This one has caught me out more than once. The default setting is “DCHP with failover” which means that if your switch comes up before the DHCP server (entirely feasible), it sets itself up on 192.168.88.1 and then doesn’t seem to change when the DHCP server comes up. This means that you run the risk of having two devices (router & switch) on 192.168.88.1…
So what’s best practise here? I’m thinking that either moving the router to 192.168.88.254 because then if you add a new switch, all that will happen is that it fights with other switches you’ve forgotten. Or a totally different subnet like 192.168.90.0/24?
Best practice is not to use any of the MikroTik / TP-Link / Ubiquity / etc. default IP ranges used in the default settings.
(This does not mean the idiocy of deleting all configuration of the routerboard, including the firewall)
In addition to using custom IP subnet as recomended above, never ever use DHCP client (not even fallback) on LAN infrastructure devices (and servers). If needed, attach a sticker with statically configured IP address printed if you fear to forget it.
More and more IT- “Expert” i get to talk to, tell me to use DHCP for both IPv4 & IPv6
Not only for Client-Devices but also for Servers and/or Infrastructure devices.
Of course for Servers and co, with other Parameters like Static Leases, longer Lease times, etc…
The reasoning being that, it provides the necessary flexibility needed in a “modern” Network.
Furthermore reducing operating costs when handling changes and IT-documentation.
a bit off Topic… nevertheless care to share your opinion?
My own best-practice. As @OP observed, a site-wide power outage can happen and in that case DHCP server might boot slower than some DHCP clients leaving them unconfigured (or mis-configured). My opinion is that essential infrastructure needs resilience against failed support services (such as DHCP server).
Many so called IT-experts take some things for granted (i.e. anything out of scope of their responsibility), so every IT-expert (we are IT experts, aren’t we? ) should take advice from fellow expert with a pinch of salt. Which includes my advice from this post
I tend to agree that DHCP is the way to go for less crucial servers (as in: not so important for operating network itself) and specially for client devices and common gadgets (e.g. networked printers, coffee machines and ATMs).
So @mkx, you prefer to assign static IPs on the devices ?
Personally never had a problem with the DHCP server since i can remember … Ofcorse if static IPs are set then you don’t care if the DHCP server fails or not and your devices will keep communicating with the network ( at least with the rest of the devices that have static IPs as well )
To the extent described in my previous post … yes. And that includes IPv6, so far my ISPs never changed prefixes (I always request static prefix and IPv4 address(es)).
) should take advice from fellow expert with a pinch of salt. Which includes my advice from this post 