管理 IP 地址

在 Redis Enterprise Software 中使用多个 IP 地址或 IPv6 地址的信息和要求。

Redis 企业软件

Redis Enterprise Software 支持服务器、实例和 VM，其中 多个 IPv4 或 IPv6 地址。

流量概览

Redis Enterprise Software 流量分为内部流量和外部流量：

  • “内部流量” 是指内部集群通信，例如用于集群管理的节点之间的通信。

  • “外部流量”是指客户端和数据库之间的通信以及与 Cluster Manager UI 的连接。

当用作 Redis Enterprise 节点的计算机上仅存在一个 IP 地址时，该地址将同时用于内部和外部流量。

多个 IP 地址

在具有多个 IP 地址的计算机上配置节点期间，必须为内部流量分配一个地址，为外部流量分配一个或多个其他地址。

如果集群将 IPv4 用于内部流量，则集群节点之间的所有通信都使用 IPv4 地址。如果集群将 IPv6 用于内部流量，则集群节点之间的所有通信都使用 IPv6 地址。

要在集群设置后更新 IP 地址配置，请参阅更改内部 IP 地址或配置外部 IP 地址。

为内部流量启用 IPv6

只有使用 Redis Enterprise Software 版本 7.4.2 或更高版本的新集群才支持用于内部通信的 IPv6。

如果服务器只有 IPv6 接口，则 IPv6 会自动用于内部和外部流量。否则，内部流量默认使用 IPv4。

要在同时具有 IPv4 和 IPv6 接口的机器上将 IPv6 用于内部流量，请将use_internal_ipv6自true当您使用引导 REST API 请求创建集群时：

POST /v1/bootstrap/create_cluster
{
    "action": "create_cluster",
    "cluster": { 
        "name": "cluster.fqdn" 
    },
    "credentials": {
       "username": "admin_username",
       "password": "admin_password"
    },
    "node": {
       "identity": {
          "addr": "2001:DB8::/32",
          "external_addr": ["2001:0db8:85a3:0000:0000:8a2e:0370:7334"],
          "use_internal_ipv6": true
       },
    },
    ...
}

When other IPv6 nodes join a cluster that has use_internal_ipv6 enabled, they automatically use IPv6 for internal traffic. Do not manually set use_internal_ipv6 when joining a node to an existing IPv6 cluster, or a NodeBootstrapError can occur if the values do not match.

If you try to add a node without an IPv6 interface to a cluster that has use_internal_ipv6 enabled, a NodeBootstrapError occurs.

The host file /etc/hosts on each node in the cluster must include the following entry:

::1 localhost

Change internal IP address

Before you change an internal IP address, consider the following:

  • Verify the address is valid and bound to an active interface on the node. Failure to do so prevents the node from coming back online and rejoining the cluster.

  • Joining a node that only has IPv4 network interfaces to a master node with IPv6 enabled causes a NodeBootstrapError.

  • Joining a node that only has IPv6 network interfaces to a master node that does not have IPv6 enabled causes a NodeBootstrapError.

  • You cannot change the internal address from IPv4 to IPv6 or IPv6 to IPv4 in a running cluster. You can only change the internal address within the same protocol as the cluster.

If you need to update the internal IP address in the OS, one option is to remove that node from the cluster, change the IP address, and then add the node back into the cluster.

Alternatively, you can use the following steps to update a node's internal IP address without removing it from the cluster:

  1. Turn the node into a replica using rladmin:

    rladmin node <ID> enslave demote_node

  2. Deactivate the rlec_supervisor service on the node:

    systemctl disable rlec_supervisor

  3. Restart the node.

  4. Follow the operating system vendor's instructions to change the node's IP address.

  5. From a different cluster node, use rladmin node addr set to update the first node's IP address:

    rladmin node <ID> addr set <IP address>

  6. Enable the rlec_supervisor service on the node:

    systemctl enable rlec_supervisor

  7. Restart rlec_supervisor or restart the node.

    systemctl start rlec_supervisor

  8. Verify the node rejoined the cluster:

    rladmin status nodes

Repeat this procedure for other cluster nodes to change their internal IP addresses.

Configure external IP addresses

You can configure external addresses that are not bound to an active interface, but are otherwise mapped or configured to route traffic to the node (such as AWS Elastic IPs or a load balancer VIP).

You can use rladmin node external_addr to change a node's external IP addresses.

Add an external IP address:

rladmin node <ID> external_addr add <IP address>

Set one or more external IP addresses:

rladmin node <ID> external_addr set <IP address 1> <IP address N>

Remove an external IP address:

rladmin node <ID> external_addr remove <IP address>
Note:
While joining a new node to a cluster during the node bootstrap process, when prompted to provide an IP of an existing node in the cluster, if you use the node's IP, provide the node's internal IP address.

Known limitations

  • Using IPv6 for internal traffic is supported only for new clusters running Redis Enterprise Software version 7.4.2 or later.

  • Changing an existing cluster's internal traffic from IPv4 to IPv6 is not supported.

  • All nodes must use the same protocol for internal traffic.

  • If a Redis Enterprise node's host machine has both IPv4 and IPv6 addresses, internal communication within the node initially uses IPv4 until the bootstrap process finishes.

RATE THIS PAGE
Back to top ↑