I just always kind of assumed idrac and I’ll where dangerous as shit and kept them air gapped.
Why the fuck would you let those touch anything else?
That sounds like a tiny number…
As someone who bought a Supermicro board 6 weeks ago and was super happy to learn about IPMI: oh no.
I flashed the latest bmc firmware and bios the day I got it running (also fuck paywalling the bmc UI for bios updates), but… 😬
E: holy shit it truncates passwords to 19 characters. One of my weakest passwords, in the bottom 10 of like 450 passwords.
E2: I can’t disable KCS, I can’t disable the IPMI (short of unplugging the ethernet cable), I can’t disable the default user or limit the privileges, so I had delete my user, login as the default, then change the username and password to my user creds. Like, wow.
IPMI/BMC have been known problems since 2015, if not before.
Matthew Garrett: IPMI - because ACPI and UEFI weren’t terrifying enough
Really interesting watch, thanks :)
Not an article about waiting staff.
Baseboard management controllers are miniature computers that are embedded into the motherboards of virtually every enterprise server. The microcontrollers, typically abbreviated as BMCs, run with their own operating system firmware, network stack, and IP address. Administrators rely on them to monitor the physical status of large fleets of servers and to perform a variety of tasks, including rebooting machines, installing updates, and even reinstalling operating systems. BMCs provide what’s known as “lights out” and “out-of-band” management because they work even when servers they’re attached to are turned off or are unresponsive.
Hmm. I don’t really understand what the major benefit is to having an integrated management system over just providing a BIOS that can expose its interface on a serial port and having a serial device server attached to the thing, which then means that you aren’t tied to a particular management system. I’d also kinda think that it’d be cheaper to use a serial device server if you have a ton of servers in a rack that need to be managed, since one serial port server can handle all of them. Use a smart PMU for power control.
Maybe interfaces are more standardized or something, for automated tasks?
The benefit is when you have a huge number of servers arround the world you can manage them from a few locations without having to have someone go physically touch them. Bonus points they often allow for remote wiping and provisioning, it makes it possible for a few peoole to manage the entier life of a server without ever touching it. This is a lot cheaper than flying people out to each site whenever something goes wrong.
Right — that’s why I’m talking about a PMU and serial device server linked to a serial console and BIOS as an alternative. That’s another way to accomplish that.
My guess was that IPMI might provide a more-standardized interface, and notabot said that this was the case.
When you’re dealing with a large fleet of servers you need a system that you can automate. IPMI is (sort of) standardised, so you can use the same tooling to monitor and manage all of your servers, regardless of model or manufacturer. A serial based interface would put a lot more friction in the way of that, and mean you have to manage and maintain even more devices.
You can mount an ISO image through the BMC and boot the server from there. A serial port would be too slow. The management ethernet is supposed to be on a separate network not exposed on the same network as the rest.
You can mount an ISO image through the BMC and boot the server from there.
I mean, you can typically do a TFTP boot on the systems that I’ve seen that have serial-capable BIOS setups.
considers
Heck, I think that all of my modern personal non-serial-port-BIOS systems can do that, even.
