CLI and/or FTP. The reset code has been simplified to make it more robust and less prone to failure.

2823 CR-60 Only: RFC2833 digits coming in for H323 outbound calls causes reset

There were situations where RFC2833 digits came in from the SIP side, and went out to an H.323 call, that could cause the CR-60 to reset. This has been resolved. Call Relay 60 only (Call Relay SP was not affected).

P103-08-10

2376 Codec Profile attachment incorrect after reboot

There were several cases where codec information in the Codec Profile was lost after a reboot. These situations have been addressed and the problem resolved.

2400 Voice Codec changes after reset

There were several cases where codec information in the Voice Codec definition was lost after a reboot. These situations have been addressed and the problem resolved.

2602 Slot 2 down in PPPoE mode

There were several cases where slot 2 (line cards) would go down when PPPoE was enabled. This has been resolved.

2626 ISDNSignalingGroup-1 and cassg-phone/line gets reattached to ChannelGroup after Reset.

There were several cases where if a Signaling Group was detached from a channel group, it would be re-attached on reset. This has been resolved.

P103-08-09

2574 Minimum Dial Digits resets to 7

If mindigits in the dialplan prompt was set to greater than 7, it would be reset to 7 on a reboot. This has been resolved.

2546 Need to use interdigit timer, as opposed to critical timer, on R2

In certain situations an R2 switch may not properly send a termination character (usually #) at the end of the DNIS the system may wait 20 seconds to process the call. This now uses the configurable interdigit timer (default 4 seconds, but configurable down to 1 second) instead of the hardcoded 20 second critical timer. This only was a problem with R2, other protocols were fine. R2 now behaves like the other protocols in this regard.

2571 Local ring back ALWAYS provided independent of ProgressTone setting

A phone to PSTN circuit-switched call via an analog Tenor, and no matter how the ProgressTone parameter was set in the TCRG and LCRG, CH always attempted to provide local ring back tone. This tone wass disabled very quickly (in about 100 msec). However, this causes a clicking noise that is annoying to the caller.

This has been resolved.

2580 Crash UPDP table in memory

UPDP table in memory could get corrupted when any dialplan DB item is changed. When the Tenor boots up, the table is good, but after db change, it may get corrupted, forcing a reboot.

This has been resolved.

2582 Busyout feature does not work in case of new passthru type Quintum

When one FXS is in conversation to another FXS or IP call, another call incoming from corresponding FXO port (via passthrough). With "busy out = 2", the calling party of the incoming call to FXO would be busy tone, but there was continuous ringing.

This has been resolved.

2479 DASS2 does not filter "X" from the Calling Party number

In DASS2 there can be an "X" character in the Calling Party field from the telco. This was not getting filtered and processed, causing invalid CLID to be relayed. This has now been resolved, the Calling Party is now properly filtered in DASS2.

2555 CallerIDType does not work when DNChannelMap is used

Calleridtype was not being used when DNChannelMap was being employed. This caused improper delivery of CallerID using certain calleridtypes. This has now been resolved.

2566 Var config for stopping SIP Re-INvites sent by Call Relay

Some SIP devices do not like having periodic RE-INVITEs being sent to them, and will disconnect the call if they are received.

As a temporary solution, a var_config variable has been added to allow this to be configurable:

EnableCRReinvite 1 enables periodic reinvites, and is the default
EnableCRReinvite 0 disables periodic reinvites and must be explicitly set for these cases.

A better solution will be introduced in P104 code, this variable set to 0 will solve the problem for now.

P103-08-08

2505 Manufacturing Test enhancements for small Tenors (MFG and DMT)

Several enhancements for pre-shipment manufacturing tests.

2549 AX: Support for new motherboard (ID 0110)

Adds support for a new AX motherboard spin coming in a few months.

2538 FXS to FXS call routing support for DN channel map

DN channel map typically uses single LCRG. CH did not allow calls to be routed on the same LCRG, therefore FXS to FXS calls did not work with DN channel map. This bug fix addresses this issue.

2343 Tenor occasionally can't maintain SIP registration with certain proxies

There were situations, especially noted with Vonage and Asterisk where the Tenor loses registration state. It could get into this state when the Proxy failed several authentications in a row. This seems to occur only in specific circumstances, usually with heavily loaded (or underpowered) proxies.

The process has been made more robust and tolerant of these failures, so this is resolved.

2527 Tenor continuously re-attempts SIP Registration (every ~100ms) upon receipt of Authentication Failure (401 - Unauthorized)

Tenor could get into a loop after receiving a 401 unauthorized to a REGISTER attempt. It would cause continuous, and frequent, re-attempts. Code has been introduced to eliminate this situation.

2535 Tenors Running P103-08-05 would drop calls after a few seconds with certain proxies

This was an unintended artifact of bug 2347 (introduced in P103-08-02). These calls were being mistaken for parallel forked calls and cancelled.

Bug fix for 2347 has been backed out, a better solution will be introduced in P104 code.

P103-08-07

2530 Call Waiting with DNChannel Map broken with new SIPSG scheme

Call Waiting did not always work properly when DNCHannelMap was in use. It would sometimes deliver the second call to the wrong channel. It now works properly.

2531 Error for some DB configuration

In some circumstances a manually created text db file could cause the system to constantly reboot after a reboot. This has been resolved.

P103-08-06

2246 Gateway description is not updated to RMSS server until box is rebooted

When RMSS server protocol was UDP, the Gateway->Description was not updated to RNMS server until box was rebooted. RMSS agent now reads this variable for every databaes update, so a reboot is no longer required to update this value.

2424 NEW problem while fixing this bug: Cause code is converted from 16 to 41. - REFIXED

As noted in earlier updates on this bug, there were still some cases where cause codes were getting converted to 41 in error on RADIUS stop accounting records. These have now been resolved.

Note: this addresses the known issues in AccountingType 5. AccountngType 5 should now be usable.

2427 Negative call duration in ivr call

Due to rounding, it was possible that the reported time in stop accounting could be slightly higher than the allowed credit-time. This would cause RADIUS servers to get a negative balance, causing problems on some servers.

The rounding algorithm was adjusted to avoid this. Also note, session time in Stop Accounting messages is now rounded to the nearest second, not always rounded down as it was in the past.

2507 Support for FTP via RMSS

The RMSS now supports ftp over a tunneled management session. This allows for firmware updates, as well as log collection. The method for doing this is documented in the RMSS documentation.

P103-08-05

Important note: AccountingType 5, introduced in P103-08-01 (bug 2424) is not working properly, causing billing issues. We expect to have a patch to resolve this shortly. For now, please do not use accountingtype 5! Watch this site for a patch to resolve these issues.

2398 Remote call id is not parsed properly

As a result, same remote call ID is reported for all the calls.

2424 Cause Code converted from 16 to 41 in error

Bug 2424 (in P103-08-01) has issues affecting all accountingtypes in RADIUS. Cause code was converted from 16 to 41 when a call is disconnected by termination side first. This leads to billing problems. This has been corrected and should no longer occur.

2501 CH crashes when submit is done when calls are in connecting stage

System may crash when a submit is executed and calls are in connecting stage. This should no longer occur.

2502 Tenors have been seen to hang during bootup--tNetTask dies

Ocasionally a Tenor will hang on bootup. Seems to happen most often immediately after a software update, but it is not restricted to that case. This has been resolved.

2503 RTP Packets from Asterisk not being processed. Bug 2428 fix introduced this problem

Bug 2428 fixed problems with certain fragmented UDP packets not being recognized by the Tenor. This fix broke RTP with an Linux-based Asterisk IP PBX, causing one-way voice. This has been resolved.

P103-08-04

2452 SIP/PSTN Interworking change is effecting passthru calls

A certain combination of events could prevent ringback from being heard on a passthrough call. This has been resolved.

2463 Duplicate pattern entries in several tables not allowed

In P103-08-00 through P103-08-02 a restriction was put in not allowing duplicate entries in multiple Hopoff number directories, DNIS translation directories, Bypass number directories, and callerid translation directories.

This restriction would not allow duplicates to be entered, and would delete them upon migration from earlier versions.

This restriction has been removed.

P103-08-03

2319 SIP Crash during Attended Transfer

It was possible for the Tenor to crash during an attended transfer, when call waiting was enabled and a call came in. Similar to 2454 below, same fix applied. This is now fixed.

2356 CANCEL was ignored by SIP server

Note: This fix was in P103-08-01, not in P103-08-02 due to a build error. This has been resolved and this fix should be in all future builds. See in-depth description below under the P103-08-01.

2428 Quintum unable to process some fragmented UDP packets

Certain large UDP packets, that were fragmented, were not being recognized and processed. Not all fragmented packets were affected, but packets fragmented by certain IP devices were having problems.

This has been resolved.

2432 HDLC driver stop sending after power up (CMS ONLY)

Under certain conditions a CMS board may not send any packets. This will be very obvious by doing "status" at the appropriate DI prompt. You will see layer 1 up, and layer 2/3 waiting for establishment. RX counters will be increasing, TX counters will not.

This can often be resolved by reboot. A much better fix has been implemented, the problem has been resolved.

2435 Need to support some punctuation in password fields

There was a restriction not allowing certain punctuation in PPPOEPassword field. This caused some problems where these characters are used. " % & < and > are now allowed in the PPPOEPassword field.

For now, this must be done through the CLI. Configuration Manager still has this restriction. It will be removed in the next GA version.

2449 SIP: Call through Secondary proxy fails

If Tenor fails to route a call through Primary Proxy and attempts to route the call through Secondary Proxy, the call always fails. Even upon receiving of 100, 183 and 180 through Secondary Proxy, retransmit timer expires and we retransmit INVITE.

This has been resolved.

2454 SIP CallWaiting: Incoming call while off hook after release causes Crash

A crash could occur if a user is off-hook, but the call has been terminated (far end disconnected) and a call-waiting signal is sent.

This only occurred in this fairly rare situation, and only if call waiting was enabled.

This has been resolved.

2457 No audio when put-thru call is made

If a put-through call was made, the call had no audio. It has been fixed, these calls should now work.

See P103-08-00 GA Release Notes for more information on the put-through feature.

2466 Crashes and hangs with CDRServer enabled

Crashes and system hangs were reported with Tenors using the CDRs (not RADIUS). This has been resolved.

2467 Facility msg w/startH245 being sent w/o H245Address

In certain situations H.323 Facility messages with starth245 were sent without a h245 address. Calls to some devices, such as Nortel CSE1000 will fail. This problem has been corrected.

P103-08-02

2425 ARJ from the configured GK resulted in incorrect attributes ine Stop-Accounting

For Accounting Type 2 and 4, where a Stop Accounting is sent for every route attempt, the Stop Accounting when an ARJ was received from the configured Gatekeeper had incorrect values.

Here are the affected attributes and their corrected values:

• Tenor-NAS-Port(VSA 2) = ip address of the GK from which ARJ is received.
• h323-call-type(VSA 27) = VOIP
• Quintum-Trunkid-Out(VSA 231) = ip address of the GK from which ARJ is received.
• h323-remote-address ( VSA 23) = ip address of the GK from which ARJ is received.

If the rejecting GK is the Tenor embedded GK, the address will show as "0.0.0.0"

2347 On outgoing Invite, accept response from 1st destination only if multiple are received

Tenor does not currently support SIP parallel forking. If we received multiple routes we attempted to do it, and failed, often resulting in a crash.

Tenor now only acts upon the first route received, ignoring other routes.

2436 DN Channel Map and One to One passthrough features do not work for DASS2

The DNChannelMap and related one-to-one passthrough mechanisms did not support DASS2 trunks. It now does.

P103-08-01

2162 Route mfg test output to event log

MFG and DMT tests now puts any errors from those tests in the event log as an exception (excp).

2343 Tenor occasionally can't authenticate registration

There were problems with Tenor occasionally losing it's registered status with the proxy. A reboot solved the problem.

This has now been fixed.

2356 CANCEL was ignored by SIP server

Some SIP proxies ignored our CANCEL message or send 481 response to CANCEL request, causing hung calls for some period (they would eventually timeout). The CANCEL message is rejected/dropped because of the presence of "tag" in the To: header. To resolve this the CANCEL now is sent without any Tag parameter in To: header.

2388 Some Databases when manually edited can crash the Tenors

There were conditions where if a user manually edited the configuration db text file, the tenor might crash. This has been resolved.

2396 Stopping local ringback when cut-thru is desired

There were times where the tenor generated local ringback, when it should have been passing through the voice path from PSTN. This has been resolved.

2399 System loses progresstonecountry config after reboot sometimes

There were situations where the setting for progresstonecountry would be lost on a reboot. This has been resolved.

2402 No dialplan (dialplancountry=255) does not work.

If one tried to disable the Dialplan by setting DialPlanCountry 255, the dialplan was still applied. This has been resolved.

2404 AX shows ??? as the line type in 'status ds1'

The command "st ds1" on an AX that has FXS ports and no FXO ports shows "???" for the FXS line type. This has been corrected.

2419 Tenor resetting on receipt of TCS w/no codecs

There were isolated cases where a tenor would receive a TCS with no codecs listed. This could crash the tenor. It no longer will.

2424 RADIUS; Generate Only One Stop-Accounting Per Leg

In P102 code we changed RADIUS behavior to generate a Stop Accounting for every outbound route attempt for a call with accountingtype set to 2 or 4. This was well received by most, but caused some RADIUS servers to behave badly.

To resolve this, we have now added accountingtype 5. This will send a Stop Accounting for each leg of the call as before, but only 1 for the outbound leg regardless of the number of routes attempted on the outbound leg.

See important note in P103-08-05 section above. For now, please do not use accountingtype 5.