Chat Server Administration

Review the following recommendations for the administration of Chat Server. See also the following topics:

• Deploying High-Availability Chat Server
• Matching Contact Attributes
• Multilingual Processing

Limitations

The following limitations are recommended for a Genesys Chat solution running on a single host with two Intel Xeon 3.0 GHz processors. Observe these limitations for optimum performance (meaning without significant delay).

Item	Maximum
Message size	4 KB (Genesys Desktop limitation; Chat Server does not have this restriction.)
Transcript size	54 KB (Genesys Desktop limitation; Chat Server does not have this restriction.)
Concurrent sessions (in a realistic simple scenario)	1000 per Chat Server
Messages per second	50 (rare temporary peaks up to 150)
Sessions opened and closed per second	10 (rare temporary peaks up to 30)

You can configure a timeout in Chat Server that prevents keeping unused connections open. Use the user-register-timeout option (default value 30 seconds) to set the maximum time between:

• Receiving a registration over the socket
  
• Receiving a flex packet over the socket
  

Connection Delay with Antivirus

It may take some time (up to several minutes on some UNIX Platforms) for Chat Server to connect to an unopened port on a Windows host that is running an antivirus program. For example, if Chat Server is running on Linux and is trying to connect to an inactive UCS instance, it could take up to three minutes for Chat Server to detect that the listening port is not open.

Setting Load Limits

Starting in the 8.5.0 release of Chat Server, you can impose load limits on Chat Server: when Chat Server reaches the specified limit, it no longer creates new sessions or restores existing sessions.

Set load limits using the following configuration options (full descriptions are in the eServices Options Reference):

• Enable or disable the general functionality of load limitation using limits-control-enabled.
• Set specific limits:
  • limit-for-flex-users—Maximum number of currently logged-in flex users.
  • limit-for-reply-delay— Configures the maximum average delay (in milliseconds) for processing requests. The average value is calculated on the limit-average-interval interval. This delay increases if the Chat Server instance is overloaded with a large number of incoming requests.
  • limit-for-sessions—Maximum number of concurrent chat sessions.

  If any of these limits is reached, Chat Server stops creating and restoring sessions.

• If Chat Server is configured in primary/backup mode (not recommended; see Deploying High-Availability Chat Server), you may want to stop it from reporting service unavailable to SCS when a limit is reached. You can do this using limits-reached-report-scs. Blocking the reports avoids a scenario in which Chat Server in primary/backup mode closes a chat session because (1) Chat Server detects disconnection from Interaction Server and/or UCS, (2) it sends a service unavailable notification to Solution Control Server, (3) SCS switches Chat Server to backup state, which closes the chat session. (This scenario does not apply if Chat Server is in N+1 mode: multiple Chat Servers with no backup configured).
• Set the point at which Chat Server returns to full functionality using limits-restore-threshold. This value is a percentage of the limit set by the three limit-for-X options.

Example

If limit-for-flex-users is set to 400 and limits-restore-threshold is set to 80, then:

1. When the number of flex users reaches 400, Chat Server stops creating and restoring sessions, and rejects login attempts by flex users.
2. When the number of flex users falls to 320, Chat Server returns to full functionality.

Masking Sensitive Data

Chat Server logs and chat transcripts might contain sensitive data such as credit card numbers, phone numbers, Social Security numbers, and so on. You can omit this data from logs and mask it in transcripts.

Logs

To omit sensitive data from logs, you must configure both UCS, as described elsewhere, and Chat Server, as follows:

• In the [settings] section, set message-log-print-size to 0. This means that logs do not show the messages sent between chat participants. Where a message occurs, the log shows [truncated from size=x], where x is the number of characters in the suppressed message.
• In the [log-filter-data] section,
  • Set StructuredText to hide so that logs will omit the transcript that UCS sends to Chat Server.
  • Set Transcript to hide so that logs will omit the transcript that Chat Server sends to UCS.

Chat Transcripts

Chat Server can mask sensitive data in transcripts by using a regular expression (regex) to find and substitute the data with a configurable replacement character. This functionality enables you to:

• Examine each chat session message with an ordered set of regex rules. Use the apply-config option to configure the source/location of regex rules that will be applied. Note: all options are located in the section [transcript-cleanup].
• Replace any part of the message that matches a regex rule with a replacement character specified by the configuration. The default is specified by the default-repchar option.
• When replacing symbols you can choose to replace all symbols or only digits. When replacing digits, you can also leave the last few digits unmasked —see the default-spec option.

This functionality can be applied in the transcript messages of an ongoing chat session and/or a transcript saved in the contact history (UCS). This is specified by the apply-area option.

Important

Prior to the 8.5.103 release, Chat Server used different options from [settings] for this functionality. Click here to view the previous description.

.

If the apply-config option has a value of cfg or is set to mix and no UCS PII configuration has been provided for the given chat session, Chat Server uses the following default rules to find sensitive data:

	Name	Regular Expression
Credit card	GCTI_CreditCards	(?>^|(?<=[\s[:alpha:](),.:;?!"'`]))(?>4\d{3}|5[1-5]\d{2}|6011|622[1-9]|64[4-9]\d|65\d{2})[ -.]?\d{4}[ -.]?\d{4}[ -.]?\d{4}(?>$|(?=[\s[:alpha:](),.:;?!"'`]))
Social Security number	GCTI_SSN	(?>^|(?<=[\s[:alpha:](),.:;?!"'`]))(?!000|666|9)\d{3}[- ]?(?!00)\d{2}[- ]?(?!0000)\d{4}(?>$|(?=[\s[:alpha:](),.:;?!"'`]))
Phone number using the North American Numbering Plan	GCTI_PhoneNANPA	(?>^|(?<=[\s[:alpha:](),.:;?!"'`]))(?:\+?1[-. ]?)?(?:\(?[2-9][0-9]{2}\)?[-. ]?)?[2-9][0-9]{2}[-. ]?[0-9]{4}(?>$|(?=[\s[:alpha:](),.:;?!"'`]))

Tip

As of release 8.5.103, in addition to the method of configuration described on this page, Chat Server also supports reading the regex rules from UCS. To do this,

• Set the apply-config option to mix or ucs.
• Use Privacy Manager, a plugin for Genesys Administrator Extension (GAX), to select and activate these rules.

Typing Preview

Typing preview allows an agent to see text that a customer types before the text is submitted to the chat session. You can have Chat Server mask all digits in the typing preview by setting the typing-preview (called transcript-cleanup-typing before release 8.5.103) option to a value other than none. Chat Server then replaces all digits in the typing preview with the character specified by default-repchar (called transcript-cleanup-mask before release 8.5.103).

Inactivity Monitoring

Inactivity monitoring closes the chat session if there is no activity by chat participants after a certain length of time. To enable this control, set the enabled option in the [inactivity-control] section to true.

• Chat Server carries out inactivity monitoring only if at least one customer and one agent are participating in the chat session.
• If there is no activity during the time specified by timeout-alert, Chat Server issues a warning comprising the text specified by message-alert. The default warning is "Chat session will be closed soon due to inactivity of chat participants."
• If there is no activity for another timeout-close seconds, Chat Server issues a notification consisting of the text specified by message-close and closes the chat session (and removes all participants from it). The default notification is "Chat session closed due to inactivity of chat participants."
• If any activity occurs, Chat Server resets both of the timeouts. Activity means any activity in the chat session that is visible to all participants—so, for example, coaching messages between agents do not count as chat activity.

KPI (Key Performance Indicator) counters

Starting with release 8.5.103, Chat Server includes KPI (Key Performance Indicator) counters that monitor activity within the server.

Accessing KPI counters

Access KPI counters in one of two ways:

• The Chat Server log. by configuring options log-output-content, log-output-proviso, log-output-timeout in the [health-service] section.
• The web REST interface which you can configure by:
  • Adding a port with the ID=health to the ports of Chat Server.
  • Adjusting the soap-* options found in the [health-service] section.

Web interface

Access the web interface through the following URL format: http://ServerName:ServerPort/Counters?<list of parameters> where:

• ServerName is the host name where Chat Server is running.
• ServerPort is the web service port, also specified as the health port of Chat Server.

Web interface parameters:

Name	Description	Valid Value	Default Value
metadata	Returns a list (in JSON format) of all supported counters, with descriptions.	true, yes false, no	false
content	Returns a list (in JSON format) of counters according to the provided parameter value.	all—returns all available counters new—returns only recently updated counters set —returns all initialized (non-zero) counters	all
reset	Resets all counters to zero.	true, yes false, no	false

If a parameter is omitted or has an invalid value, then the default value is used for that parameter. Parameters are processed according to the following rules:

• If reset is true, then all the counters will be reset and then the rest of the parameters will be processed.
• If metadata is true, then the content parameter will be ignored and the metadata will be sent.
  

Example URLs:

• http://hostname:7000/Counters?content=set
• http://hostname:7000/Counters?metadata=true&reset=true

How counter values are calculated

All counters, except the process memory counter, are cumulative. The value begins to accumulate the moment the application is launched or the counters are reset. To calculate the difference, the user must use two sample counters from different times and subtract the earlier sample from the later one. To find the counter's rate value, divide the difference by the number of elapsed seconds between the two samples.

Important

On some platforms, the time for processing internal activities may be reported as zero. This does not indicate an issue with the counter. On the contrary, a rapidly growing value on the counter for internal activity indicates that the server is overloaded.