Verba Release Notes

Summary

A query error leads to an increasing size of the local database cache on the Media Recorders storing VOX/voice activity events which leads to performance degradation and affects media processing which can cause data loss in recorded conversations. The problem was introduced in v9.6 with the new feature which can avoid creating unnecessary CDRs for long open line calls without any media activity.

Am I affected?

All v9.6 trader voice recording deployments (BT ITS, BT IPTrade, IPC Unigy and Speakerbus) are affected where VOX/VAD triggered media recording is configured and Media Recorders and Recording Directors are deployed on separate servers. Note: v9.6 Unified Call Recorder service executable can be potentially deployed on top of previous installations too.

Impact

- The issue manifests after many weeks or months of deployment (depending on server load and server performance) and once the size of the cache reaches a critical size the issue becomes more and more evident - Parts of affected media records might become choppy/bad quality/some seconds missing -Voice quality check reports packet loss and affected media records in UI/search/call details are marked with packet loss media processing error - In the case of BT ITS integration TTP packet loss alerts are triggered

Workaround

It is recommended to manually delete the local cache database in a maintenance window, every 1-2 weeks depending on server load on Media Recorder servers (Recorder Servers acting as Media Recorder role for Unified Call Recorder service): 1. Stop the Verba System Monitor and Unified Call Recorder services 2. Delete verba_install_folder\work\unified rec\dbcache\voxhistory.db file 3. Start the stopped services

OpenSSL library is updated to the latest v1.1.1j

OpenSSL library is updated to the latest v1.1.1l

The private key password for the SSL certificate configuration can be defined in an encrypted format in the Apache Tomcat server.xml configuration file (previously it was plain text only)

OpenSSL library is updated to the latest v1.1.1n

Passive Recorder does not release/invite conference announcement again at recording segmentation

Support for multi-fork recording through CUBE Media Proxy to allow 2N recording using Cisco network-based recording

If the capture interface cannot be recovered then the Media Collector & Proxy service will be automatically restarted and an alert is raised

Luware: Call As a Service - Process new metadata information to create proper CDR for these call flows

Recording announcements can no longer be configured for Teams recording due to Microsoft restrictions. Configurable announcement policies for P2P calls are coming later in Microsoft Teams as a built-in feature.

The Microsoft Teams Bot service is now able to reread the following configuration values without restart: - Log level - Compress RAW audio to G.711 - Bidirectional/Stereo Recording - Number of Recorded Video Participants - Preferred Video Resolution - Preferred Screen Share Resolution - Separated Screen Share Record - Start Recording After Recorded User Joined - Block Calls when Recording Server Unavailable

Microsoft Teams Bot Service is now running on .NET 4.8. All dependencies (Teams Calling SDK v1.2, Teams Bot Media SDK v1.19) updated to the latest versions.

The Microsoft Bot Media SDK was updated to 1.20.0.348

Microsoft Teams bot grouping feature added for voice recording. The bot grouping allows grouping the recorded streams of the recorded users participating in the same meeting. When enabled, a single audio stream is sent to the bot until the group limit is reached. When the bot grouping is enabled, the following features are not supported for the user handled by the bot grouping: - Video/ Screen Share recording - Controlled recording - Never record (recording mode) - record only if external user is participating - record video only for external participants - record only scheduled meeting

New configuration: Internal Domain, Numbers Pattern - Defines the tenants and number ranges that have to be considered as internal in the case of direction identification of the calls.

The Microsoft Teams Bot service now utilizes a counter to track the number of used video sockets. If the number of sockets reaches the limit's 99% it sends an unavailable response to the health probe requests.

The Microsoft Media Platform SDK health status was added to the Microsoft Teams Bot service's health probe so the load balancers can detect if the Media SDK is in a heavily loaded state and the bot should not receive any new invites.

Microsoft Teams instant message, attachment capture and archiving: - Event/webhook (DLP) based API integration, where the Teams platform sends all updates once the recorder subscribes - Support for on-prem, cloud and hybrid deployments - Supports always-on recording mode only - Supports all instant message scenarios, teams, channels, P2P, and meeting chats - Supports all formatting options, emojis, giphys, stickers, and other apps - Limitation: no selective capturing/archiving support, only tenant-wide capturing - Limitation: chat sessions initiated from external domains won't be captures

New service counters for Unified IM Recorder: - Number of new events in last 15 minutes - Number of new events in last 1 hour New service alert for Unified IM Recorder for inactivity alarm (disabled by default) with the following service level settings: - Enable Recording Inactivity Alert - Minutes Between Recording Inactivity Checks - Recording Inactivity: New Message Events Threshold

Conversation Detail Record is no longer created for bot participants

Added support for P2P and Group chat replies

Microsoft has changed the license model for chat and channel archiving integrations which have to be defined when using the APIs. The following licensing models are available: - model=A is restricted to applications performing a security or compliance function, and requires a supported license. In the future, apps will also be required to pay for the messages they consume beyond the seeded capacity. - model=B is restricted to applications that do not perform a security or compliance function. model=B is free to use today; however, in the future, apps will pay based on the number of messages they consume. There are no licensing requirements for model=B. - Evaluation Mode (default) enables access to APIs with limited usage per requesting application for evaluation purposes. Change notifications will not be sent if the limit is exceeded.

Improve the Avaya DMCC Recorder to not try to register non-numeric Extensions, and not go to the next CM in case of InvalidActiveDeviceState and InvalidParameterValue errors

The default value for the "Do Not Keep CDRs without VOX Activity" setting is changed to "Only at Call Segmentation" (from "Yes")

Minimum RTP port range limits are removed when configuring the RTP port range for the service (Media Port Range Begin, Media Port Range End) to allow deployments where the RTP port range has to be kept as narrow as possible. Previously, the RTP port range was limited to a minimum of 64 / CPU core.

Integration with the Verizon Meeting platform using SIP based dial-in recording where the meeting host ID and the meeting ID are transmitted using DTMF codes

In case of a large amount of CTI/CDR records and absolute call splitter timer, the splitting might have taken 1-2 seconds causing such length of gap between termination of previous and starting of current segment's recording

BT IPTrade TPO recording: metadata reflects the TPO side known display name of the recorded party as it might be more valuable for open lines/DDIs (provides line/DDI name instead of trader/username)

In the case of BT ITS integration, which does not enforce recording rules, the recorder still tries to lookup recording rules/recorded extension to support direct upload/export, shared server features

Local SIP Port can be configured in the IPC Zone configuration widget

IPC Unigy - There was no alerting if media channel establishment (SIP) failed

Dial-in Recorder: add support for call segmentation and improved silence gap filling for silence suppression/DTX

Canceled Symphony calls are recognized and no longer marked as failed recordings

New integration with the Genesys Pure Engage on-premise contact center platform: - Integration with Genesys SIP Server for media stream recording - Integration with Genesys T-Server for CTI - Always-on and selective recording

T-Server connection alert throttling is introduced. The service will send alerts only in the case of a lost connection.

Added support for "individual recordings" change on Zoom API for Meetings - now if an extension is set for audio-only recording, the system will import the recording specifically created for the recorded user covering the period the given user was present at the meeting. This feature is only supported for audio, the video file will still contain the entire meeting.

The Zoom Phone and Zoom Meeting import source can raise an alert if there are only 2000 API calls have left to reach the API rate limit

Zoom Meeting participant list is now cached during import and reused during CDR reconciliation, reducing API load if CDR reconciliation is also enabled

Apache Tomcat version is upgraded to 9.0.37

The duration filter now supports milliseconds to make sure the specified seconds are correctly filtered in the Data Retention Policy, Conversation Export, Labeling Rules, Role features

Start Search Automatically is disabled by default for new installations to improve search performance for large databases. It can be enabled in the service configuration.

New Identity Provider (IdP) option: Authentication with Reverse Proxy which allows single-sign-on deployment where the IdP is providing the authentication information in HTTP request headers via a reverse proxy deployed in front of the Tomcat application server

Updated Web Application dependencies: Puppeteer (v7.0.4), embedded Chromium (v90.0.4403.0)

Apache Tomcat version is upgraded to 9.0.43

Default datetime settings for search: for Today, and Last X Minutes filters use the end of the current day time instead of the current time

Updated Web Application dependencies: Puppeteer (v9.0.0), embedded Chromium (v91.0.4469.0)

Custom legal disclaimer messages can be added to the login screen via the web UI customization/brandind feature

Updated Web Application dependencies: Puppeteer (v10.1.0), embedded Chromium (v92.0.4512.0)

The Data Retention Policies showed all Background Task records for a given policy, and that can result in a huge dataset. It will show the last three tasks from now on.

Improvements for built-in accounts: - Verba Administrator user account cannot be deleted (only invalidated) - Verba API User account cannot be deleted - Verba System account cannot be managed through the UI and cannot be used to login into the system

Login ID is no longer stored in a browser cookie. It can be turned on under Web Application / Password Policy / Store Login ID in Cookie

OPTIONS HTTP method has been restricted

Apache Tomcat version is upgraded to 9.0.54

Support for the X-Forwarded-By request header is added to generate full URLs when it is present (e.g. when using a reverse proxy in front of the web application server like IIS)

The prevent simultaneous login with the same Login ID feature now supported across multiple Media Repository servers

New configuration option (Mobile Host Prefixes) to redirect requests to the mobile web application

Ability to remove a Legal Hold Label from conversations individually. It is tied to a new permission, Remove Legal Hold From Conversations, which is not granted to any roles by default.

Try reading from the Active Directory again for three times if there was a network problem during AD Synchronization

Apache Tomcat version is upgraded to 9.0.62, APR 1.2.33 (OpenSSL 1.1.1o)

New Users Conversation Volume report showing the total number of recorded conversations for users

Conversation Count Trend widget improvements: - New filtering options: Platform, Import Source - Date filtering did not work

Group filter added to Users Without Recording Report

All internal and external APIs now use HTTP Basic Authentication by default

Provisioning API can be used by tenant-level users (previously it was only available for service/hosting providers)

Legal Hold API allows creating and removing litigation holds in the system programmatically

User Custom Fields added to the Search API, and End Time can be filtered by the endTimeFrom and endTimeTo parameters

Performance improvements for Azure AD synchronization to support a large number of users

Revised NIC/capture interface listing to provide more information and interfaces which might be discarded due to special drivers involved (e.g. NIC teaming)

Microsoft JDBC driver (used by Java based services such as Web Application, Cisco JTAPI, Cisco IM&P, Avaya DMCC) is updated to version 8.2

The database Index Rebuild process is automatically stopped after 24 hours to avoid blocking normal operation by the database maintenance process

Improved call record insert performance by removing the possibilities for long running queries (e.g. legal hold management)

Data management policies are no longer paused during database maintenance jobs to better utilize the SQL Server resources

Improved database table index rebuild process which runs every days (instead of weekly)

When the Archived Conversations checkbox was turned on in the Search settings, then the search was slow in case of large databases

Multi-tenant license allocation and enforcement which can be used to allocate licenses to individual tenants from the license pool. License enforcement is also tenant-based when this option is enabled.

New Intel based MP4 encoder added for video transcoding

Customizable alert severities and selective alert sending option added for the System Monitor service on the servers

The event ID attribute is added to SNMP traps

The import service sends an alert when an import source cannot be started due to a bad configuration

Microsoft Azure Blob Storage support (in addition to existing Azure File Storage)

VSFT secure file upload protocol connection timeout is now configurable

New SMTP export target field ISPLATFORM([value], [true], [false] optional) to define conditional values based on the value of the platform field

Export task page showed end time as 1 January 1970 01:00:00 for ongoing advanced export tasks. From now on it will be blank. Changed both start time and end time column formatting to match the logged in user locale.

Storage upload optimization: if too many records couldn't be found in the database or no upload policy was matching, the upload could be stuck or was slow. These records are not processed separately with lower priority, so other/normal records can be processed faster.

Improved SFTP connection error and timeout handling

Significant performance improvement in data retention management for the advanced voice (incl. trader voice) data model which makes the deletion process much faster. After this improvement, the system will not try to check for each Media-Only record if there are related CDR-Only records with higher retention during the execution of the deletion, instead it relies on the retention period configuration of the Media-Only records.

Improved log database table handling for advanced export jobs to avoid database locking

Updated SFTP library to support more 'Key Exchange' algorithms. Now "diffie-hellman-group14-sha256, diffie-hellman-group16-sha512, diffie-hellman-group18-sha512" are also supported.

Encryption and signing capabilities are added for external Recording Server configurations. Both encryption and signing are available as server-level settings (instead of data management policies) and can be applied to all recorded data only (no filtering) during the upload process. The crypto information is also uploaded and re-used during the import process.

The Data Retention default schedule will be Hourly from now on (previously it was weekly, configurable)

WORM capabilities are added to Amazon Web Services (AWS) S3 storage target. It supports WORM features such as object lock and legal hold as well.

New Addressing Mode configuration setting for Amazon S3 and Amazon S3 Compatible storage targets to define if Virtual Hosted Style or Path Style addressing mode should used

Vulnerable ciphers are disabled by default for the Media Streaming service

Microsoft .NET Framework updated to version 4.8 in the install kit

Updated Web Application dependencies: Puppeteer (v5.4.1), embedded Chromium (v87.0.4272.0) and Node.js (v14.15.0)

Multiple improvements for the PowerShell Deployment Toolkit: - The PowerShell Deployment Toolkit now searches also for the AdoptOpenJDK JRE 11, ODBC Driver 17, and .Net 4.8 when checking the prerequisites. - 3rd party certificates now can be provided in the PowerShell Deployment Toolkit as Web Application SSL certificate (HTTPS) - The PowerShell Deployment Toolkit now uses the same process for installing the Verba database as the MSI installer. This includes several improvements: Support for Always-on Multi-subnet Failover setting, Support for database partitioning, Running configuration and IM schema update scripts when upgrading from Verba 8.x -SQL script execution log - The PowerShell Deployment Toolkit now searches also for the Oracle JDK 11 when checking the prerequisites. - The PowerShell Deployment Toolkit no longer asks for confirmation for CredSSP authentication when starting the script

Playback, download, and export support for legacy calls stored on Hitachi Content Platform and migrated from Verint v11

Verint import (dual consolidation) improvements: - changed metadata parsing - added VOX metadata enrichment from CTI records - added ANI, DNIS hidden number handling - added new metadata fields: Cdm Id, External Call Id, Interaction Type

Migration from Verint v15 WFO platform allowing importing of historical recordings: - Support for archived calls only - Supported archive mediums: SMB folder, EMC Centera, Hitachi Content Platform (tapes, DVDs, or any other removable media is not supported) - Storage targets are automatically created based on the archive configuration in Verint - All Verint file formats and codecs are supported: wave file using G.729, G.723.1, or G.726 - Encrypted calls are not supported - Both back office and front office (trader voice) calls can be imported - Users, Groups, and Extension can be migrated from v15.2 systems. The Users' conversation access scope is not migrated. - Migrated calls are assigned to users defined in Verba based on their associated recorded extensions (Trader ID / Extension or Phone Number / SIP URI)

When using the Verint Import Source, the Verint Recorder Addresses configuration field can be used to provide the serial number and Verint recorder server FQDN/IP mapping

Verba Import Source has a new option to reinterpret/update recorded participant/owner of a call where it is already set

In the case of Verint WFO 15.2 import, the shared line extension information (presented as ext#/MAC) is imported as separate fields (From/To and From/To Device fields)

Integration with TeleMessage mobile archiving system to allow the capturing and archiving of WeChat, Whatsapp, TeleMessage Enterprise platforms

Azure storage certificate check can be disabled optionally for Azure Blob and Azure File Storage for Verba import policies when the certificate revoke checks don't work due to network firewall limitations

Generic import execution time significantly improved when CSV files were used

Multiple low or moderate Web Application vulnerabilities, affecting specific pages only, are resolved: - SQL Injection in a single API call - Insufficient Cache Controls - Cross-Site Scripting (XSS), Reflected - X-Forwarded-For Injection for API calls - Vertical Privilege Escalation on a few read-only pages

Microsoft Teams Bot Service logged internal API credentials (including the password) in plain text

VLAN decapsulation might have missed the last 2 bytes leading to processing problems

Passive Recorder placed the recorded calls into the wrong folder for remote cluster (shared recorder) calls and the upload did not work

In the case of Cisco SIP Proxy based recording, when the domain part of the TO URI was invalid, the call setup failed due to strict validation of the destination domain of the URIs

Database records were not created for calls recorded by Cisco network-based recording if the call was started during daylight saving change in UTC

Verba Cisco JTAPI service stopped recording in very rare circumstances when the Keep-Alive from the Unified Recorder timed out and at the very same time a new Keep-Alive process started

Cisco JTAPI Call Recording Rules based on Genesys information did not work for Consult call scenarios

Cisco JTAPI Service did not reconnect to the CUCM after a Provider Shutdown event

The Verba Cisco JTAPI Service sent a CUCM Connection Down alert during service startup

Cisco JTAPI records are inserted into the database with a delay when a Controlled Extension started to record automatically, and the call is considered finished by the JTAPI Service

The Cisco JTAPI Service ran out of memory and crashed when the Cisco UCCE or UCCX integration was enabled. The system raised an alert after the service failure. During the service failure, the system could have missed metadata.

The service did not properly close the ongoing sessions during restart after an ungraceful shutdown of the service and generated empty records in the database

Video was flickering in Acano meetings with Skype for Business interop when only voice recording was enabled for the user

When RTP Proxy with advanced relaying was enabled, the Teams inbound calls to SfB failed to connect if media could reach the Verba proxy

The From Info field was missing (display name of the caller party) if initial the INVITE did not contain the display name but it was available later in the call setup sequence

In the case of controlled recording, sporadically the Microsoft Teams Bot service made two control database records when the call was started as video. After the call ended, the control record made first, stuck in the ongoing calls.

The Verba Lync Call Filter and IM Filter Services didn't reset the keep-alive state in the case of non-graceful disconnection. This could lead unsuccessful connection attempts. This state could be resolved by restarting the filter service.

During Skype for Business application share recording if RDP was in use and the sharer/viewer role changed and meanwhile the storage upload cycle ran, the initial segment might have been lost

The Skype for Business IM recorder could not create an XML file for the conferences when direct export/upload configuration was used

HSTS response headers were missing for HTTPS listeners on 9440 and 10100 ports for the Microsoft Teams Bot service

Microsoft Teams Bot service was installed with the wrong default value for the Recording Notification setting and recording was still controlled by the system while it should not have been

The phone numbers were not parsed properly in the case of conference calls. Therefore, the phone numbers were not added to the participants list.

The Microsoft Teams Bot Service didn't take into account the recording rule configuration of the delegator in the case of delegated calls. Now the bot service applies the recording rules defined for the delegator for the delegated calls answered or initiated by a delegate.

When a user configured for recording only audio in Verba, the recording notification was shown in the case of app share only calls too.

When the bot was not able to answer a call because of missing parts of the call invite, the bot didn't register the call as a non-answered call. It caused the bot to still allocate resources for handling the call and the allocated resources were not disposed until the service was restarted.

The video stream of the guest users was ignored and the bot never subscribed to them

The Verba Microsoft Teams Bot Service took the Advanced API configuration into account for recorder connections. When the Advanced API was turned off the Verba Microsoft Teams Bot Service didn't start a secure listener and the Verba Unified Recorder Service could not establish a TLS connection.

PSTN participant numbers were not stored in the case of peer-to-peer calls if there was no Teams tenant ID presented

Microsoft Teams Media SDK generated invalid timestamps for audio leading to media processing issue and continuously growing audio files

When a user started a call with audio modality and added video or application sharing modality later, the Verba Microsoft Teams Bot Service started the controlled recording automatically.

The Verba Microsoft Teams Bot service did not provide the source or destination caller IDs, when it identified the recorded user's tenant as a non-provisioned tenant.

In the case of controlled recording, the Verba Microsoft Teams Bot service created an additional control record when federated user or guest user entered the meeting.

The Verba Microsoft Teams Bot service automatically started controlled recordings in the case of specific extension configuration. When the 'Record Only If External Use is Participating' option was enabled the dominant speaker changed event triggered the controlled recording.

Screen share recording has stopped when a participant has left the meeting, and screen share recording did not continue for recorded users after that point.

In the case of an unknown caller ID or UPN, the Phone Number field in the participant information contained invalid characters when non-Latin characters were used for the display name of the participant

When the service failed processing an item of the participant list, it didn't process the rest of the participant list.

The audio stream ID was not properly updated in the case of P2P calls. It caused that in the case of stereo recording the bot didn't use the dedicated channel for the recorded user.

The Microsoft Teams Bot service started the recording in the case of undefined call direction. The service will wait until it can gather enough information to calculate the direction of the call. The affected call scenarios: P2P VBSS-only and call queues.

Microsoft Calling SDK is rolled back to v1.2.0.1702 due to performance issues with the latest version in handling call invites and participant updates. By rolling back to this version, the bot grouping feature will be entirely disabled.

In the case of controlled recording, adding the screen share modality to the call automatically started the recording.

In the case of transfer scenarios, the party which initiated the transfer was stored in the From Info field instead of the To Info.

In the case of a non-configured recorded user, the video modality was not recorded when the Record Non-configured Extensions configuration setting was enabled.

Large Microsoft Teams meetings were not recorded if the list of participants is truncated, and the recorded user was not on the participant list sent to the bot. The system did not raise an alert or log an error when an invite to the bot was not followed by a related participant list update. The issue can only be detected if call/meetings logs are compared with the database records. The issue affected only the call recording taking place for the specific user who was not listed as a participant. If there were multiple recorded users joining the same meetings, the meeting could be recorded from the other user's point of view if that user was listed as a participant. In order to provide an immediate workaround, the system uses a different recording start logic in the bot to not wait for the recorded user to appear as a participant. There is an open support case with Microsoft on how to get the full list of participants because the system stores the participants in the database, and the information is incomplete in this way. The Start Recording After Recorded User Joined settings must be changed to No in order for the bot to ignore the join event for the user. The setting defines if the bot waits for the join event of the recorded user before it starts recording a meeting. This setting is only applied for meetings, for P2P and PSTN calls, the bot always waits for the recorded user join event. If the configuration is set to 'No', the bot will trigger recording after it processed the first participant of the first roster update. The service will also stream every media packet to the recorder without waiting for the recorded user to join.

When a Microsoft Teams tenant was moved from one Verba environment/tenant to another, the ongoing chat messages were kept stored in the original environment, only new conversations were recorded correctly to the new environment.

The message processing queue was stuck when federated Skype for Business users were participating in the chat room

Stickers could not be downloaded and archived for P2P and group chats

Attachments could not be accessed for standard users, only for system supervisors

Microsoft Teams instant messages were missing sender information intermittently

Preferred Codecs list was incorrect in the hint of the Avaya Recorder / Avaya DMCC / Preferred Codec server configuration

The Avaya DMCC service cannot record after the Unified Recorder service was restarted meanwhile the Avaya DMCC service was communicating with it

The duration of the stereo recordings (AU files) was double the length of the call

If the Recording Director and the Media Recorder is separated and either File Share, Poll/Q&A, Whiteboard or SMS modality was enabled in addition to Voice/Video or Screen Share (but not all modalities), then the recording rule became invalid end the call was not recorded at all. Alerts were raised with "Cannot register call on media recorder" error.

Unified Call Recorder applied call segmentation for video and screen/application share recordings, which led to keyframe/video decoding issues after splitting the record into a new one. Call segmentation is disabled for video and screen/application share streams.

If VOX/VAD triggered recording generated invalid start time, it could have led to Unified Call Recorder service crash and memory leak at call segmentation. The system generated call processing and low memory alerts.

Using the E-Dialer integration, the Cisco master call and any non trader-voice recordings where VOX was enabled might have triggered "no media/RTP received" call processing alerts if the call was just pure silence and so no media file was created

Partially recorded files for remote cluster/shared recorder were not recovered after the recorder crash. After applying the fix, the previous partial recordings can be recovered.

IPTrade turret based playback was very slow due to the wrong execution plan compiled by SQL Server

BT IPTrade call playback on turret did not work in multi-tenant environments

BT ITS LDAP provisioning queries could take too much time during service start which could prevent the service from starting as Windows considered it a service failure after a timeout and killed the process prematurely

BT ITS CTI recorder might crash if CLI info was received on not IP recorded line or VR ID was not set for recording. It affects v9.6.11 or later.

BT IPSI media recording did not work from version 9.6.13.6740, media could not be recorded at all

In some case, the system sent a false call processing error alerts if a device change happened on the line

IPC Unigy recording did not work when AudioCodes SBC was in the media path

Cloud9 Recording System API integration did not work and C9 clients were unable to upload data to the Recording Server (affects 9.6.0.5983 or later builds only)

Cloud9 Call Data API import did not use the Work Folder setting and used the default application work folder instead for media files

Cloud9 Recording System API based integration did not work intermittently due to an uninitialized connection timeout

Cloud9 Recording System API based import issues resolved: - import did not work when multiple C9 clients were uploading the media file for the same call (shout down) at the same time - in the case of database errors after database failover, database entries were multiplicated - improved work folder cleanup process - improved network connection error handling - participant information was missing in some call scenarios

C9 Call Data API did not handle the HTTP proxy error code 504 (Gateway Timeout)

Cloud9 Recording System API based import source did not insert the data to the right tenant in multi-tenant environments.

The Auto-Pause configuration was not taken into account properly when the screen was still recorded after the call has ended due to wrap up timer configuration

The Verba Screen Capturing Service crashed if "silent/non-visible" mode was configured

Metaswitch Perimeta: caller/callee was swapped in case of interconnecting calls

Screen/application share modality was not properly determined for SIPREC sessions and the system stored these streams as video recordings

Symphony chat messages with multiple attachments could not be displayed

The Verba Genesys CTI service failed to insert database record in the case of database record duplication check.

The Genesys CTI service didn't start recording based on CTI event fields mid-call. The service only applied the recording rules at the beginning of the conversation.

Fixed handling of CTI triggered recording in the case of hold-resume, blind transfer, consultative transfer, consultative conference scenarios for Genesys Active Recording based integration

Multiple fixes for Zoom Meeting and Phone import/archiving: - Fixed individual recordings choosing based on time: when a user joined a single meeting multiple times, the correct individual recordings are chosen for the resulting CDR (previously always the same was chosen) - Fixed a bug when sometimes a meeting or phone call that was already imported was downloaded and attempted to be imported multiple times - Import/reconciliation stopped progressing indefinitely when there were no meetings/phone calls for the entire max filter period (Meetings 7 days / Phones 28 days) - Some configurations of safety lookback with phones reconciliation caused reconciliation to stop progressing indefinitely when calls happened near midnight - Fixed potential data loss in meetings, when a very high amount of meetings were imported in a single cycle, and the load was unbalanced between "past" and "pastOne" type of meetings - Fixed a SQL problem that could cause crashes, and prevented reconciliation tasks to appear in background tasks - Fixed Meetings reconciliation that did not work when recorded extension configuration was set to voice only - Solved a problem that caused reconciliation to cause false-positive alerts when Zoom took a too long time to archive meetings or phone calls. A new parameter is introduced for both Zoom Phones and Meetings import: Reconciliation Delay Behind Import in Minutes (60 minutes by default), which causes the reconciliation to ignore elements within this period before where import progress is currently at. Never set this lower than how long it takes for Zoom to process the archiving of a meeting/phone call.

Zoom import source stopped working after receiving an unkown archived file type for Zoom meeting import. Unkownfile types are no longer causing an error, but a warning alert

Workaround for missing participant leave date (unexpected) for Zoom Meeting imports

Video playback area in Internet Explorer didn't fit properly to screen size. In some cases, when the screen resolution was under 1080p, it overfleww into the conversation details section.

Send URL feature did not work

Audio segment download in player did not work for trader voice recordings with stitched media

Ad hoc transcoding jobs were executed by any of the Media Repository / Application Servers (whichever has seen the job first) and the transcoded video was not available for playback on the server where the user was connected to

The scroll bar was missing in the conversation metadata tab in the player

The selected Identity Providers were displayed as "Available" too when cloning an existing Role

Users were able to bypass playback restriction when the multiple playback feature was used

Bulk Media download was available even if the user had no "Download a Conversation" permission

Copy URL icon was not shown on conversations

Advanced Export was visible on the Search screen even if the user did not have the "Conversations List Export" permission. The server threw an error when the user tried to use it without a permission.

The newly created users were not put into the groups which were selected on the AD profile's New Users' Properties when the Synchronize Security Groups checkbox was not turned on.

Media files bigger than 2 GB cannot be downloaded and played back from the web application

If the user did not have "Customize Conversation Export Target Folder" permission, then the Direct Download feature did not work

In the case of large voice transcripts (approx. 20k words), the search hit highlight could not process the transcript, therefor the transcript didn't contain any highlighted words.

When a Controlled auto-start recording started, only the Administrator was able to stop the recording

When the Active Directory Synchronization had a custom schedule in a redundant deployment, and Media Repository 1 executed the sync, then Media Repository 2 computed the wrong "Next Time of Execution" and locked the database for a few hours unnecessarily

The CSV user export did not differentiate the "User/Agent ID" type extensions

The silent monitoring button on the Agent View page was added multiple times for each on-hook agent.

Background Tasks page End Time column was converted to the user's timezone two times

Active Directory Synchronization: groups not re-added after the user is re-activated in the same Active Directory Synchronization Profile

Microsoft Teams Instant Messages could be duplicated when searching for a text for a long time period

The Verba Web Application Service could not send emails after upgrading Java from jdk-11.0.8+10 to jdk-11.0.11+9 due to wrong default TLS configuration for Java Mail API

The Verba Finesse Gadget didn't fallback to the legacy API authentication method in the case of Verba Web Application older than version 9.6.

The transcription did not load into the player when the user did not use the Text Search

When the IWA authentication was used, then the Session Monitor stored the wrong Session ID and so after a Logout and Login, the User saw the "Other user(s) have been already logged in to the system with the same login id" message

The Reports / Create New Report menu showed a blank page when using Internet Explorer

The meta XML file for CDR-Only records could not be downloaded in the Conversation Detail File format section, and if the files were stored on non-SMB storage, then no files were shown at all in the File format section.

The fetch following or previous SMS messages did not work in the Conversation View

Error code 400 was received after opening a new tab when Extensions, Users or Alerts List were clicked (Open in New Tab or CTRL + Click)

When searching for Metadata Fields in the Search feature with a negative operation (Not equal to, Does not match a value in the list, Not like), then the calls where the metadata value did not exist were not shown

Import from Every Source checkbox on the Extension configuration page did not work when no import source was selected and the goal was to not use the extension for any import source configured

The reports list downloaded the wrong file when pagination was used

Dashboard did not work when the Date Interval was set to Hour

If a User was created manually, and the "Synchronized by Active Directory" was turned on, then this User was not deactivated when not found in the Active Directory, because the Users that are synchronized by failed AD Sync Profiles are never deactivated by the system.

Data Management Policy delete gives no error explanation when referenced by another policy as "Execute Only After Another Policy Executed"

Private checkbox did not mark the conversations as Private

The Conversations Search Export / Media feature downloaded previously selected recordings' media files after the user performed new searches

A "Database Maintenance Error" alert was generated after a Conversation that was already marked for deletion got a Legal Hold Label

Direct Search link did not work because the colon character triggered a false vulnerability blocking

Viewer for advanced IM (Microsoft Teams) didn't respect the advanced search timezone display settings

Pressing the Find button on the Automation Rules - Cases redirected the user to the Automation Rules - Labeling page

The existing custom metadata fields were not shown after an upgrade from v9.3 or older version

Right-click options did not working properly on the Conversations screen

Active Directory Synchronization User read failures were not detected if paging was required (more than 1000 users) and it did not happen on the last page. This led to user invalidation which could have led to recording and data loss. The underlying LDAP protocol did not provide any error.

Right-click options / Same day features on the conversations screen chose the day in UTC timezone instead of the user's timezone

Waveform always showed on peak when static there was noise in the call. Now data is normalized by the visualization script.

Missing waveform with SMB network storage for MP4 or MP3 media (did not apply if VMF was available)

The Server Capacity report did not work and the system generated 0kb reports (no report output) with exceptions in the log files

Recorded Users and Extensions report did not show the Distinct Total value in case of the CSV format

Dashboard snapshot schedule was late when the creation time converted to UTC resulted in a negative time. For example, 7:30 AM Hong Kong time should be created at -0:30AM UTC. That was not converted to 11:30PM.

Agents Evaluation Details report's CSV format contained the column headers multiple times

Minor reporting problems with a few Quality Management and the License Usage Summary reports: - Quality Management reports: when the Available Points or the Weight of a question is zero, then the report is not generated (0 byte file) - License Usage Summary report: when the Database Server's default collation is different from the Verba database's collation, then the report is not generated (0 byte file)

The Audit Log Report XLS version did not display the Search Query details

The Permission Configuration Report displayed only reference environment users

The User Instant Messaging Details report did not show the participants of the conversation. It only showed the user who sent the message.

Voice Quality Check Summary report generated 0KB file when filtered for groups

Reporting did not work with Firefox. The issue only affects 9.6.17.7145 and later.

The Roles and Permissions report did not work when there was a Role with the "Unable to Access Conversations Older than

Non-numeric values were not filtered out in a few Quality Management reports so those reports did not work when a non-numeric value was written to the scorecard. Affected reports are: - Agents Evaluation Details - Agents Skills Performance Summary - Average Score by Question - Question Score Details - Skills Performance Summary - Skills Performance Trend

If the case of video recording with mixed audio, the voice transcript included everything twice

In the case of Cisco announcement, if callee was forwarded from one to other recorded participant, multiple announcements were played (one for each hop)

The Skype for Business announcement white list considered the PSTN numbers as internal numbers because of the domain part appended by the mediation server

The Verba Lync/SfB Announcement Service could not insert announcement audit log to the database because of a password encoding issue.

SearchCalls API threw an error when searching for ongoing calls without the end time parameter

Import REST API did not work intermittently due to an uninitialized connection timeout

The Provisioning API v3 GetUsers call did not return the Roles of the Users

The Provisioning API passwordExpires was not written into the database, and when querying the User, the workflowId and retentionDays attributes were set to 0 instead of null

The Business API did not return archived conversations

Provisioning API v3 - Invalid import sources could be added for extensions.

SQL Server Failover Partner connection setting was not adjusted for ODBC connection string scheme if a custom port number was specified causing persistent SQL connection failure. This only affects C++ based services such as Storage Management Service, Unified Call Recorder Service, Passive Recorder Service, Import Service, Node Manager Agent Service.

Certificate validation: revocation status was checked only for issuers but not for the actual certificate which caused the system to continue operating with revoked certificates

When site configuration was used and the Recording Server recorded a call for the user using an integration that supported shared server recording rules, the system could have recorded calls on the server which should not have been recorded on that site.

The following deletion functions did not work since 9.6.4.6145: - On-Demand call deletion after call buffer expires (Storage Service) - Deletion of CDRs where there was no voice activity for trader voice recording (Unified Call Recorder Service) - Deleting records during deduplication (Storage Service) - Manual deletion on the Web UI (Web Application Service) In all cases, the related services raised an alert on failed deletion attempts.

Recorder and Import service-specific built-in metadata fields were inserted into the reference environment/tenant in the case of multi tenant deployments, instead of the users' tenant

License Usage was not counted after a Web Application restart until the daily license check or visiting the license screen

Transcoded MSRDP based Skype for Business screen/application share recording could not be played back (showed green screen) if the original stream (shared window) height was an odd number

Very short (<500msec) video/screen share transcoding might have stuck

Memory and resource leak fixed in MP4 transcoding. MP4 transcoding might have stuck in the processing state at the end of the process when SMB storage was used.

Video frames could be missing when decoding VMF files when the aspect ratio of the video stream was changed and become noticeable with low framerates such as application share streams

When an SNMP trap was sent successfully it was inserted into the database with unsuccessful SNMP sending status

System Monitor service at detail level logged the credentials for the API user requests (injecting alerts into the database)

Media stitching for playback and export was very slow for trader voice conversations due to the wrong execution plan compiled by SQL Server

Advanced Export manifest file contained the temporary file path of the media files if the file was to be transcoded or stitched

In case of data was modified by other processes during executing the deletion transaction in the deduplication policy, the failed transaction was not retried, instead of alert was raised and processing was reattempted only at the next policy run

On-demand conversation processing SQL queries might fail after a long idle time. Alerts were raised and queries were executed again.

Policy based export ignored the "Encrypt XML metadata" checkbox on UI and encrypted all the time when the media was encrypted

The system showed wrong export statistics for Bloomberg Voice on the user interface (multiple tasks were checking the same storage target level files and updating their task status). Improved error and SFTP handling.

The wrong file was played back in some cases in the player in case of Verint v11 migrated calls which were stored on Hitachi Content Platform

When the user tried to replay a media for a trader voice call which was not yet placed on the storage that caused the media to be unplayable for any further attempts

The priority of the upload policies was not properly handled and in case multiple upload policies were matching the call the system did not take into account the configured priority setting

Transcript of voice calls could not be exported to an SMTP target

Azure Blob based storage target was slow in case of a large number of files in the container (>500K)

Advanced Export: CSV metadata was not properly escaped if it contained delimiter, newline, or " characters

When performing a Search of Conversations using for example Singapore time zone and performing an advanced export, the result in the CDR CSV file export displayed the time zone of each call in GMT rather than Singapore time zone

Deduplication policy related improvements and fixes: - the policy did not check if the media had been uploaded for all CDR/Media records, so it might have failed to delete or copy the related ones and the processing policy was retried in the next schedule - the policy did not give up processing (committing conversation as processed) conversations that were under retention - processing secondary records might have continued with the next policy before the related primary pair had been processed by the policy

In rare cases, when the Storage Management Service was looking for matching policies related to a call, and the call hadn't been updated in the database yet, a Policy processing error: no upload policy alert was raised. The call was processed and uploaded later just fine.

Copy policy failed when the call was not stored on a storage target (it was stored in the local media folder on the Media Repository)

Storage Management service crashed due to database connection issues during export

Microsoft Azure File Storage playback/download did not work after Microsoft changing the API

Verba Secure File Transfer (VSFT) based upload issues resolved: - Storage Management service crashed when upload failed and retried in the same transaction - High CPU load caused by an infinite loop in SSL error handling - When advanced API security was disabled, the system fell back on the wrong certificates

Storage Management service crashed when voice quality check was enabled for upload and it processed a CDR-Only record

A retention period expiring after 2038 could not be set when using file-level retention based storage integrations (EMC Unity FLR and iTernity iCAS)

The "Move", "Copy", and "Archive in DB and Move Media" policies could not move the files after upgrading to a version that supports Dual Archiving, because the "Which Copy" was initialized to "Both" (that is 0)

Advanced IM Export (Microsoft Teams) issues resolved: - If the original filename was not specified in the attachment and a random filename was generated, the extension was not properly added during export - If the conversation export did not find any IM or attachment to be exported, the export was not counted as success nor appeared in the total to be exported conversations counter under background tasks - The attachments which has not yet been downloaded or could never be downloaded are marked in the exported messages - The export did not work for conversation where an attachment download failed or was in progress

Trader voice export could have contained ongoing, partial conversations

If a Verint-Migrated call was Archived in Verba, and a secondary copy was deleted, then the deletion was not registered in the database

Upload did not work when no matching policy was found

Playback, download, and export was slow when using Azure File Share Storage

Manifest file for advanced export did not contain the end time of the export job

Metadata XML files for CDR-Only records were not uploaded to EMC Centera and the generated file based on the database during export did not contain custom metadata, participants, etc.

Transcoding of application share recordings resulted in a video where the screen share stream was not visible if the aspect ratio of the app share stream was greater than 9:20 (WxH)

Audio transcoding was not available for screen and application share recordings during export configuration. Audio formats are now listed and the audio portion of the VMF files can be transcoded to standard audio files.

Erroneous background task counters were displayed for the deletion policy for migrated Verint v11/v15 TAR files. In addition, potentially high memory usage when querying a large number of records to be processed.

Advanced export ignored the Recent Than, Older Than, Elapsed Time Since Transcoding filter options when the schedule was set to "Once immediately"

Advanced IM export e-mail attachments sometimes were ZIP files with wrong (.png) extensions

If custom thresholds were set for voice quality check then false alerts might have been raised because RTP Decoding Errors and Beep & Clicks scores were swapped when evaluating thresholds

Deduplication could process both primary and secondary recording at the exact same time. If both tasks tried to delete a file or swap something in the database then the system raised alerts about missing files.

Imported Cloud9 recordings (via the Call Data API) could not be uploaded when there was no media file referenced

Migrated TAR files from Verint v11/v15 were not deleted when two TAR files with the same name were archived two times

When the system was unable to determine the size of the file to be uploaded to a storage target, the upload process failed due to the invalid file size parameter. Affected storage targets: Amazon S3, Verint, Bloomberg Voice, O365 Compliance Archive.

When the key was not available during decrypting legacy encrypted files (VF), the system didn't recognize the missing key and generated invalid files

Using advanced export, the naming pattern for files could have caused the media file name to be not unique

Advanced IM export did not take into account the Do Not Export Attachments option

Participant information was missing for advanced IM export (Microsoft Teams) when the UPN information was not available when the CDR entry was created (it only become available later)

Auto-delete after the retention period is over did not work in multi-tenant deployments. Those conversations were not deleted where the retention has expired. The auto-delete feature is used when the retention of the conversations is set by an upload policy, instead of using a delete policy.

Storage Management Service could have raised false alerts on upload failures (no matching policy) due to race conditions in database and file operations

Transcoding from video/app share/desktop screen to audio format did not convert the call to voice only modality and did not delete the original video file

When both Verba\LocalIP and LocalIP6 were empty, the Media Streamer service started normally but did not start listening on. Now it will throw a configuration error/alert.

Test SQL connection in the installer showed an error when the database had already exist

The Installer displayed a SQL error when the Genesys Metadata Template was already assigned to Groups. To fix the problem just run the installer/scripts again.

Installer did not detect partitioning support on Azure SQL Database and Azure SQL Managed Instance

Multiple issues fixed for the PowerShell Deployment Toolkit: - The PowerShell Deployment Toolkit included the wrong version of the lyncfilter.dll - The PowerShell Deployment Toolkit used the old hashing algorithm for the user passwords created during the install process, therefore it wasn't able to update them. - File copying failed randomly in the PowerShell Deployment Toolkit. The Toolkit now uses xcopy instead, which is more stable than the PowerShell Copy-Item cmdlet. - The PowerShell Deployment Toolkit failed to install the database when using windows credentials. - The certificate generation failed randomly in the PowerShell Deployment Toolkit, because the CNG Key Isolation service wasn't running. The script now checks and starts the service.

Microsoft SQL Express installer did not work from the Prerequisite Tool

The CDR reconciliation process did not take into account correctly the direction configuration of the extensions

Lync/SfB CDR import on the rare occasion of reconciliation was not set in the data management policy, the customer could see a long reconciliation cycle until 2100.1.1.

Migration from Verint v11 did not pull all of the Verint "Media" as Verba "Storage Target" so calls possibly could be migrated without a Storage Target property

Generic Import from CSV files showed incorrect information about the total number of importable calls

Verba import did not work for previously exported calls from a Verba system when calls were encrypted originally and were decrypted during export

Verba Import did not work for recordings where there was a partial mismatch (e.g. timestamp) in the name of the media file and metadata XML file

Generic Import XML reading issue in case of single line XML files

In the case of multi-tenant setup, the imported media and metadata XML files were not put under the tenant ID specific folder

Verint Import Source had crashed when a custom metadata field was updated in the database and the database connection was already in use

When using the Verint Import Source, not all calls were properly assigned/mapped to users

Cisco Webex Teams API authentication token was not refreshed in certain cases and caused the import to stop working, no alert was raised

Cisco Webex Teams instant message export did not work for malformed database fields

Verba import did not work when the standard metadata template type was used in the source system

The generic import source was setting the Native Call ID wrong and the information was not stored in the database

RingCentral import source did not handle API throttling responses

Ringcentral import source did not handle API throttling responses

Migration from Verint v11/v15 may chose an unsupported (tape) storage during migration, and that led to an empty Storage Target

HTTPS connections could not be established when HTTP headers exceeded a certain size

Verba Offline Player could not playback Microsoft Teams video (or any other where "unknow" directional audio streams were present)