It is not easy to put such things as server hardware maintenance. It is not easy because most of our energy is concentrated on systems and applications. Hardware maintenance can be described in words. It doesn't seem to be much. However, based on my experience, a good maintenance strategy can make the server prolong life. Let's start with two common accidents and then summarize some common maintenance measures.
a accident: a research institute of the center room on the top floor of a building, the building during the holiday 5.1 engaged in the decoration, the action is very large. At that time, I was on vacation in Chongqing, and I was recalled urgently because of the failure. When I saw it in the computer room, it was all day, and the water in the computer room was all water! It was originally renovated on the roof and the roof was opened up, so that a lot of tap water and rain leaked into the machine room. The server's outer casing and motherboard are all water, not bad.
accident two: a company placed inside the engine room servers, switches and other equipment dozens of sets, large measuring fever, damage due to air conditioning, the heat has become a big problem, because there are many times the room temperature is too high and cause the server to switch The experience of the strike. Just the machine room is by the window, opening the window to heat is the only choice (waiting for the property to repair the air conditioning is very time consuming). Unfortunately, finally there was a heavy rain. The four servers, two switches and one router on the window of the window were all wet. Three people quickly dehumidified the equipment with paper towels, rags and hair dryers. There are 2 servers destroyed (haha, the switch quality is awesome!).
When I first walked into the room IDC, was surprised why so many machines hosted here. After the above two accidents, I realized how safe it is to put the server in a safe place. To be straightforward, if you want to rely on the server to make money, such an important business, it is strongly recommended to host it in the IDC room. After finding a good place for the server, the most worthy of attention is the hard disk. Although the scsi hard disk supports hot swapping, it is not allowed to do this during the running of the server. Wait until the system stops the hard disk and then plug it in and out. The hard disk should be kept in mind. The power supply is also a troublesome place. If the server is equipped with dual power supplies, it is best to use both power supplies. Once one power supply is damaged, another power supply continues to work.
back and then talk about the heat. People always like to stack a lot of servers in a rack to save valuable space. Several machines are stacked directly together (I usually work three 3U machines directly stacked), so that they can dissipate heat. Certainly not good, if it is placed in the cabinet, it is better to put a device in a partition, so that there will be a gap between the devices, the heat dissipation effect will be much better. Summer is coming, we must install air conditioners in the computer room to make the server have a good heat dissipation effect.
server running over many years, the chassis will absorb a lot of dust and other debris, the debris will first affect the heat, and sometimes causes the machine does not start up, such as memory stick Goldfinger dust will cause the machine can not get up . In general, dust is cleaned approximately every six months. Clean the server with a brush, vacuum cleaner, screwdriver, etc., open the cover, remove the cpu, memory, etc. from the motherboard, then use the vacuum cleaner to clean the motherboard and slot; then check the cpu fan, fiddle Whether it rotates smoothly or not, if it is not smooth, it should be soaked and washed with anhydrous alcohol. If the shaft is loose after cleaning, it should be replaced with a new fan; the rack server, fan and cpu are not in one piece, it is a separate module, it is easy to insert and pull. Also need to clean up. The most dusty thing is the cpu heat sink. Take some time to brush the dust in the gap, then use a vacuum cleaner to clean the dust as much as possible. When it comes to warranty issues, it is best not to disassemble the power supply and use a vacuum cleaner to suck it. Remove the memory card, network card, etc., and wipe it off with the antistatic cloth.
inspection is a routine task, by looking at the status of the indicator light on the front panel of the server to understand the operation of the server; touch the server shell to see if it is overheated; listen to the sound of the server with the ear, determine the fan, hard disk and other machinery Whether the part is normal. A better IDC room will put a thermometer in the cabinet, which is a good idea. These are just finishing
from a hardware point of a number of measures to maintain, some tools can remotely monitor the current cpu temperature, fan speed and other indicators, but also to set thresholds based on these values, once an index exceeds the threshold The monitoring platform will send emails and SMS messages to automatically alert the system administrators to deal with them in a timely manner. Due to space limitations, it will not be detailed.