1. RPM配置

解决场景：
Juniper多链路出口，路由主备，链路故障后，自动切换，同时不影响多链路的Dnat
关键词解释：
Successive-Loss 连续丢包数
Total-Loss 总丢包数

联通链路rpm
set services rpm probe RpmProbeCU test cu-gateway target address 114.114.114.114 \\探测的目的地址
set services rpm probe RpmProbeCU test cu-gateway probe-count 4 \\发送4个探测包
set services rpm probe RpmProbeCU test cu-gateway probe-interval 5	\\4个探测包的间隔为5秒钟
set services rpm probe RpmProbeCU test cu-gateway test-interval 5	\\完成4个探测包后，距离下次探测的间隔时间
set services rpm probe RpmProbeCU test cu-gateway source-address 10.1.1.2 \\已10.1.1.2为源进行探测
set services rpm probe RpmProbeCU test cu-gateway thresholds successive-loss 3 \\连续丢包数为3个
set services rpm probe RpmProbeCU test cu-gateway thresholds total-loss 3 \\总丢包数为3个
电信链路rpm
set services rpm probe RpmProbeCT test ct-gateway target address 114.114.114.114
set services rpm probe RpmProbeCT test ct-gateway probe-count 4
set services rpm probe RpmProbeCT test ct-gateway probe-interval 5
set services rpm probe RpmProbeCT test ct-gateway test-interval 5
set services rpm probe RpmProbeCT test ct-gateway source-address 20.1.1.2
set services rpm probe RpmProbeCT test ct-gateway thresholds successive-loss 3
set services rpm probe RpmProbeCT test ct-gateway thresholds total-loss 3

rpm监测切换策略
链路故障后，被选中的路由优先级会被系统重置为1，切换时间大概20秒；当链路故障恢复后，会重新切回原来的路由
set services ip-monitoring policy RpmPolicyCU match rpm-probe RpmProbeCU \\如果满足rpm probe监测条件，则执行以下路由，
set services ip-monitoring policy RpmPolicyCU then preferred-route route 0.0.0.0/0 next-hop 20.1.1.1
set services ip-monitoring policy RpmPolicyCT match rpm-probe RpmProbeCT
set services ip-monitoring policy RpmPolicyCT then preferred-route route 0.0.0.0/0 next-hop 10.1.1.1