First, the general rules of fault finding
Ethernet troubleshooting can be divided into five steps:
1. Collect all the valuable information that can be collected, and analyze the phenomenon of failure.
2. Locate a fault to a specific network segment, either a single standalone function group (module) or a user.
3. Verify that a specific hardware failure or software failure.
4. Positioning and repairing faults.
5. Verify troubleshooting.
Introduction to Serial Ethernet Technology
Serial port Ethernet conversion equipment and modules, because taking into account the existing serial port network, communication, and into the Ethernet network there are a variety of new situations (such as IP address allocation, domain name resolution, network security, network performance), so need to apply A lot of new technology, we have a product on the market serial Ethernet products, for example, introduce these technologies.
Here are a few of these representative of the representative of the technology:
For existing serial communication networks (or separate serial devices), how to upgrade to the lowest cost of the network, is a question worth considering: the existing equipment is not easy to re-add the circuit, the newly developed products do not have the time and effort to develop the network Interface, it is best to have a way to make everything look exactly like the previous serial port, but also use the Ethernet.
Third, the typical network failure
Fault 1, can not access the server or a service
It is normal to set up a server or a service here and have done the following:
Re-start the PC (hot start can not reset all the adapter card).
Verify that the PC does not have its own hardware failure.
Make sure that all network cables are connected correctly.
Make sure that all network card drivers are loaded properly and no errors are reported.
To confirm that the server or service has not changed, such as reconfiguration to increase hard
To test whether the failure affects the workstation (local failure) or other sites (large-scale failures), this can be demonstrated by loading the server or service from another workstation. These workstations are on the same network segment or on the hub. If the fault is on the same network segment or another site on the Hub, try to test from another Hub or site on the network.
(A) local failure
Before making a hardware fault lookup, make sure that other users can not log on from this machine, which eliminates the user account error. For a single site, typical failures occur on bad cables, bad NICs, driver software, or incorrect workstation settings.
(B) global issues
In general, physical layer failures in coaxial networks can lead to catastrophic network failures. Use the "dichotomy" to find such a failure can be quickly resolved to solve. Intermittent failures are more difficult to isolate.
Half of the problem comes from the cable
(1) Visual connectivity
Check the connectivity commonly used method is to check the Hub, transceiver and the recent production of the status of the card on the card. If it is 10BASE5 cable, to carefully check all the AUI cable is firmly connected, the lock to be locked at the same time, a lot of problems as long as the simple part of the re-tight again to solve the problem.
(2) damaged cable or connecting parts
When you check the physical layer of the problem, pay attention to the damaged cable, the wrong cable type (such as Ethernet RG62 or RG59), did not play the RJ45 head, not according to the prison BNC head. Cable for suspected problems can be tested with a typical cable tester.
(3) connection pulse polarity problem
Both the NIC and the Hub's connection pulse polarity can be measured with the test, and the connection polarity fault is usually caused by a wiring error in the cable.
Check the link layer problem
(1) collision problem
If the average collision rate is greater than 10% or a very high collision is observed, further testing is required. If possible, try to isolate the problem area by reducing the size of the segment (dividing the network into segments) and monitoring changes at any time. In order to track the collision, you must know the network traffic. You can use the background traffic generator to add the appropriate amount of traffic (100 frames per second, 100 bytes of traffic), and observe the network statistics display at the same time. Some media-related failures are proportional to the size of the traffic. You can use the control keys to change the flow while observing the collision and the wrong change. This is especially the case, because you can easily add heavy traffic to the network. Solving the problems associated with collisions is often very laborious because the test situation depends to a large extent on the position of the observation. Perhaps in the same network segment a few feet away from the different observation points to see the situation is different, to find a few points to test and pay attention to the changes that have occurred.
If the collision is proportional to the flow, or the collision is almost 100%, or almost no normal flow, it may be a problem with the cabling system. For UTP cabling, disconnect the cable from the Hub and perform a cable test. For the coaxial cable to carry out impedance measurement, you can use the digital meter or other instrument DC off function to test. If both ends of the cable have a terminator, the T-connector should be measured at about 25 if 50 is measured from one end of the cable.
(2) frame level error
If there is a frame level error, you will run the error statistics test, and through the detailed search function to find the workstation's MAC address, and then tested to determine the fault down. You can try to use the "clean" drive to reload the workstation into the workstation, to confirm the configuration is complete. If all this does not work, you can try to replace the doubtful card.
(3) utilization is too high
If the utilization is too high (the average is greater than 40%, the instantaneous peak is higher than 60%), then the network segment load is too heavy. You should consider installing a bridge or router to reduce traffic in the segment or divide the segment into smaller segments.
The integrity of the client server connection
If the link layer is intact, then we have to look at the agreement whether there is any problem will affect the communication between the server and the customer.
(1) connectivity
Using a professional meter, you can run the Server List or Novell Ping test function to verify that the server or service function can be accessed from the test point. If the server does not respond, re-start the server and pay attention to observe all the necessary driver software is loaded correctly, with or without error. Novell's version 3.11 and previous versions, using the 802.3RAW package in the default configuration, and version 3.12 and later use the 802.2 package type in the default configuration.
In addition, you can try to ping the server several times to confirm the request signal and return the number of response signals equal, the results are inconsistent indicates that sometimes bad card or Hub failure caused by the loss of the frame. Critical states and bad bridges or routers can also be used to locate quickly.
(2) the accessibility of the server or service
If you use a protocol analyzer, you will need to capture a 3 to 4 minute packet for analysis. Look at whether there is a delay request from the server and find out which server, if there is a delay request, then the server can not fully handle the task loaded, each request to cancel a request.
(3) network response is slow or poor performance
First of all, to determine whether it is related to the network media failure, is a workstation problem or a server or service problems. From the other workstations to visit the same server or service, and asked the network or other users on the network segment, if the problem is not related to the media, to determine whether the problem is software, check the network utilization and software errors to find performance Poor network problems.
Run tests such as network statistics to see if there are high traffic and abnormal high collisions, and then run these tests after increasing the background traffic. If the collision and PCF are found to increase as the traffic increases, you should run a cable test to check the connectivity of the workstation.
sales 