Summary of failure when a website cannot be opened on the Internet

Let's talk about the company's website architecture. The company's front-end uses Keepalived + LVS for load balancing, followed by WEB servers. The WEB and IMG are independent, and the IMG uses NFS + storage servers, the IMG front-end uses Squid for caching. The entire website uses the Alibaba Cloud security firewall, and the entire website architecture is not complex.

At on July 22, September 22, I received a fault message and reported that the entire site was unavailable. At this time, I can't open my website with my mobile phone. I rushed to the company to handle the problem. below is my summary of the solution and ideas.

Step 1: First, we use curl/lsof and other tools to test whether the backend WEB services are normal and eliminate WEB faults. It is also OK for the company to open the website in a browser.

Step 2: Do you think of DNS resolution problems? The company's internal access points), so that my colleagues did a test elsewhere. The DNS resolution was normal and the ping domain name failed, first, the DNS issue is ruled out. Then let him traceroute to check whether the line is smooth. The VIP cannot be found at the upper-level route of our cabinet.

650) this. width = 650; "src =" http://www.bkjia.com/uploads/allimg/131228/00235420H-0.jpg "title =" 00.jpg" alt = "160531184.jpg" style =" float: left; "/>

Step 3: ping the domain name fails. Is it a problem with lvs vip? Then, use tail-f/var/log/messages in Load Balancer to view the log. No exception information is found, viewing the VIP with ip addr is also bound to the eth0: 2 Nic, and then accessing the website with curl tool in Load Balancer is also OK, indicating that LVS is normal.

650) this. width = 650; "src =" http://www.bkjia.com/uploads/allimg/131228/00235424D-1.jpg "title =" 01.jpg" alt = "100154577.jpg"/>

Step 4: DNS, WEB, and LVS are normal. It must have been an issue with the Alibaba Cloud security firewall in the data center. Then, find the firewall logs and find that the firewall has restarted, it may be caused by a policy problem after the restart that prevents access to our website.

650) this. width = 650; "src =" http://www.bkjia.com/uploads/allimg/131228/00235449D-2.jpg "title =" 1.jpg" alt = "160657928.jpg"/>

Step 5: In order to save time, contact Alibaba Cloud security's technical support to help us see if there is a problem with the firewall, and find that there is a problem with the rule that shields all the ip addresses of the client, the rule setting date is not found. It may be because the previously set policy takes effect after restart.) delete all the OK rules.

650) this. width = 650; "src =" http://www.bkjia.com/uploads/allimg/131228/002354L53-3.jpg "title =" 12.jpg" alt = "160723810.jpg"/>

Conclusion: when a problem occurs on the entire website, first report the problem to the website leader. It is best to determine the problem in the first place and check whether the problem can be solved in time, if you cannot solve the problem, first release a temporary solution so that the entire website can be accessed by users.

This article from the "simple dream catcher" blog, please be sure to keep this source http://sfzhang88.blog.51cto.com/4995876/1301121