跳转至

清理失效IPsecSA

清理失效sa

一、ROS-A 新增 SA 专用 Netwatch

ROS-A 当前参数:

WAN 源地址:10.240.0.48
IPsec peer:peer-ct
GRE 对端:198.19.13.1
GRE 源地址:198.19.13.2

执行:

/tool netwatch
add name="SA-A-WAN1" host=223.5.5.5 src-address=10.240.0.48 type=icmp interval=10s timeout=2s packet-count=5 packet-interval=200ms thr-loss-percent=50%

add name="SA-A-WAN2" host=1.2.4.8 src-address=10.240.0.48 type=icmp interval=10s timeout=2s packet-count=5 packet-interval=200ms thr-loss-percent=50%

add name="SA-A-WAN3" host=119.29.29.29 src-address=10.240.0.48 type=icmp interval=10s timeout=2s packet-count=5 packet-interval=200ms thr-loss-percent=50%

add name="SA-A-GRE" host=198.19.13.1 src-address=198.19.13.2 type=icmp interval=10s timeout=2s packet-count=5 packet-interval=200ms thr-loss-percent=50%

检查新增条目:

/tool netwatch print detail where name~"SA-A-"

二、ROS-A SA watchdog 脚本

如果确认还没有 sa-watchdog-a,执行:

/system script
add name="sa-watchdog-a" policy=read,write,policy,test,sensitive source={
    :global saAWdBad
    :global saAWdCooldown

    :if ([:typeof $saAWdBad] = "nothing") do={
        :set saAWdBad 0
    }

    :if ([:typeof $saAWdCooldown] = "nothing") do={
        :set saAWdCooldown 0
    }

    :if ($saAWdCooldown > 0) do={
        :set saAWdCooldown ($saAWdCooldown - 1)
    }

    :local wan1 [/tool netwatch get [find where name="SA-A-WAN1"] status]
    :local wan2 [/tool netwatch get [find where name="SA-A-WAN2"] status]
    :local wan3 [/tool netwatch get [find where name="SA-A-WAN3"] status]

    :local internetOk false

    :if (($wan1 = "up") || ($wan2 = "up") || ($wan3 = "up")) do={
        :set internetOk true
    }

    # All independent WAN probes are down.
    # Treat it as an Internet failure and do not reset IPsec.
    :if ($internetOk = false) do={
        :set saAWdBad 0
        :return
    }

    :local greStatus [/tool netwatch get [find where name="SA-A-GRE"] status]

    :local activeCount [:len [/ip ipsec active-peers find]]
    :local establishedCount [:len [/ip ipsec active-peers find where state="established"]]
    :local matureCount [:len [/ip ipsec installed-sa find where state="mature"]]

    :local saHealthy false

    :if (($establishedCount > 0) && ($matureCount >= 2)) do={
        :set saHealthy true
    }

    # IPsec SA is healthy but GRE endpoint is down.
    # This is more likely a remote GRE problem, so do not reset SA.
    :if (($greStatus != "up") && ($saHealthy = true)) do={
        :set saAWdBad 0
        :return
    }

    :local tunnelHealthy false

    :if (($greStatus = "up") && ($saHealthy = true)) do={
        :set tunnelHealthy true
    }

    :if ($tunnelHealthy = true) do={
        :set saAWdBad 0
        :return
    }

    :set saAWdBad ($saAWdBad + 1)

    :log warning ("SA-A abnormal: count=" . $saAWdBad . " gre=" . $greStatus . " established=" . $establishedCount . " mature-sa=" . $matureCount)

    # Three consecutive abnormal checks.
    # Scheduler interval is 30 seconds, so approximately 90 seconds.
    :if (($saAWdBad >= 3) && ($saAWdCooldown = 0)) do={
        :local peerCfg [/ip ipsec peer find where name="peer-ct"]
        :local policyCfg [/ip ipsec policy find where peer="peer-ct" and protocol=gre]

        :if ([:len $peerCfg] = 0) do={
            :log error "SA-A peer-ct configuration not found"
            :set saAWdBad 0
            :return
        }

        :if ([:len $policyCfg] = 0) do={
            :log error "SA-A GRE IPsec policy not found"
            :set saAWdBad 0
            :return
        }

        :log error "SA-A resetting peer-ct"

        /ip ipsec policy set $policyCfg disabled=yes
        /ip ipsec peer set $peerCfg disabled=yes

        :delay 3s

        # Current ROS-A has only one active IPsec peer.
        /ip ipsec active-peers kill-connections

        :delay 2s

        /ip ipsec peer set $peerCfg disabled=no
        /ip ipsec policy set $policyCfg disabled=no

        :set saAWdBad 0
        :set saAWdCooldown 10

        :log warning "SA-A peer-ct reset completed"
    }
}

添加调度器:

/system scheduler
add name="sa-watchdog-a-scheduler" start-time=startup interval=30s on-event="/system script run sa-watchdog-a" policy=read,write,policy,test,sensitive

三、ROS-B 新增 SA 专用 Netwatch

ROS-B 当前参数:

WAN 源地址:10.240.0.46
IPsec peer:peer-cm
GRE 对端:198.19.13.9
GRE 源地址:198.19.13.10

执行:

/tool netwatch
add name="SA-B-WAN1" host=223.5.5.5 src-address=10.240.0.46 type=icmp interval=10s timeout=2s packet-count=5 packet-interval=200ms thr-loss-percent=50%

add name="SA-B-WAN2" host=1.2.4.8 src-address=10.240.0.46 type=icmp interval=10s timeout=2s packet-count=5 packet-interval=200ms thr-loss-percent=50%

add name="SA-B-WAN3" host=119.29.29.29 src-address=10.240.0.46 type=icmp interval=10s timeout=2s packet-count=5 packet-interval=200ms thr-loss-percent=50%

add name="SA-B-GRE" host=198.19.13.9 src-address=198.19.13.10 type=icmp interval=10s timeout=2s packet-count=5 packet-interval=200ms thr-loss-percent=50%

检查新增条目:

/tool netwatch print detail where name~"SA-B-"

四、ROS-B SA watchdog 脚本

/system script
add name="sa-watchdog-b" policy=read,write,policy,test,sensitive source={
    :global saBWdBad
    :global saBWdCooldown

    :if ([:typeof $saBWdBad] = "nothing") do={
        :set saBWdBad 0
    }

    :if ([:typeof $saBWdCooldown] = "nothing") do={
        :set saBWdCooldown 0
    }

    :if ($saBWdCooldown > 0) do={
        :set saBWdCooldown ($saBWdCooldown - 1)
    }

    :local wan1 [/tool netwatch get [find where name="SA-B-WAN1"] status]
    :local wan2 [/tool netwatch get [find where name="SA-B-WAN2"] status]
    :local wan3 [/tool netwatch get [find where name="SA-B-WAN3"] status]

    :local internetOk false

    :if (($wan1 = "up") || ($wan2 = "up") || ($wan3 = "up")) do={
        :set internetOk true
    }

    :if ($internetOk = false) do={
        :set saBWdBad 0
        :return
    }

    :local greStatus [/tool netwatch get [find where name="SA-B-GRE"] status]

    :local activeCount [:len [/ip ipsec active-peers find]]
    :local establishedCount [:len [/ip ipsec active-peers find where state="established"]]
    :local matureCount [:len [/ip ipsec installed-sa find where state="mature"]]

    :local saHealthy false

    :if (($establishedCount > 0) && ($matureCount >= 2)) do={
        :set saHealthy true
    }

    :if (($greStatus != "up") && ($saHealthy = true)) do={
        :set saBWdBad 0
        :return
    }

    :local tunnelHealthy false

    :if (($greStatus = "up") && ($saHealthy = true)) do={
        :set tunnelHealthy true
    }

    :if ($tunnelHealthy = true) do={
        :set saBWdBad 0
        :return
    }

    :set saBWdBad ($saBWdBad + 1)

    :log warning ("SA-B abnormal: count=" . $saBWdBad . " gre=" . $greStatus . " established=" . $establishedCount . " mature-sa=" . $matureCount)

    :if (($saBWdBad >= 3) && ($saBWdCooldown = 0)) do={
        :local peerCfg [/ip ipsec peer find where name="peer-cm"]
        :local policyCfg [/ip ipsec policy find where peer="peer-cm" and protocol=gre]

        :if ([:len $peerCfg] = 0) do={
            :log error "SA-B peer-cm configuration not found"
            :set saBWdBad 0
            :return
        }

        :if ([:len $policyCfg] = 0) do={
            :log error "SA-B GRE IPsec policy not found"
            :set saBWdBad 0
            :return
        }

        :log error "SA-B resetting peer-cm"

        /ip ipsec policy set $policyCfg disabled=yes
        /ip ipsec peer set $peerCfg disabled=yes

        :delay 3s

        # Current ROS-B has only one active IPsec peer.
        /ip ipsec active-peers kill-connections

        :delay 2s

        /ip ipsec peer set $peerCfg disabled=no
        /ip ipsec policy set $policyCfg disabled=no

        :set saBWdBad 0
        :set saBWdCooldown 10

        :log warning "SA-B peer-cm reset completed"
    }
}

添加调度器:

/system scheduler
add name="sa-watchdog-b-scheduler" start-time=startup interval=30s on-event="/system script run sa-watchdog-b" policy=read,write,policy,test,sensitive

五、判断逻辑

SA-A/B-WAN1、WAN2、WAN3 全部 down:
    认为外网整体故障
    不清理 SA

至少一个 WAN up,GRE up,active peer established,mature SA >= 2:
    隧道正常
    不处理

GRE down,但 peer established 且 mature SA >= 2:
    更可能是远端 GRE 问题
    不清理 SA

GRE down,且 peer 不 established 或 mature SA < 2:
    连续 3 次异常后重建

时间参数:

Netwatch:10 秒
watchdog:30 秒
连续异常:3 次,约 90 秒
重置后冷却:10 次,约 5 分钟

六、验证命令

先只检查新增 Netwatch:

/tool netwatch print detail where name~"SA-"

手动执行 watchdog。当前隧道正常时不会重置:

A:

/system script run sa-watchdog-a

B:

/system script run sa-watchdog-b

查看日志:

/log print where message~"SA-"

查看调度器:

/system scheduler print detail where name~"sa-watchdog"

查看实际 SA:

/ip ipsec active-peers print detail
/ip ipsec installed-sa print detail
/interface gre print detail

正常状态应为:

SA-* WAN probes:至少一个 up
SA-* GRE:up
active peer:established
installed-sa:至少两条 mature
GRE:RUNNING