Quick problem isolation and resolution is one of the most important practices in a successful network maintenance strategy. The words sound simple, but reality can provide some hair-pulling problems for network engineers and technicians.
Isolating and analyzing these problems require the proper troubleshooting tools. Here are the top 10 network problems encountered today and some specific troubleshooting methods that you can use in your overall strategic network maintenance program.
10. The Mysterious Case of the Dropped Sessions
Complaint: Slow service and intermittent dropped sessions.
Symptoms:Only present on one workstation and occurs regardless of which server is being accessed.
Disconnect the patch cord from your workstation and connect an active tester (transmits traffic as part of the diagnostic procedure), such as the Fluke LANMeter, a portable network management tool. Determine if the network segment is healthy by monitoring the utilization levels, collisions and the presence of error frames. The connection typically will pass a passive monitoring test but fail an active one, indicating a physical-layer problem.
Run a cable test on the entire installed link, including the patch cords. In this case, the length measurement may fail, showing more than 100 meters for unshielded twisted-pair (UTP) cable or more than 185 meters for coax. This indicates that the IEEE maximum allowable lengths have been exceeded which can cause severe attenuation of the LAN signal transmissions.
Problem Resolution: Install a repeater in the middle of the cable run to create two legal lengths of cable, or replace the copper or coax cable with a fiber-optic link.
9. I’ve Fallen and I Can’t Log On!
Complaint: Cannot log onto the desired server.
Symptoms: Usually occurs on a new installation or after a workstation is reconfigured. The problem is present on only one workstation in the area.
Troubleshooting Approach: Disconnect the patch cord from your workstation and connect a portable network verification and diagnostic tool, such as the Fluke OneTouch, or portable network management tool. Perform an Internet Packet eXchange ping or obtain a server list and check what network layer protocols the server is configured for.
Note the encapsulation frame type (IEEE 802.2, for example). Now perform a network-interface card test to verify that the driver loads and operates correctly. Again note the frame type bound to the card (IEEE 802.3, for example). These two frame types must match. If the server is not running the same frame type as the workstation, there can be no connection.
Problem Resolution: Open the workstation’s NET.CFG file with a text editor and change the line FRAME=ETHERNET_802.3 to FRAME=ETHERNET_802.2. Now connect the workstation to the network and reboot. Verify the server connection.
8. Does This Patient Have a Heartbeat?
Complaint: Slow access to network services during peak hours.
Symptoms: Affects all workstations on a particular segment regardless of whether or not the server is local to that segment or on a remote segment.
Troubleshooting Approach: Connect your network verification and diagnostic tool into the hub of that network segment during peak usage hours. Run a network health test on the network segment and note the levels of utilization, collisions, error frames and broadcast frames. In this instance, the average number of collisions is likely to be unusually high. Possibly up to 50% of the transmissions are collision fragments. This is evidence that a transceiver’s signal quality error (SQE) “heartbeat” signal has inadvertently been left in the on position.
Problem Resolution: Locate each transceiver attached to that segment and verify that the SQE switch is in the off position. This signal originally was used on a 10BASE-5 Ethernet network, but is not needed on new Ethernet equipment.
7. Wire Bundles That Get Bumped in the Night
Complaint: Dropped connection or cannot log on.
Symptoms: Affects only one workstation on the segment.
Troubleshooting Approach: Disconnect the patch cord from the workstation and connect a portable network verification and diagnostic tool or a portable network management tool. No link pulse is detected from the hub.
Run a test to see if the cable is open between the office wall plate and the wiring closet. If the length seems right, use the tool or a tone test set to generate a tone on the cable.
Move to the wiring closet and use an inductive amplifier to pick up the tone and identify which connection goes back to that office. When things are changed in a wiring closet, the wire bundles can be jostled and cause a connector to come unseated.
Problem Resolution: Reseat all connectors in the path between the hub and that station. Verify that the link pulse LED lights. Then make sure that you can log on.
6. Fiber-Optical Illusions
Complaint: Generally slow network operation.
Symptoms: Affects all stations connected to one segment.
Troubleshooting Approach: During normal network usage hours, connect a portable network verification and diagnostic tool or portable network management tool into the hub of that network segment. Have traffic present on both sides of the connection and test the health on each side of the connection, one at a time, when the fiber is connected and disconnected.
This problem will exhibit itself as large numbers of error frames whenever the fiber-optic repeater port is included in the segment. Poor fiber-optic link connections cause junk to be transmitted on the attached segment, forcing stations to resend frames.
Problem Resolution: Clean or reterminate the fiber connectors, reseat all connections and check the network health again. There should be very few error frames present.
5. The IP Network Administrator’s Nemesis
Complaint: Intermittent failure, unable to log on or cannot access remote server.
Symptoms: Affects only one station and usually occurs during adds, moves or changes.
Troubleshooting Approach: To eliminate potential problems with your account and security, log on and attempt the same operation from another working station. If that succeeds, use a network verification and diagnostic tool or network management tool to verify the link from your patch cord to the desired server or service. Then perform a network interface card (NIC) test to verify the hardware and driver functionality.
If all checks out, examine the Internet Protocol (IP) configuration to ensure that your IP address is valid and not a duplicate, that the subnet mask is right for this segment, that the default router address points to the right router for this segment, and that the right domain naming system server address (if used) is specified. An error like configuring a new workstation with the same IP address of a print server can route everyone’s print jobs to the new client.
Problem Resolution: Make the appropriate changes to the configuration and reboot the workstation.
4. Daisy Chains Are Only for Gardening
Complaint: Generally slow network service.
Symptoms: Affects all users connected to one Ethernet segment. Network service is slow regardless of which server you are connected to.
Troubleshooting Approach: Connect a network verification and diagnostic tool into the hub for that network segment during peak usage hours and run a network health test. In this instance, many error frames are present.
On a 10BASE-2 network, the type of error frame may be classified as a late collision. On a 10BASE-T network, the type of error frame typically would be a frame- check sequence error. As networks grow, new hubs often are added by connecting them to existing hubs to create a daisy-chain effect.
Problem Resolution: Count the number of repeater “hops” between any two distant network devices. There should be no more than four hops between any two stations on a 10-Mbs Ethernet network or no more than two hops on a 100-Mbs Fast Ethernet network. Correct the network topology to abide by these Ethernet standard rules, then verify that the error frames no longer exist.
3. Bad Watchdog
Complaint: Dropped connections to a Novell NetWare remote server.
Symptoms: Affects all workstations in this segment connected to remote servers.
Troubleshooting Approach: Determine the health of the network segment and note the levels of utilization, collisions, error frames and broadcast frames. In this instance, many collisions or other errors could cause delays of “hello” frames between the client and the server. Repeat this test on the remote segment where the desired server resides.
If both segments appear clean, use a portable or fixed network management tool to query the signaling network management protocol agent(s) within the router(s) in the path between the two segments. Make sure that no congestion problems are along this path. To use resources efficiently, NetWare servers will log off clients who do not respond to these “hello” frames before a watchdog timer runs out.
Problem Resolution: The short-term solution is to change the “set delay between watchdog packets” parameter in the Novell server to a longer time. If the situation improves, locate and correct the cause of lost or delayed watchdog timer frames.
2. My Fast Ethernet Is Slower Than My Slow Ethernet
Complaint: Slow service during peak hours on 10-Mbs Ethernet or no connection on a 100-Mbs Ethernet.
Symptoms: Can affect new workstation connections or anyone who upgrades to Fast Ethernet.
Troubleshooting Approach: For plants wired with Cat 5-rated UTP cable, use a cable meter that can certify the installed links to meet the EIA/TIA TSB67 standard for 100-MHz data transmission rates. Some installed links work fine at 10 Mbs but fail at 100 Mbs due to excessive near-end crosstalk. The transmit wire pair tends to couple signals onto adjacent wire pairs that cause high-frequency transmissions to fail. Depending on the traffic load, any UTP cable misterminations resulting in split pairs cause even low-frequency LAN transmissions to fail.
Problem Resolution: Replace or reterminate any cabling links that do not pass. After you make these fixes, retest all failing links.
1. Problem: Arrest Those Physical (Layer) Abusers
Complaint: Cannot log on or intermittent operation.
Symptoms: Affects only one workstation that has been operating with no previous problems.
Troubleshooting Approach: Cable terminations only last so long. Users move, bend, spindle and mutilate network terminators every day, especially laptop owners. An inexpensive cable meter can identify this problem. But before you can safely say that there is a cabling problem, find out what it is not, namely all other potential sources of errors.
This requires using a personal network verification and diagnostic tool or portable network management tool to verify that the health of the local network segment is good, that the hub port is functioning, that the NIC and its respective drivers are all okay and, finally, that no problems exist with the network interconnect devices. Only then will you know where to begin performing cable tests to isolate the marginal termination, broken connector or ground-loop problems.
On coax networks, look for stub cables connected between the coax T and the NIC. On UTP networks, run a wire-map test in a continuous mode and have someone wiggle all of the terminations to identify intermittent connections or shorts.
Problem Resolution: Replace the bad patch cord or the bad connector or reterminate the horizontal wire run as necessary. Double-check that the fix takes care of the problem entirely.
Tools equal to the troubleshooting task at hand are necessary to identify the source of common network errors. Along with education on how to apply the tools to isolate problems, this is a cornerstone of any network maintenance strategy.
About the Author
John Giles is the senior technical support manager for the Networks Division at Fluke. He teaches network technology seminars and is responsible for maintaining Fluke’s research and development network. Fluke, P.O. Box 9090, Everett, WA 98206-9090, (206) 347-6100, [email protected].
Copyright 1997 Nelson Publishing Inc.