I set up Immich on a spare mini PC running Linux Mint with a spare external HDD to store the library. Used Docker compose as recommended on the website. Everything was working great last night. Then this morning I decided to move the mini PC to a more permanent location. I powered down and moved it. When everything restarted though I discovered that I was unable to c9nnect to the immich server. I went onto the mini PC and the local host and all it says is, “connection was reset.” I have tried restarting docker from terminal several times and it says the immich services are running. I still cannot connect to the server though. What can I do to fix this?
don’t do this, use DHCP reservations instead so you actually have a list of all your servers and most routers register hostnames in DNS for you which is even better.
Confused what you mean.? I have a range of IPs for my servers and such IE 192.168.2.2-12. And then the DHCP pool 192.168.2.13-xxx. I set the servers IP static on its OS. And then set a static in my routers settings for the same IP I set in the OS. What is it that I would be doing incorrect in this setup?
Instead of doing a manual action in two different places and having to keep them in sync, just do it once on the DHCP server. Setting a static IP on the server is superfluous.
While handy on a personal net, on a larger corporate net this isn’t practical and even adds a security risk. By having servers request leases you run the chance that someone gets into a segment, funds the ARP association for an IP/MAC combo and can take over a server’s spot simply by spoofing their own MAC to match at the time of lease renewal.
In the post above about setting a static address in two spots that in itself isn’t required either. So long as there are no duplicates you would just set the static address on the end device, then the network will sort it out with ARP ‘who has’ requests in local segments, or routing in the case of distinct subnets.
Edit: the duplicate I suppose could be referring to putting names into a DNS registry, in which case yes you would need that double entry, or just reference things by IP if the environment is small enough for it to be practical.
Corporate nets use 802.1X authentication, risk of a DHCP hijack is very low.
As someone who works in large corporate networks, we absolutely don’t assign static IPs outside of core network gear, it’s impossible to manage a fleet of servers in this way with scaling in mind.
Indeed they do use 11x but it’s still a possibility to cause issues. It’s entirely possible to manage a fleet of IPs across a net but it takes a solid plan organization plan. My company is big on the acquiring companies game where IP overlaps are a perpetual challenge when merging sites in and you need a mess of snat/dnat conversions to keep routing from getting in a knot.