Troubleshooting

Table 4-1: Common Problems

Problem	Take these Actions
Subscriber Access message inventory is slow	The more messages there are in the subscriber's inbox, and the larger they are, the slower the message inventory process will be. Suggest that subscribers delete messages or save them locally on their PC. Inventory functions will be slow during a broadcast message distribution (messaging server processor limitation). Any processor intensive operations on any server used by the subscriber's uOne session will also slow down functions (processor limitation).
Message delivered late or not at all	1. Is call from a remote AMIS-A system? Check the remote AMIS-A system for problems. Is the message addresses to a remote AMIS-A system? Check the AMIS-A queue on the messaging server for back-log of messages. Telnet to the proper messaging server and change directory to the AMIS-A administrator mailbox. Check the inbox and AwaitTran folders. Notify subscriber that the queue must clear for delays to be resolved. 2. Was the message a broadcast message? If so, the MWI notification will not occur. When another message is delivered, turning on the MWI indicator, the earlier broadcast message may appear as if it were delayed. 3. Check the time of day on different servers. Their time of day clocks may be out of synch, giving false delivery times. Cisco recommends the use of NTP to keep the clocks in synch. 4. Gather details to analyze log files. Ask the subscriber: How do you know the message was sent? Who sent it? When was it sent? Note   Messaging services are inherently store and forward systems. Any added load conditions, such as sending broadcast messages, can increase delays.
During Caller Access, the caller does not hear the called subscriber's greeting or spoken name	The subscriber must record and activate an All Calls greeting, or both a Busy greeting and No Answer greeting, or activate one system greeting. The subscriber must record a spoken name.
Quota exceeded	The subscriber should either delete messages or save messages locally on their PC. The quota for an individual subscriber can be expanded through uOne Administration.
MWI doesn't turn on when a message is received	1. Was the message a broadcast message? MWI is not designed to come on for such messages. 2. See if the MWI_Notify and MWI_OnOff application agents are running on the Gateserver on which they are configured. Log on to the machine that you want to check as spmaster. From the command line type: umcli watch acb globals exit 3. Look at the counts for MWI_OnOff and MWI_Notify. Type: umcli watch notify exit 4. Check the Event logs to see whether notifications are being sent (MWI application) or whether coming in/out of the RS232 port (logs generated by CMA).
Saved message no longer there	Check the Event Log to verify that the message was actually saved, and not inadvertently deleted. Was the message a broadcast message? Broadcast messages can deliberately expire after a period set by the sender.
MWI doesn't turn off	1. Ensure that all new messages of the type and severity for which the subscriber has configured MWI notification (e.g., normal and urgent voice and fax messages) are saved or deleted. uOne requires an explicit save or delete command to move a message out of the "new" state. Have the subscriber save or delete the messages as appropriate. 2. Reading such messages with an e-mail client will not turn off MWI. The subscriber must log in to and exit the uOne application to turn off MWI.
Blank messages are received	The most common cause of blank messages is callers attempting to leave a message through a speaker phone. Due to background noise, callers often do not hear the beep indicating that uOne is recording. To verify that a "blank" message is simply background noise, examine the .wav attachment to the message that is in the subscriber's mailbox.
Calls are not answered	The CMA agent may be down. 1. Use umcli, uOne Manager, or SNMP to view the status of CMA. 2. Start CMA if it is down. Note   To prevent the need for manual restart of the ACB and agents, such as CMA, you should use the AutoRestart feature.
Subscriber hears "dead air" when dialing the access number	1. If there are complaints from multiple subscribers, check the directory server. 2. Check the CMA Sof logs for call allocation failures. 3. Disable any VoIP dial peers on your Gateway that refer to the CMA host. 4. After calls have been diverted from CMA, cycle CMA to clear its state. Note   This step may will impact services, and should be done only during "off" hours, if possible. 5. Re-enable the disabled VoIP dial peers.
Paging not working	1. Is the problem the user, the whole system? Is the queue backing up? Use the umcli tool (watch notify). Check the Event logs. 2. If the subscriber is provisioned for numeric paging or SNR paging, ensure that the HylaFAX Paging Server processes are running and the modem pool is functional. 3. If the subscriber is provisioned for alphanumeric paging (SMTP), check the SMTP paging server. 4. Determine: Has subscriber ever received pages? If yes, has subscriber ever received pages from uOne? If no to either of first two questions, which paging provider is used by the subscriber? When was the paging notification sent? By whom? Where was the subscriber when the page was sent? Is the subscriber at the same location? What is the PIN (pager ID)? 5. Dial end-user pager from a telephone and send a page. If page does not go through, end user could be in a location that blocks pages, or the paging provider may be having problems. 6. If the page from step 5 was received, try paging from uOne message notification. 7. If page from step 5 was not received, call the paging provider to verify the paging information required for provisioning; update it if necessary. 8. If the page is not received, then ask the paging provider technical support service for a modem dial-in number (IXO/TAP) and the required modem settings to use it. Call this number to verify that it is a modem line. 9. Check the paging server configurables particularly the pagermap file on HylaFAX and modem settings; update as necessary. Check the paging server log files. 10. Try sending a page directly from the paging server. Telnet to the server port or by using test commands, such as the HylaFAX send page command. Send e-mail to the SMTP paging server's e-mail address for a subscriber. 11. Check network connectivity between the Gateserver and paging server.
Subscriber login fails	For a single subscriber: 1. Use uOne Administration to change the subscriber's voice mail password, and/or reset the subscriber's bad login counter to zero. Ask the subscriber to log in. Note   When dealing with login and passwords, it is the responsibility of the service provider to ensure security. Your corporate policies will dictate the proper procedures for handling the change described above. For more than one subscriber: 1. Test the messaging server directly to ensure that it is accessible over the network. Bypass the Gateserver using any IMAP client (such as Netscape Communicator) to log in with a subscriber's e-mail login and password. Ensure that the messaging server processes (e.g., imapd) are running. If necessary, restart the messaging server processes. 2. Test the directory server directly to ensure that it is accessible over the network. Bypass the Gateserver using any LDAP search tool, such as those provided in Netscape Communicator to perform a search on the subscribers telephone numbers. Ensure that the directory server processes (e.g., ns_slapd) are running on the directory server. If necessary, restart the directory server processes.
uOne gives upper-level menu all the time (The upper-level menu is an initial uOne menu offering the choices: (1) to leave a message, and (2) to login).	1. Test from a phone known to give the correct menu. 2. Ensure that NetworkInfoExpected is set to Yes in the UM.ini file of each Gateserver. 3. If you are using H.323, use watch h323 states to view the caller RDNIS and call type are set to non-empty and non-* values for forwarded calls.
An attempt to send a fax to a mailbox fails	Ensure that the on ramp MMoIP fax gateway is configured properly and that the subscriber has a proper inbound fax number alias of the form FAX=<n><d>, where <n> is their personal fax number and <d> is their e-mail domain.
Server agents won't start up after a failure	There may be no more space free in the /home2 file system. 1. Check space with df -k. 2. If space is low: a. Remove the log files in stderrout. b. Remove Sof log files. c. Ensure that the LogRemover application is configured properly.
Single Number Reach doesn't work when enabled	1. Check the Event logs to see if any attempts were made or if there are any indications of failures. 2. Note that Single Number Reach is not designed to activate for the call type BUSY. 3. Make sure the subscriber's phone line is administered on the PBX/switch to go to coverage on No Answer or send all calls and that uOne is the first coverage path. 4. Verify device type for reach numbers: If there is an answering machine at a reach number and it picks up after 4 rings, SNR should be configured to ring only 3 times. If cellular phones are being used, the maximum number of rings should be set to at least 5. On average if the number of rings is set to 5, the cellular subscriber is hearing only 3.
Subscriber Access inventory says "no new messages" but MWI is on	Message may have been marked as "read" on PC when the subscriber checked e-mail. (This does not cause MWI to be turned off.) uOne will treat this as a "saved" message and add it to the Saved Messages list. Log in to and exit uOne to turn off the MWI.
Single Number Reach doesn't work with cellular phone	1. Check the Event logs to see if any attempts were made or if there are any indications of failures. 2. Typically, it takes longer to connect to cellular service. Using uOne Administration Modify a Subscriber, set the Single Number Reach Allowed Number of Rings attribute to at least 5 rings. If calls still don't get through to the cellular phone, use the ldapmodify command to set the snrringnumber attribute to 6, 7, or 8. See directory server documentation for information about the ldapmodify command.
Playing a voice message or play out of a greeting is delayed by several seconds (Specifically because the IMAP fetch preceding the play is delayed)	Such delays may be caused by bulk data transfer flow control problems resulting from mismatched Ethernet link modes, typically when not all links are 100 Mbps full duplex. If you have a Catalyst 2900, use the Mode button and the LEDs to ensure that all links are consistently 100 Mbps full duplex. If all links are not consistent, you can force consistency via the Ethernet switch, gateway, gatekeeper, or Solaris configuration, or via the Ethernet port and the device to which it is connected. For the Cisco Ethernet switch, gateway, or gatekeeper, use the IOS interface fastethernet configuration to force 100 mbps full duplex on the mismatched switch port or gateway/gatekeeper interface. For a Solaris host, use set commands in the Solaris /etc/system configuration file to force 100 Mbps full duplex on the mismatched hme interface. See the Solaris ndd man page for information about the hme driver configuration parameters for auto negotiation, speed, and full/half duplex.

.
Table 4-2: Isolating the Problem

If You Suspect	Take these Actions
a hardware problem.	Check for failed components on all machines in your system configuration. Check all physical connections, e.g., power and data cables.
a network problem.	Do spot checks, e.g., test name service, ping network nodes, send e-mail, telnet to server ports. Check Ethernet full vs. half duplex and link speed.
a software problem.	Check for operating system errors (e.g., in syslog log files or on the consoles). Check for errors in 3rd party software (Netscape Directory Server, Netscape Messaging Server, HylaFAX Paging Server, etc.). On the messaging and directory servers, check the syslog log files for any errors. The location of such log files is configured in /etc/syslog.conf. These log files may be large on an messaging host with MWI_PlugIn and SMS notify errors. Check for errors in Cisco software (ACB, CMA, UM, MWI_Notify, MWI_OnOff, FaxPrint, LogRemover, SMS_Notify and MWI_PassOff processes, and the MWI_PlugIn and notify plugin libraries). To do so, use your SNMP Manager to view the SNMP trap messages. View the log files. Check for amisain and amisaout errors on the UM 3.0 hosts. Check the configuration files on the Gateserver(s) for recent changes. Run the following command on the Gateserver(s): umcli watch acb entries exit Report to TAC any UM processes running longer than 2 hours. Run the following commands on the Gateserver(s): ls -l /home2/spmaster | grep core. ls -l $PARMLIB/logs/cfiles/* |grep core Report any core files found. Check root and spmaster mail on the Gateserver(s). Ensure that no items exist in the root or spmaster dead letter file. Use the umcli watch command to view the real-time status of uOne resources. Use vmstat, mpstat, iostat, top, and df to view operating system resources.