高数量的aws目标群体健康检查

Question

我有一个包含几个注册目标组的应用程序负载均衡器(以及6个可用性区域，请注意)。

有一个ec2实例，它是所有目标组的注册目标。在ec2实例中，有一个nginx正在运行。

对于每个目标组，我使用自定义url定义了健康检查，间隔为60秒。

当我查看nginx日志时，我希望每60秒查看一次特定目标组的健康检查url。但令我惊讶的是，在60秒内，有8个呼叫组出现了这样的情况：

172.31.25.32 - - [14/Feb/2022:16:00:29 +0000] "GET /path/target-group-X/ HTTP/1.1" 200 4 "-" "ELB-HealthChecker/2.0" rt=0.118 uct="0.000" uht="0.120" urt="0.120"
172.31.89.13 - - [14/Feb/2022:16:00:35 +0000] "GET /path/target-group-X/ HTTP/1.1" 200 4 "-" "ELB-HealthChecker/2.0" rt=0.080 uct="0.000" uht="0.080" urt="0.080"
172.31.75.210 - - [14/Feb/2022:16:00:43 +0000] "GET /path/target-group-X/ HTTP/1.1" 200 4 "-" "ELB-HealthChecker/2.0" rt=0.050 uct="0.000" uht="0.052" urt="0.052"
172.31.88.219 - - [14/Feb/2022:16:00:44 +0000] "GET /path/target-group-X/ HTTP/1.1" 200 4 "-" "ELB-HealthChecker/2.0" rt=0.059 uct="0.000" uht="0.060" urt="0.060"
172.31.9.236 - - [14/Feb/2022:16:00:51 +0000] "GET /path/target-group-X/ HTTP/1.1" 200 4 "-" "ELB-HealthChecker/2.0" rt=0.059 uct="0.000" uht="0.060" urt="0.060"
172.31.15.138 - - [14/Feb/2022:16:01:02 +0000] "GET /path/target-group-X/ HTTP/1.1" 200 4 "-" "ELB-HealthChecker/2.0" rt=0.010 uct="0.000" uht="0.008" urt="0.008"
172.31.49.23 - - [14/Feb/2022:16:01:07 +0000] "GET /path/target-group-X/ HTTP/1.1" 200 4 "-" "ELB-HealthChecker/2.0" rt=0.062 uct="0.000" uht="0.064" urt="0.064"
172.31.47.189 - - [14/Feb/2022:16:01:13 +0000] "GET /path/target-group-X/ HTTP/1.1" 200 4 "-" "ELB-HealthChecker/2.0" rt=0.094 uct="0.000" uht="0.092" urt="0.092"


172.31.25.32 - - [14/Feb/2022:16:01:29 +0000] "GET /path/target-group-X/ HTTP/1.1" 200 4 "-" "ELB-HealthChecker/2.0" rt=0.050 uct="0.000" uht="0.048" urt="0.048"
172.31.89.13 - - [14/Feb/2022:16:01:35 +0000] "GET /path/target-group-X/ HTTP/1.1" 200 4 "-" "ELB-HealthChecker/2.0" rt=0.049 uct="0.000" uht="0.048" urt="0.048"
172.31.75.210 - - [14/Feb/2022:16:01:43 +0000] "GET /path/target-group-X/ HTTP/1.1" 200 4 "-" "ELB-HealthChecker/2.0" rt=0.280 uct="0.000" uht="0.280" urt="0.280"
172.31.88.219 - - [14/Feb/2022:16:01:44 +0000] "GET /path/target-group-X/ HTTP/1.1" 200 4 "-" "ELB-HealthChecker/2.0" rt=0.050 uct="0.000" uht="0.048" urt="0.048"
172.31.9.236 - - [14/Feb/2022:16:01:52 +0000] "GET /path/target-group-X/ HTTP/1.1" 200 4 "-" "ELB-HealthChecker/2.0" rt=0.508 uct="0.000" uht="0.508" urt="0.508"
172.31.15.138 - - [14/Feb/2022:16:02:02 +0000] "GET /path/target-group-X/ HTTP/1.1" 200 4 "-" "ELB-HealthChecker/2.0" rt=0.176 uct="0.000" uht="0.172" urt="0.172"
172.31.49.23 - - [14/Feb/2022:16:02:07 +0000] "GET /path/target-group-X/ HTTP/1.1" 200 4 "-" "ELB-HealthChecker/2.0" rt=0.061 uct="0.000" uht="0.060" urt="0.060"
172.31.47.189 - - [14/Feb/2022:16:02:13 +0000] "GET /path/target-group-X/ HTTP/1.1" 200 4 "-" "ELB-HealthChecker/2.0" rt=0.057 uct="0.000" uht="0.056" urt="0.056"

有8种不同的本地IP-来自于呼叫。如果我单独使用每个IP (例如172.31.25.32)，那么来自该IP的健康检查电话确实是在60秒后到达的。但其他电话是怎么回事？为什么这么多？

Answer 1

我认为目标组至少要从每个可用性区域，或者可能是每个VPC子网进行健康检查。您可能可以将这些in映射回VPC中的特定子网。

这显然看起来过度了，但你必须意识到，在幕后，多个负载均衡器实际上是多个服务器，每个服务器都在对目标服务器进行自己的健康检查。