Shulou(Shulou.com)06/01 Report--

I haven't written a diary for a long time. I have come to the 51cto Forum for such a long time and learned a lot from my predecessors. For many problems encountered in practical work and the process of dealing with them, it is also necessary to learn from each other and learn from each other to make up for their weaknesses. Take advantage of this time and opportunity to share an example of a strange network problem that occurred in the company a few days ago.

Here a brief account of the background, I am in charge of IT work in a brand 4s store (more than a month), due to some reasons, also take into account the IT maintenance of another 4s store (in fact, the two stores belong to the same group, the former is called A store, the latter is called B store). I work in store A, but the equipment in store B is relatively old, and the combined workload of the two stores is actually quite a lot (mainly because there are too many things managed by IT).

Just before lunch, I was told to change a ribbon rack for a needle printer in store B (seeing that the picture sent was a broken ribbon rack, I couldn't help but observe a moment of silence for it). When I was about to get off work, I found a ribbon rack of the same model, put on a new ribbon, and found a printer in store A to test it. I planned to change it into a ribbon rack when I ate in the past. (the canteen of A store has not been built yet. All the staff dined in the canteen of shop B.

While installing the ribbon rack, the beautiful girl next to her suddenly said that she could not get on the net, and then received all kinds of notices that she could not get on the net. I was surprised at that time. I was installing the ribbon and never touched anything related to the network. Why did I suddenly say that the network was cut off? At this time, it is generally a time for network managers to have a headache, because the disconnection of a large area of the network is more troublesome and urgent, on the one hand, it will affect the normal business of the company, and on the other hand, all kinds of people will urge when to get it done and whether they can do a good job or not. All right, stop gossiping and talk about the process.

First of all, it is necessary to determine which and how many computers are off the network, all or part of it. After understanding that there are about six or seven places that can not be connected to the network, because the network is disconnected at the same time, and these locations are not in the same place, there are also those on the first floor and the second floor, so I checked several computers that could not be connected, and there was nothing different on the physical connection (to put it bluntly, to see if it was connected to a variety of inexplicable switches), ping the gateway and found that ping could not be connected. At this time, it is basically locked that is the failure of the physical connection, and there is a good chance that it is the problem with the switch. But at this time, there is also doubt, if there is something wrong with the switch port, several ports will not be broken at the same time, if it is the overall failure of the switch, why only a few computers of the 24-port switch have been cut off (it is not so coincidental that more than 10 other computers have been turned off).

With doubt, he took the key and opened the door of IT's room, meditating all the time. Here is a brief description of the computer room. Because the manufacturer is responsible for the construction, the IT that handed over before is not clear about the configuration of the specific equipment and the direction of the network cable, and there is no authority to configure it, that is to say, it was like this before. A cabinet has a route and three 24 × × replacements (the brand does not say, this has nothing to do with the theme). There are about 100 outlets in B store, some of which are not connected to the switch temporarily, and 24 ports of the three switches are all connected to the network cable. It is understood that they will be turned on and used. In addition, as the computer room has been for several years, and the network cable and network structure are relatively messy, the biggest headache is that there is a pile of network cables about 30 cm thick next to the cabinet, in which there are all kinds of optical cats, small routes, small switches and other equipment (B store a total of three optical fiber inputs, one for the office network, one for the staff wifi, the other for the customer wifi). Before the IT also confirmed that there is no network port, but the line is not long enough to connect to the ground, basically wifi lines, and the company's main network has nothing to do.

Checked the operation of several 24-port switches and found that the state was normal. At this time, I was wondering if the switch was overheating (the air conditioner often tripped) or the cpu memory was too high. I decided to restart 3 switches to see if the problem could be solved. As the line is really complex, can not find the power plug, can only simply and rudely unplug the power cord and restart, the failure remains after the restart of the three switches.

After simply rebooting the switch, we must use a relatively stupid method, that is, to find out the network cable from the disconnected network to the computer room. at this time, some people may say, isn't it easy? there will be labels. However, people who have patience must be able to guess that these lines basically do not have labels, and the only labels are not necessarily correct. Of course, not every line should be checked, and the priority here is the unlit line on the switch. Connect the wire tester at a disconnected point, and go to the computer room to test the light-off network cable one by one. Although there are not many lights-out lines, there are nearly 20, which is strange that they can not be found (during this period, because the cashier has been pressing for a good job or not, the guests are in a hurry to type the invoice. I temporarily connected a long network cable, let them connect to the normal network, first solve the guest's problem! This is what I was thinking, if these lines are in these lights-out ports, it is impossible that all 8 cores will not pass unless they are completely cut off. In the case of making sure that there was no major construction and damage inside the B store, I found another network to continue to look for the line, and the result was that I couldn't find the line. At this time, what I can be sure is that the network cables of these two points are not in the three switches. I turned my attention to several 8 × × × changers in the random wire stack on the ground. Because the network cables are very chaotic, I can hardly find where the network cables of these switches lead to. It feels like connecting to the three switches in the cabinet, and some lead directly to the outside of the computer room. With a fluke in mind, I rebooted those switches all over. A few minutes later, a colleague said that he could surf the Internet. Then I woke up and checked carefully. I found that one of the switches connected to 6 ports, and one of them was connected to a 24-port switch. Brother, it's hard to find you.

Now that the problem is solved, I think it may be that the small switch is overloaded and should be able to operate normally for a period of time after restart. However, the doubt still exists. According to the report, a total of seven points cannot be connected to the network, but this switch is only connected to five ports. according to reason, if there is a problem with this switch, there will only be five points that cannot be connected to the network. and there is no possibility of physical connection of other switches outside the computer room. With this doubt, although the fault has been temporarily solved, it can not be let go.

After seeing that everyone was able to surf the Internet, I began to deal with some small problems that had not been done well. as a result, it was not long before I reported that I could not surf the Internet, and it was still the same computers just now. So go back to the computer room, carefully examine the switch that connected six lines, connect to the 24 × × × switch port, the indicator light is normal, then remove the network cable connected to the 24 × × × replacement machine, and test with the line tester to find that the eight cores are all connected. At this time, I was basically sure that it was the problem of replacing the 8xxx. since there was no extra switch for backup, I asked my colleagues in the administrative department to apply for a switch without thinking. At the moment she left the office, it suddenly occurred to me that it was not necessarily the switch.

I said, wait a minute, and I'll make sure it's the switch. So go back to the computer room, remove another 8 × × switch (switch B) next to that switch (switch A), connect the previous level networking cable, and then connect to the five faulty network cables. We should know here that if there is a problem with switch A, those computers can access the Internet in theory, on the contrary, if they are still unable to access the Internet, it has nothing to do with switch A. The moment I asked to stop the purchase and return to the computer room, I probably thought that it was not the problem with that switch, and as I expected, the problem with switch B remained the same. If I had to say that switch B also had a problem, it was obviously not the work of a responsible IT.

If you see here can accurately find the cause of the failure, it should be said that half a foot stepped into the door of IT. In fact, by now, it is not far from the truth. At that time, my eyes were fixed on the network cable connecting switch An and 24 × ×. I removed the network cable and examined it carefully. it was very hard. I couldn't find a harder network cable than that one in the computer room. I had to bend it a little bit. I re-tested the network cable with the line tester, and the 8-core was connected, so I pulled the cable and bent it again and retested it. The sixth line (green line) was found to be cut off. At this moment, I finally found the answer: due to the excellent texture of the wire (the copper core is too thick), the crystal head metal blade is not easy to press in, a little twist will cause the pressing to lose contact with the copper core, and the contact will be intermittent due to poor contact. so there will be all the above problems.

As a result, I replaced an ordinary network cable that had been connected before, replaced the cascade cable that had a problem, and then heard the reply that I could access the Internet.

I didn't breathe a sigh of relief until that moment that day. At this time, can not help but sigh, do IT operation and maintenance, in addition to a little bit of superficial technology, but also need a little bit of meticulous and luck. After talking for a long time, I finally finished the story, and I am very grateful to the readers who can see it here.

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.