DHCP Failover Configuration
DHCP Failover performs seamless failover, real-time database synchronization, and load balancing that ensures high availability of the DHCP service. DHCP Failover is available for IP v4 addressing. Besides providing more robust DHCP services, DHCP Failover avoids duplicate addressing when two EdgeConnect gateways at the same site enable their DHCP servers.
DHCP Failover server peers are configured on two EdgeConnets, one on each. Configure consistent subnet IP pool ranges, peer IP addresses, port numbers, and other failover settings across the peer pairs for deterministic behavior of the DHCP service. Settings are not compared and validated between the failover peers. DHCP Failover settings examples are provided below.
DHCP Failover can be configured in a Deployment Profile or directly on a LAN interface of an EdgeConnect. The procedure below provides the steps for configuring DHCP Failover directly on EdgeConnect LAN interfaces.
Preparation
Before starting this process, record the IP addresses for the EdgeConnect LAN interfaces that you plan to use for the DHCP Failover servers. Use the same interface label for each physical interface.
NOTE: DHCP failover is time sensitive. For DHCP failover to function properly, both the Primary and Secondary DHCP servers must be configured to use the same NTP server.
Procedure
On the DHCP Failover dialog box of each EdgeConnect, configure the settings listed below to provision your DHCP failover servers.
- Navigate to the DHCP Failover configuration screen.
- Go to the Configuration > Interfaces screen.
- Click the edit icon of the interface you will use. This opens the Interfaces configuration screen.
- Click the IP address of the Hardware interface you will use (for example, lan0). This opens the Deployment screen.
- Click the NoDHCP/No RA link to open the DHCP Settings / Router Advertisements screen.
- Configure or update the DHCP settings as needed.
- Select the DHCP Failover check box.
- Click the Failover Settings link to open the DHCP Failover settings screen.
- Configure the DHCP Failover settings listed in the table below. Click Save to save your settings. Upon saving the configuration, the associated EdgeConnect LAN interface(s) show that they are in use for a DHCP server. Repeat this procedure to configure the DHCP Failover server peer on the other EdgeConnect.
Field | Use | Comments |
---|---|---|
Role | Select Primary or Secondary. | Differentiates the DHCP Failover peer servers, particulary when the optional SPLIT option is used to determine the percentage of DHCP requests each peer DHCP Failover server handles. |
My IP | IP address of the EdgeConnect LAN interface you assign to the DHCP server. | Must be on the same Layer 3 subnet as the Peer IP below. If you have multiple VLANs/Sub-Interfaces configured under one physical interface (for example, LAN0), choose any of the interface IP addresses as My IP and the corresponding EdgeConnect LAN interface IP of the peer as the Peer IP. |
My Port | TCP port number of the LAN interface. | TCP port 647 is the default port EdgeConnects use to establish a peer DHCP server state and synchronize DHCP leases between them. |
Peer Port | TCP Port number of the peer DCHP server. | TCP port 647 is the default port EdgeConnects use to establish a peer DHCP server state and synchronize DHCP leases between them. |
MLCT | Optional. The Maximum Client Lead Time (MCLT) default is 60 minutes. This field cannot be zero. | This setting only impacts DHCP renewal requests when one of the DHCP servers is down. It defines the maximum amount of time that an EdgeConnect can extend a lease for a DHCP client during a failure of a peer DHCP server that issued the original lease. Using a low value causes the DHCP renewal clients to too frequently look for lease renewals. The default of 60 minutes (3600sec) typically allows enough time for the failed DHCP server to have recovered. |
SPLIT | Optional. Controls the percentage of DHCP requests that each EdgeConnect handles. | The range is 0 - 255, with 128 being the default value. A SPLIT value of 128 distributes the load evenly distributed between the two EdgeConnects. For example, a SPLIT value of 64 for the Primary Dhcp Failover server would result in it handling 75% of the client requests and the secondary server handling 25%. |
Max Response Delay | Optional. Determines how long the EdgeConnect delays its response to a client request, such as a lease request or renewal, if it hasn’t received a timely acknowledgment from its failover partner. | By waiting for a defined period, the EdgeConnect can ensure that it does not prematurely take over leases or assignments that the failover partner might still be handling. This setting provides a buffer to accommodate temporary network issues or delays in communication between EdgeConnects. |
Max Unacked Updates | Defines the maximum number of update messages (BNDUPD) that one EdgeConnect can send to its failover peer without receiving an acknowledgment (BNDACK) before it takes action to mitigate potential issues. | Limiting the number of unacknowledged updates helps ensure that both EdgeConnects maintain consistent lease information, reducing the risk of IP address conflicts. |
Load Balance Max | Optional. Allows you to configure the number of seconds to wait for a cutoff, after which load balancing is disabled. | The cutoff is based on the number of seconds since the client sent its first DHCP DISCOVER or DHCP REQUEST message. This only works with clients that correctly implement the SECS field (most clients support this). Set this in the range of 3 to 5. The result is that when one of the EdgeConnects is responding to failover keepalives but not responding to actual DHCP requests from clients, the peer EdgeConnect takes over its client load automatically as the clients retry. |
NOTE: In a DHCP Failover configuration, if failover is checked but the ‘my IP’ and ‘peer IP’ are not provided, and you try to remove the pair, the system complains that ‘My IP’ and ‘Peer IP’ need to be added because failover is checked. In this case, uncheck failover, save the changes, and then remove the pair.
DHCP Failover Settings Examples
The DHCP failover settings below are for the DHCP servers of two EdgeConnects, each with identical failover settings and correct Role, My IP, and Peer IP settings. You can configure one or more failover groups. The examples provided below are for a single failover group and for two failover groups.
Each failover group must be configured on a per physical interface basis. For example, if you select the DHCP Server for any of the LAN0 subnets and configure the Failover Settings, then those failover settings are retained for all the DHCP servers under LAN0 interfaces. If LAN0 has VLAN2301 and VLAN2302 on it, they would have the same DHCP failover settings. Likewise, a LAN2 interface DHCP server has its own failover settings. Use the examples below as a reference for how to configure DHCP failover settings. You can also use a pre-configuration template to standardize DHCP failover settings.
Single DHCP Failover Group Example
-
EdgeConnect 1 DHCP Server
- Interface: lan0 (IP/Mask: 172.23.2.3/25)
- Role: Primary
- MY IP: 172.23.2.3
- Peer IP: 172.23.2.4
- Port: 647
- MCLT: 3600
- SPLIT: 128
- Max Response Delay: 10
- Max Unacked Updates: 10
- Load Balance Max: 5
-
EdgeConnect 2 DHCP Server
- Interface: lan0 (IP/Mask: 172.23.2.4/25)
- Role: Secondary
- MY IP: 172.23.2.4
- Peer IP: 172.23.2.3
- Port: 647
- MCLT: 3600
- SPLIT: 128
- Max Response Delay: 10
- Max Unacked Updates: 10
- Load Balance Max: 5
Two DHCP Failover Groups Example
-
EdgeConnect 1 DHCP Server LAN0 Group
- Interface: lan0 (IP/Mask: 172.23.2.3/25)
- Role: Primary
- MY IP: 172.23.2.3
- Peer IP: 172.23.2.4
- Port: 647
- MCLT: 3600
- SPLIT: 128
- Max Response Delay: 10
- Max Unacked Updates: 10
- Load Balance Max: 5
-
EdgeConnect 2 DHCP Server LAN0 Group
- Interface: lan0 (IP/Mask: 172.23.2.4/25)
- Role: Secondary
- MY IP: 172.23.2.4
- Peer IP: 172.23.2.3
- Port: 647
- MCLT: 3600
- SPLIT: 128
- Max Response Delay: 10
- Max Unacked Updates: 10
- Load Balance Max: 5
-
EdgeConnect 1 DHCP Server LAN2 Group
- Interface: lan2 (IP/Mask: 172.23.3.3/25)
- Role: Primary
- MY IP: 172.23.3.3
- Peer IP: 172.23.3.4
- Port: 647
- MCLT: 3600
- SPLIT: 128
- Max Response Delay: 10
- Max Unacked Updates: 10
- Load Balance Max: 5
-
EdgeConnect 2 DHCP Server LAN2 Group
- Interface: lan2 (IP/Mask: 172.23.3.4/25)
- Role: Secondary
- DHCP Failover: Secondary
- MY IP: 172.23.3.4
- Peer IP: 172.23.3.3
- Port: 647
- MCLT: 3600
- SPLIT: 128
- Max Response Delay: 10
- Max Unacked Updates: 10
- Load Balance Max: 5
DHCP Failover Fundamentals
EdgeConnect DHCP Failover settings under DHCP Server Configuration enable you to provision DHCP Failover server peers in your network. THe topics below explain how DHCP Failover servers operate.
DHCP Failover Primary or Secondary Roles
In the DHCP Failover settings screen, Role specifies how the pool of IP addresses is managed across the DHCP Primary and Secondary servers.
NOTE: The DHCP server Failover role does not refer to which server is active, and which is backup.
The pool of IP addresses the DHCP Failover servers manage is specified by the Layer 3 subnet that both the primary and secondary DHCP servers must be configured to use. That is, the interface subnet mask specifies the pool of addresses the DHCP Failover servers manage. For example, if a DHCP Failover server is configured to use lan0 with its IP/Mask set to 172.23.2.3/25, the pool would have 126 addresses (128 minus two, one for the network address and one for the broadcast address).
DHCP Active and Backup Server Behavior
During failover events, observe the following distinctions between the backup and active server designations:
-
Active-Backup Pair: In DHCP Failover mode, you are deploying two DHCP servers, designated as “active” and “backup.” The Primary DHCP server does not always service the requests. The active server handles all DHCP requests and maintains the DHCP lease database.
-
SPLIT: The configuration option that specifies the percentage split of the DHCP requests that each EdgeConnect handles. The SPLIT values are 0 to 256. The default of 128 distributes 50% of the load to each DHCP Failover server.
-
Lease Assignment and Failover: The active server assigns IP addresses and other configuration parameters to clients. If the active server fails, the backup server seamlessly takes over, ensuring uninterrupted DHCP service.
-
Lease Database Synchronization: The active server continuously synchronizes its DHCP lease database with the backup server. This ensures that the backup server has the latest information and can take over without any lease conflicts or disruptions in case of a failover.
DHCP Database Synchronization
DHCP Failover synchronization does not rely on an Edge-HA link. In DHCP Failover mode, DHCP servers synchronize their databases in these ways.
-
Real-time Updates: The active DHCP server sends real-time updates to the backup server whenever a lease is assigned, renewed, released, or expires. This ensures that both servers have the most up-to-date information.
-
Full Synchronization: Periodically, a full synchronization of the DHCP lease database is performed between the active and backup servers. This acts as a safeguard to ensure complete consistency in case any updates were missed during real-time synchronization.
FAQs
In a split-brain scenario, would both EdgeConnects provide IP addresses for DHCP requests?
A split-brain scenario is when the DHCP Failover servers are not communicating and synchronizing their data. Once the DHCP Failover servers detect the peer is unavailable, both serve all clients. New clients get DHCP offers from both. The client chooses one. Client DHCP lease renewal is different. If the EdgeConnect that provided the previous DHCP lease is not responding, the renewal fails for its clients. If the client eventually sends out new DHCP Requests (DORA process Discover-Offer-Request-Ack), an active EdgeConnect responds.
In a split-brain scenario, could an EdgeConnect assign a duplicate IP to a second device?
No. The EdgeConnect DHCP-Server uses a standards-based implementation that provides a mechanism to avoid duplicate IP addressing scenarios. EdgeConnects do a standard ARP request prior to issuing DHCP requests, thus preventing assignment of duplicate IP addresses.
What happens if one sub-interface is disabled?
If a disabled sub-interface whose IP address is not configured as My IP under the failover settings, the DHCP server does not provide DHCP services for that sub-interface range. If a disabled sub-interface whose IP address is configured as My IP under the failover settings, then DHCP failover is triggered for all the DHCP servers configured on that interface. This causes the local DHCP server to be unreachable from the DHCP peer perspective. The peer DHCP server then serves all new clients and honors the DHCP renewals that were issued by the previous DHCP server that is now unreachable. Likewise, if the physical interface goes down or the peer IP is unreachable, then the DHCP failover trigger happens.