Common problems and processing of WEB

First, the application of fault

HTTP 502 Failure

502 Bad Gateway

Fault Detection:

First locate the front-end failed server node and access the backend service port response time on the front-end server (Telnet). If the response time timeout is found >10s. Indicates a back-end application failure. Need to see the backend server and identify the situation.

Ps:http 502 Bad Gateway fault is generally divided into the following 2 scenarios:

Network problem: Front end cannot connect back-end service, network 100% packet drops

Back-end service problem: The backend service process is down. such as the nginx,php process hangs out.

HTTP 503

503 Service temporarily unavailable

Fault Detection:

First locate the front-end failed server node, ping the backend server on the front-end server to see the network latency drops, and the backend service port response time. If delay >100ms is found, drop packet >5%. Describes a problem with the front-to-back network.

Ps:http 503 Service Temporarily unavailable failure is typically caused by a front-end access back-end network delay. The preferred troubleshooting is whether the backend traffic overload is causing. If not, it is the front-to-back network problem.

HTTP 504

504 Gateway Time-out

Fault Detection:

Check the back-end server nginx PHP MySQL Resource usage and review the related error log.

PS: This type of failure is less likely than a small HTTP 504 Gateway time-out fault is typically due to a back-end server response timeout. If the PHP program executes too long, the database query times out. The need to increase the PHP execution timeout when communicating with the program.

Second, attack failure

DDoS attacks

Fault Detection:

View monitoring network card traffic, network latency/packet loss, number of packets/s

#iptraf-D eth0 View network traffic and packets/s, the number of traffic and packets that come in is very large, and the traffic and packets are very small. Such as:

Ps:ddos attack fault, generally comes in the traffic is relatively large, out of the traffic is small. The number of packets received by the network packet is large, the packet is small, the network delay is high, and there is a packet loss phenomenon.

CC Attack failure

Fault detection: View monitoring network card traffic, Nginx connection status, CPU load

PS:CC attack fault, generally out of the traffic is relatively large, incoming traffic is small. The number of nginx connections soared.

