netctl (简体中文)

创建了一个配置文件之后请尝试用它建立一个连接。下例中的 profile 仅使用配置文件的文件名，不要带全路径名。

# netctl start profile

如果上面的命令返回失败，可以使用 和 命令获取进一步的失败原因信息。

下列命令实现开机时自动启动配置文件：

# netctl enable profile

这条命令将创建并启用一个随计算机启动而自动运行的 systemd 服务。对配置文件本身的修改无法自动关联到前述创建的服务文件，修改之后，需用下述命令重新启用该配置：

# netctl reenable profile

启用的配置文件将在下次引导时自动启动。显然，只有线缆已连接或已处于无线信号覆盖区域时，配置文件才能成功启动。

如果需要在多个配置文件之间频繁切换（比如携带笔记本电脑旅行），应改用 #特殊 systemd 单元 一节的方法代替本节所述方法。

netctl 提供了特殊的 systemd 服务以实现有线与无线连接的自动切换。这些特殊的 systemd 单元的完整清单请参阅 。

安装 包，并且启动/启用 netctl-ifplugd@interface.service systemd 单元。网线插入/拔出时，DHCP 配置文件将被启动/停止。

• netctl-ifplugd@interface.service 优先启用使用了 DHCP 的配置文件。
• 若要自动启动一个静态 IP 配置文件，需要在其中增加 配置项。
• 若要使某个静态 IP 配置文件的优先级高于使用 DHCP 的配置文件，可以增加 配置项，这将使其优先级高于使用 DHCP 的配置文件默认的 。

安装 包并启动/启用 systemd 单元。当在不同网络覆盖区域间移动（漫游）时，netctl 配置文件将会自动启动/停止。

• netctl-auto 要求配置文件必须使用 或 Security=wpa 配置项才能工作，不能使用 配置项。

• 如果希望某些无线网络配置不要被 自动启用，需要特别在该配置文件中加入 。
• 如果存在多个无线访问点可用，可以在 WPAConfigSection 配置节中加入 配置项（参阅 /etc/netctl/examples/wireless-wpa-configsection）。

注意，服务单元名称中的 interface 一词不要原文照抄，应当替换成实际的接口设备名，如 。详情参阅 。

通过 netctl-auto 的命令动作可以在不停止 服务的情况下手工控制一个不受 netctl-auto 管理的网络接口。完整的 netctl-auto 命令动作列表参阅 。

For a DHCP connection, only the has to be configured after copying the example profile to .

For example:

/etc/netctl/''my_dhcp_profile''

Interface=enp1s0
Connection=ethernet
IP=dhcp

For a static IP configuration copy the example profile to and modify , , and DNS) as needed.

For example:

Take care to include the subnet notation of . It equates to a netmask of ) and without it the profile will fail to start. See also CIDR notation. To alias more than one IP address per a NIC set .

The following applies for the standard wireless connections using a pre-shared key (WPA-PSK).

You can also follow the following step to obfuscate the wireless passphrase (wifi-menu does it automatically when using the flag):

Users not wishing to have the passphrase to their wireless network stored in plain text have the option of storing the corresponding 256-bit pre-shared key instead, which is calculated from the passphrase and the SSID using standard algorithms.

Calculate your 256-bit PSK using wpa_passphrase:

The pre-shared key (psk) now needs to replace the plain text passphrase of the variable in the profile.

如果你想使用图形用户界面管理 netctl 和你的网络连接，并且不在意使用非官方体验版软件包的话，可以从 AUR 安装 。注意：它毕竟还只是一个 beta 版，你应该熟悉 netctl 的语法以便解决可能出现的问题。另一个图形用户界面程序的替代品是 ，它提供了基于 Qt 的图形界面、DBus 守护进程和 KDE 桌面小部件。第三个替代品是 ，它使用 作为图形界面。

参阅 Wpa supplicant#Problems with eduroam and other MSCHAPv2 connections

引自 内核文档:

The Linux bonding driver provides a method for aggregating multiple network interfaces into a single logical "bonded" interface. The behavior of the bonded interfaces depends on the mode. Generally speaking, modes provide either hot standby or load balancing services. Additionally, link integrity monitoring may be performed.

（Linux bonding 驱动提供了把多个网络接口聚合成一个“绑定”的单一逻辑接口的途径。绑定后接口的行为取决于绑定的模式，一般来说，提供“主备”和“负载均衡”两种模式。另外，可以提供对连接总体情况的监测功能。）

要用 netctl 配合 bonding，需要从官方软件源安装 ifenslave。

复制 到 然后进行编辑。例如:

/etc/netctl/bond0

Description='Bond Interface'
Interface='bond0'
Connection=bond
BindsToInterfaces=('eth0' 'eth1')
IP=dhcp
IP6=stateless

现在你可以停用之前的配置文件。然后设置 bonding 为自动启动，切换到新的配置。例如：

# netctl switch-to bond0

这一部分探讨怎样用bonding来实现当有线以太网无法工作时自动切换至无线网络。我们假设所有的网络接口默认启动dhcdpcd服务。

你需要从官方源安装软件包：ifenslave 和 wpa_supplicant.

首先设置 驱动使用 :

选项避免了  错误。如果使用了MAC过滤，应当添加设置 fail_over_mac=active。

接下来，编写一个netctl配置文件来绑定两个网络接口：

设置该配置文件自启动：

# netctl enable failover

将 wpa_supplicant 配置为关联一个已知网络，可以通过 netctl profile (记得设置 IP='no'), 和一个长期运行的 wpa_supplicant 服务或者 wpa_cli 命令实现。具体方法请访问 wpa_supplicant 页面。想要长期运行 wpa_supplicant 服务，创建一个wpa_supplicant配置文件 然后运行：

# systemctl enable wpa_supplicant@wlan0

在有线网络配置中设置。IP地址应当只被分配到bond0接口。

如果你有一个有线连接和无线连接连接到同一个网络，现在可以将有线网络断开连接，然后重新连接而依然保持网络通畅。在大多数情况下，甚至连流媒体音乐都不会卡顿。

In some cases it may be desirable to allow a profile to use any interface on the system. A common example use case is using a common disk image across many machines with differing hardware (this is especially useful if they are headless). If you use the kernel's naming scheme, and your machine has only one ethernet interface, you can probably guess that eth0 is the right interface. If you use udev's Predictable Network Interface Names, however, names will be assigned based on the specific hardware itself (e.g. enp1s0), rather than simply the order that the hardware was detected (e.g. eth0, eth1). This means that a netctl profile may work on one machine and not another, because they each have different interface names.

A quick and dirty solution is to make use of the directory. Choose a name for your interface alias ( in this example), and write the following to a file with that name (making sure it is executable).

/etc/netctl/interfaces/en-any

#!/bin/sh
for interface in /sys/class/net/en*; do
        break;
done
Interface=$(basename "$interface")
echo "en-any: using interface $Interface";

Then create a profile that uses the interface. Pay special attention to the directive. The rest are only provided as examples.

When the profile is started, any machine using the two files above will automatically bring up and configure the first ethernet interface found on the system, regardless of what name udev assigned to it. Note that this is not the most robust way to go about configuring interfaces. If you use multiple interfaces, netctl may try to assign the same interface to them, and will likely cause a disruption in connectivity. If you do not mind a more complicated solution, is likely to be more reliable.

netctl supports hooks in /etc/netctl/hooks/ and per interface hooks in . You can set any option in a hook/interface that you can in a profile. They are read the same way! Most importantly this includes and .

When a profile is read, netctl sources all executable scripts in , then it reads the profile file for the connection and finally it sources an executable script with the name of the interface used in the profile from the directory. Therefore, declarations in an interface script override declarations in the profile, which override declarations in hooks.

The variables , , and $Profile are available in hooks/interfaces only when using

Using this, systemd services requiring an active network connection can be ordered to start only after the is reached, and can be stopped before the connection is brought down.

To set or change the DHCP client used for all profiles:

/etc/netctl/hooks/dhcp

#!/bin/sh
DHCPClient='dhclient'

Alternatively, it may also be specified for a specific network interface by creating an executable file with the following line:

DHCPClient='dhclient'

Some people have an issue when they connect to a network with netctl, for example:

When then looking at , either of the following are shown:

1. If your device ( in this case) is up:

network[2322]: The interface of network profile 'wlan0-ssid' is already up

Setting the interface down should resolve the problem:

# ip link set wlan0 down

Then retry:

# netctl start wlan0-ssid

2. If it is down:

dhcpcd[261]: wlan0: ipv4_sendrawpacket: Network is down

One way to solve this is to use a different DHCP client, for example . After installing the package configure netctl to use it:

Adding the ForceConnect option may also be helpful:

Save it and try to connect with the profile:

# netctl start wlan0-ssid

On some systems dhcpcd in combination with netctl causes timeout issues on resume, particularly when having switched networks in the meantime. netctl will report that you are successfully connected but you still receive timeout issues. In this case, the old default route still exists and is not being renewed. A workaround to avoid this misbehaviour is to switch to dhclient as the default dhcp client. More information on the issue can be found here.

If you are having timeout issues when requesting leases via DHCP you can set the timeout value higher than netctl's 30 seconds by default. Create a file in /etc/netctl/hooks/ or , add to it for a timeout of 40 seconds and make the file executable.

If you are having timeout issues that are unrelated to DHCP (on a static ethernet connection for example), and are experiencing errors similar to the following when starting your profile: Then you should increase carrier and up timeouts by adding and to your profile file: Do not forget to reenable your profile:

# netctl reenable profile

Sometimes netctl-auto fails to reconnect when the system resumes from suspend. An easy solution is to restart the service for netctl-auto. This can be automated with an additional service like the following:

To enable this service for your wireless card, for example, enable netctl-auto-resume@wlan0.service as root. Change to the required network interface.

If the device is not yet running on resume when the unit is started, this will fail. It can be fixed by adding the following dependency in the After line:

This problem seems to be related to a recent wpa_supplicant update (see FS#44731), but a work-around is quite trivial. Just create a file for your interface (e.g. wlp3s0) in /etc/netctl/interfaces with the following content and make it executable:

After that, try to restart your netctl-auto service and WiFi auto detection should work well again.

Many laptops have a hardware button (or switch) to turn off wireless card, however, the card can also be blocked by the kernel. This can be handled by rfkill.

If you want netctl-auto to automatically unblock your wireless card to connect to a particular network, set option for the wireless connection of your choice, as specified in the netctl.profile(5) man page.

This is a very misleading response, it really means that you have assigned a default gateway in an earlier netctl control file. When netctl starts up the n-th NIC and goes to set its local route, it fails because there is already a default route from n-1.

Remove it and everything works, except you no longer have a default route and so cannot access things such as the internet. does not work as it gets executed for each network card.

A possible solution is creating a new service: