Things to keep in mind when a UNIX system restarts

Things to keep in mind when a UNIX system restarts

It is easy for system administrators to manage their own servers, but how to manage them is not a simple matter. It has become a normal life for Windows Server administrators to regularly restart Windows devices, but this is not the norm in Unix systems, and restarting by default does not bring any form of improvement.

This is a hot topic for every server administrator, but in the eyes of Unix geeks It is a deeper issue-perhaps because Windows administrators tend to reboot as one of the first steps in troubleshooting, and the UNIX team generally tries only when it is helpless.

Two Scenarios for UNIX server restart

The reality is that the server restart operation should rarely occur-please note that there are very few. Here I enumerate kernel updates and hardware replacements as examples, as they are the two main reasons for restarting in the UNIX domain. There is no basis for some people to think that there are some serious security risks that can occur without restarting the server. If there is a security risk in the service project and the application, then a bug fix will solve the problem, and the patch often does not require a restart of the device. And if the security risk exists in the kernel module, in general, simply uninstall the corresponding module, install patches, and finally reload the module. Yes, the restart operation is really necessary once there is a security risk in the kernel. But beyond that, there is no more precise reason to restart the UNIX server at all.

Some people think that if you do not restart, other forms of risk tend to follow, such as some critical service items are not properly enabled at boot time, which leads to a series of pitfalls. Of course, this is true in itself, but as long as the management work is in place, it will not happen at all. Only novices who have just picked up the server device will forget to set the startup parameters of the service item correctly. But then again, if everyone's server is in the construction phase, and there is no production-related content, then you may want to do all kinds of restart testing, which will not have any adverse effects. And I think this is the best time to be familiar with the restart mechanism.

But there's another thing to consider: the guy who takes the restart action as one of the important steps in troubleshooting is a lose mentality and intends to expose the problem at once. Just say a set of UNIX devices that already have problems, and some of the services that are still in service are actually not able to start again, and this is revealed after a reboot-perhaps due to fragmentation or other bizarre reasons.

Causes of UNIX server restart

If we simply look at a few minutes after a pat on the head to restart the device, then perhaps the real cause of the failure is completely lost in time-perhaps a junior administrator in running a set of their own stupid script, inadvertently deleted the//////////// Part of the lib64 directory. This is the culprit that caused fragmentation and device instability. However, once we choose to restart the server directly without digging into the problem, it is clear that the problem will become even more serious, and then you should start the recovery image-which means that there is a lot of recovery work to be done-and that the production server will be in a state of downtime at the same time.

This is just one of the reasons we should try to avoid restarting operations in the UNIX world. Rather than a troubleshooting approach, think of it as a kind of desperate gamble-either finding the problem or destroying it yourself and rebuilding it slowly. In short, no one can use the/var partition to restart the device to completely correct the error.

　Please do your job before the server restarts.

In most cases, it is extremely important not to restart, because the critical content of the system that helps us fix the problem must exist before the reboot, but it may not be there after the reboot. The problem will definitely reappear after the reboot, but once the solution has dissipated with the restart behavior, the fault itself is stuck in a dead loop without a solution. Unless someone decides not to restart, try to find the root of the problem. Unfortunately, there are few people who can make this wise choice. The actual situation is: A small fault memory bar can give the system normal operation and the device start state brings great trouble. This time, the remedy is the best policy, blindly restart will only bring additional losses.

Small part said a few words: Later in the face of the problem, if there is a guy said what is better than first reboot to see, may as well direct to him two punches. Rebooting is certainly one of the scenarios, but it is important to ensure that we have taken all the measures we can think of before implementing the restart, after all we have saved our own time and energy!

Share from bbs.dbfen.com reprint please specify the source

This article is from the "Big Meatball" blog, please make sure to keep this source http://12478147.blog.51cto.com/9663367/1600336

Things to keep in mind when a UNIX system restarts