Verba Release Notes

Support for CUCM 10.0

Support the load balancing in proxy recording mode

Load balancing and failover for Lync Announcer services

Conference chat recording

Provisioning API Version 2.0

Auto-search integrated with automatic labeling framework

Summary

Call length mismatch due to silence suppression If there is no RTP sent in any direction since both side are on mute/do not talk Lync stops sending RTP (or at call start does not start sending RTP). These gaps between voice parts (where RTP is sent) are filled, but the gap (if there is any) between call start - first voice RTP and last voice RTP - call end cannot filled.

Impact

- this issue leads to losing silent segments, but not actual voice recording - it leads to time difference between presented recording time and actual recording time of media - mostly conference calls are affected, especially the first participant who is on mute until others arrive

Workaround

- there are currently no known workarounds

Summary

Using a very large AD might cause timeout issues in the AD sync process. An AD timeout can lead to unwanted reconfiguration of the recorded users when only a subset of the configured users are retrieved from the AD during the sync process due to timeout.

Impact

- if the AD query runs too long, it can lead to unwanted reconfiguration of the recorded users, which can result to loss of recordings - there is no indication or alert if timeout occurs

Workaround

- include (ObjectCategory=person) in the search filter to reduce search time - increasing the timeout value on the AD side might prevent this issue

Summary

There is one way audio in recordings after SIP re-invites when media bypass is active. The SfB/Lync Filter service does not recognize the SIP re-invite messages properly when media bypass is active, and the call is only partially rerouted through the proxy. This results in one way audio in the recordings after the first re-invite. The re-invite period is controlled by the session timer configured for the connection between the gateway and the SfB/Lync system.

Am I affected?

All 7.0 and later SfB/Lync recording deployments using proxy based recording are affected where media bypass is configured with SIP session timer.

Impact

The Verba SfB/Lync Filter does not recognize SIP re-invite messages properly when media bypass is active and the call is only partially rerouted through the proxy. This results in one way audio in the recordings after the first re-invite. The re-invite period is controlled by the session timer configured for the connection between the gateway and the SfB/Lync system. When the default SIP session timer setting (1800s) is used, the first re-invite is sent after 15 minutes. Thus all inbound or outbound calls longer than 15 minutes are affected. Prior to the re-invite, recordings contain both directions.

Workaround

Disabling the session timer or media bypass completely resolves the issue

Summary

Siren7 decoding problem is causing garbled decoding of voice in certain cases.

Am I affected?

All Lync/SfB recording deployments are affected.

Impact

- Intermittently causes garbled voice recording when Siren7 voice codec is used for the call - The recording quality is varying for the garbled recordings, from light impact to severe degradation of quality - Siren7 voice codec is mainly used for Lync 2010 Windows endpoints and Skype for Business 2015 IOS/Android devices when network is degraded - Siren7 voice codec is also used for Lync 2010/2013 and Skype for Business 2015 meetings during poor network conditions

Workaround

- currently there are no known workarounds

Summary

Lync / Skype for Business IM conversations might not be recorded after recording and processing 10,000 RTF based messages.

Am I affected?

All Lync / Skype for Business IM recording deployments are affected. This issue affects IM conversations which are using the RTF format. Lync 2013 or newer desktop clients are using the RTF format for P2P IM conversation when both participants are using a desktop client. Conferences, mobile, and consumer Skype conversations are not affected by this issue.

Impact

- When the instant message is transmitted using the RTF format, the Verba Lync / SfB IM Filter application (on the FE servers) can use all available Window handles due to the improper deallocation of the RTF parser. - The service stops processing RTF based instant messages after approx. 10,000 RTF messages (after all Windows handles are consumed) - No alert or notification sent when the issue occurs

Workaround

- The RTF message format can be disabled by a client policy, for more information see https://docs.microsoft.com/en-us/powershell/module/skype/set-csclientpolicy, DisableRTFIM

Summary

Certain calls between Skype for Business and Teams or Azure VoiceMail cannot be recorded

Am I affected?

All Sykpe for Business recording installations are affected where the recorded users can call Teams users or place voicemail messages in Azure VoiceMail.

Impact

Certain Skype for Business calls cannot be recorded when a recorded Skype for Business user is calling a Teams user and one of the participants is outside of the corporate network, or a recorded Skype for Business user is placing an Azure VoiceMail message. This limitation is caused by the new call setup procedure, and specifically in ICE negotiation, introduced in Teams and Azure VoiceMail, which prevents the recording system to redirect and force the calls to the Skype for Business Edge Server where the Media collector can fork the related media streams. Since the system is not able to capture the media streams related to these calls, these calls are not recorded. No alerts are raised unless CDR reconciliation is enabled.

Workaround

Currently there is no workaround other than disabling Teams or Azure VM calling entirely for the recorded users. We are actively working on implementing a new solution which extends the capabilities of the Proxy Server to be able to relay these type of calls too. It requires a major change in the architecture by allowing the Proxy Server to relay calls with external participants through a public interface. It also means that that calls which are currently routed through the Skype for Business Edge Server and forked by the Media Collector Service will be routed through the Proxy Servers that same way as calls with internal or PSTN participants. We are currently targeting July 2020 with the enhanced version of the Proxy Server.

Openssl updated to 1.0.1h

Cisco-GUID SIP header is now gathered for each call if present. Intended to use for UCCE/X integration

Optional SIP message reordering support

Lync application sharing sessions now generate call events to trigger desktop screen recorder

Load balancer will route all data of a call to the same recorder proxy to avoid duplicates

Call direction can be detected now based on user-agent type, in Lync environment PSTN call direction is detected based on Mediation server participant now by default

Recorder queue size, signaling and RTP wait times are now configurable

Optionally recorder can now detect participant roles (caller/callee) for BiB based recording

Calls to Hunt Group address displays the Agent Extension

JTAPI logging improvements

Automatic reconfiguration on the fly in Cisco IM recorder

Multiple Cisco IM&P servers can be configured on a single Recording Server to provide support for failover scenarios

Chat sessions are handled separately based on participants' point of view. If both participants are recorded the service creates separate recordings for each party.

More recording control options for Lync IM recording

GetCall Refresh interval is configurable now, configure in the jsp with the iniParams "refreshInterval", default is 2000 ms

P2P Announcement support is added, the filter can reroute calls to announcer service

Lync 2010 support is added to the RGS only recording mode

Recording announcement feature is added for incoming PSTN calls

Automatic provisioning support is added to the UCMA based announcement service

Proxy loadbalacing is added to the filter service. The filter supports multiple proxies, and the filter balance the load between them based on the network configuration

Option not to show Lync Trusted-UCMA users (announcements, music....) in conference participant lists

Lync Announcer changed from .NET 3.5 to .NET 4

Forwarded calls from non-recorded extensions to recorded extensions are now properly recorded in proxy mode

Recorder reliability improvement when the SQL Server is offline

Avaya Recorder configuration auto-refresh for extensions

CDR information is now sent to media receiver

Desktop agent mute function can now "mute voice only" or "mute voice and discard screen"

Firefox support for auto pause feature has been implemented

Desktop screen pop-up is now shown on all terminal sessions, when the user has more than one, screen capturing works only on the last session

New Active Directory Synchronization attribute: "User Matching ID". Use "ObjectGUID" in this field when the login name can be changed in the AD. If "User Matching ID" attribute is set up, Verba will match AD users with Verba user using this attribute.

New option in Active Directory Synchronization: "Screen Capture Enabled"

Terminology change: "Caller" changed to "From", "Called" changed to "To"

SearchCalls API call accepts new parameters

HTTP Business API supports a new parameter when searching for calls: nativeID (Technical Call Identifier)

"Record non-configured extensions" checkbox removed from Verba Server Configuration. Defaults to "false", should be manually turned of off existing servers.

Support for LDAP over SSL is added to Active Directory synchronization

Bulk phrase update

Playback reporting has been added

New Webapp configuration parameter: continue or Web player can now either continue or pause playback at the end of markers

Time of creation and time period included in the report email template

Initial version of Ethical wall configuration parser

Added Ethical Wall rules for Cisco phone calls

Agents Quality vs. Productivity (Calls Handled) report shows the total number of calls selected for evaluation for each agent

Decryption of SRTP streams not followed from the beginning (e.g. silent monitoring, controlled recording)

Minor Speech Analytics UI changes

Additional CDR fields populated with voice analytics results.

SQL connection requirement removed from node manager agent, now it uses HTTPS for configuration

New CDR parameters

Improved snapshot isolation handling.

Verba web application is now compatible with Tomcat 7, tested with Tomcat 7.0.39, APR 1.1.27 (still compatible with Tomcat 6)

Licensed edition will not be shown in license summary, only in article list

Memory allocation optimizations

Heuristics to select the most probable codec mapping if RTP source does not use negotiated dynamic payload type

Fixes for stability issues related to intermittent memory corruption

Support for L8 and L16 RTP/AVP audio and video payload type

Manipulating STUN binding success response's mapped address field can totally hide remote endpoint so instead of peer reflexive candidate via proxy to remote endpoint a direct host candidate is negotiated in ICE connectivity check.

Improved SDP handling for better compatibility, c line is added both at media and global level now

New session termination cause: Video escalation (84)

The RS role's contains the latest filter settings

Tomcat7, and Java7 support is added to the installer

The recorder dialog window has been removed from the Lync Filter installer

Command line option for LyncFilter installer to avoid asking for SQL credentials

UCMA calls can now optionally be detected based on configured user agent names. To avoid duplicate/unexpected CDRs, caller and callee information is retrieved from non-UCMA participants' URIs. B2B call ids are registered for each session, if Geomant Contact Expert does not return metainfo based on SIP call id0 we try id1

OpenSSL TLS ctx is now set not to resuse existing sessions

Change Password menu is not shown to users who are only allowed to log in with SSO

Access control problem related to Private tag

Video codec ID in CDR is filled only if video was recorded in the current call

Better timeout mechanism when multiple remote capture sources are used for recording

Caller and callee might be swapped in Lync 2010 RGS calls

Lync RGS fix

Failure to start record when one of the configured NICs are not available

Issues with SIP UPDATE and reINVITE based B2B transfers

Rare issue in passive recorder load balancing

Fixes for handling escalation of audio calls to video

RTP proxy was not aware of SDPs with application/gw-sdp content type (related to Lync Mediation bypass)

SDP compiler changes due to compatibility issues with Kirk Wireless Server

Recording decision issues with Lync dial-out conference join, where callee is rewritten to "conference"

RTP relaying modification to make possible ICE connectivity check in early SIP dialog phase (when SDPs are partially exchanged in B2B calls)

IPv6 candidates might cause parsing issue in RTP proxy leading to allocating port only for one participant causing connection failure

All B2B call legs are now assigned to the same recorder, recorder decides how to merge them

Improved purging of unused RTP connections

When an extension was set up in different Partitions, only the first occurrence was observed.

If the same extension number was registered in multiple partitions, then the CDR information was not collected for one of those addresses

Recording may not get inserted into database if the JTAPI info did not arrive and the call was put on Hold and then Resumed quickly.

Controlled recording after a Hold / Resume operation might not continued recording

Patch for Lync 2010 Response Group recording

Lync filter might crash when third party tool tries to connect to configuration port

Log level of keepAlive sent log entry changed from Debug to Insight

Simulring issue is fixed when one of the clients is not audio/video capable

Improved keepalive resolves timeout issues

SSL read timeout set for the proxy connection (15 sec)

Resolved issue in recording rule handling, Asserted identity field is checked with URI and without uri (forwarded calls to PSTN)

IPv6 address parser issue caused skipping the whole call, even if media connected via IPv4 candidates. This affected edge and mediation based recording only.

Memory leak in P2P announcement service

Bug fixed when the non-recorded party cannot join to a conference

Lync announcement service might can make announcement twice for P2P calls

Some sessions might stuck in ongoing conversations during service shutdown

Issues with SQL Server reconnections

IM service did not start up with certain valid configuration parameters

RTP player component might crash

Auto-pause the window event listener is disabled to save CPU time if monitored URLs are not configured

Autopause does not pause recording when when filtered URL is already open in a browser when a new call starts

If auto-pause URLs are not specified when the desktop agent service is starts, configuration updates were not enforced until agent restart

If ACL was disabled calls recorded via SIPREC/Cisco CUBE were duplicated as if both participant were to be recorded. Now if ACL is disabled only one instance of the call is recorded.

Cisco SIPREC parsing improvements

Agents Call Activity Summary and Recording Servers Call Activity Summary reports did not work when there was a Group filter set up and the output format was other than PDF.

Group Membership History did not work under Tomcat 7

Recording Minute Usage report did not work when there was a Group filter set up and the output format was other than PDF.

User lockout (because of wrong login attempts) did not work.

Some API calls related to Markers did not release the database connection properly. This may led connections to be exhausted so the web application may became inaccessible.

Search Calls API did not work when returnSharedItems was set to 1

Configuration Profiles might not get updated when parameters with "multiple" value type are not stored continuously in the database

Cosmetic changes in call sharing email templates

Error message was displayed when user disabled a service on the web interface. The service got disabled anyway.

Minor style fix on login screen

Event log details were not shown if the first detail was "Call ID"

Possible integer division by zero exception fixed in waveform generator

Simultaneous Calls Dashboard/Report might show incorrect values

Yes/No/NA type of question on an evaluation form could not be saved if the default selection was N/A

Silent monitoring media is now always sent from the same IP address where the monitoring session request was received, this improves support for multiple NICs in recorders and NAT.

Speech search tasks can not be canceled in certain cases

Quick speech search from List Phrases screen did not work

For improved interoperability, HTTP responses from all components now adds Content-Length: 0 if content/body is empty

Node manager agent might return false error codes to GUI

Transcoding issue when one call matches multiple storage policies

Database issue related to chat recording

Automatic call markers might not get inserted if a call is not assigned to a user

Corrected licensing error, that led to unfair calculation of licensed always-on users

If number of processing threads are not explicitly specified, we default to 'number of cores x 2' (instead of 'number of cores')

If multiplexer does not find a voice file and the task is rescheduled, sometimes it might be reprocessed only after a very long delay

At codec change during playback the previous codec was not properly flushed out before initializing the new one

Fixed RTP payload types are always added to RTP processor, because CUBE does not handle calls properly if not the first codec is selected from the SDP answer

Video playback via VMF framework might freeze/jump in case of voice stream with silence suppression (Lync video), since video playback is synchronized to audio and audio timestamps might jump too big in case of long silence periods. Mixing algorithm redesigned to support silence suppression in video calls properly.

Rare deadlock scenario fixed in networking code

VMF recorder sink might overwrite the initial stream descriptors with the final ones at closing the file leading to improper decoding parameters at the first segment in case of mid-call codec/stream changes

In certain situations H.263 decoder swapped U and V channel resulting in blueish colors

VMF buffering and mixing changes to improve silent monitoring

DB and call activity monitoring configuration issue found when weekend days were not configured

CallTimeouts increased from 30 to 300 sec in proxies, 30 to 180 sec in passive recorders

Resolved problems multipart content in SIP

Resolved potential deadlock in multithreaded SIP processing

If primary record has to be overwritten with secondary one in media sync mode now the start and endtime is also updated

The Verba Dial-in Recorder is now available on all Verba server roles

Geomant outbound campaign, consultative agent transfer fix

PrairieFyer contact center metadata parsing fix

Enhanced logging of Geomant Contact Expert integration, missing Geomant call ID is not a fatal issue anymore