My EPYC Quest (for failure?)

2»

Comments

  • I'm a little late but will add my 2c anyway given I have a small time 2U colocation as well.

    @Freek said: 2️⃣ KVM & IPMI Access in a Datacenter

    Any half decent provider will give you a secure option to access your IPMI. With my colo I can access the iDRAC interface through an IPsec VPN. You're better off using the built in Supermicro IPMI as it'll have access to hardware parameters and diagnostics if the machine becomes unbootable. No IPMI nor KVM can be safely exposed to the Internet, nothing which has that kind of access to your hardware can ever be secure enough to be trustworthy.

    @Freek said: 3️⃣ Power Estimation for Co-Location

    Most IPMIs can tell you the system power usage. Alternatively you can get a cheap power meter from the hardware store that's accurate enough. What you want to do is measure the system's power usage at idle with your OS booted to ascertain the baseline, then start adding load to it and see how it changes.

    Generally the power consumption will be quite high while POSTing and then settle down once the OS has booted and the fans ramp down, power saving features kick in etc. Buy power according to that as generally you can get away with brief spikes above your allocation, but check with your provider that they won't cut the juice if you go over 1A or whatever for a minute.

    @Freek said: 4️⃣ How Does IP Addressing Work in a Datacenter?

    So for your needs, you'll generally be working with a provider who partners with a datacenter and can handle the networking, racking, etc. for you. Generally they'll give you a static IP address or a small subnet like a /29. There's some smaller datacenters that'll handle everything for you, but big boys will want you to go via a partner to handle everything.

    This is all to say that my recommendation is that you do not colocate this hardware! The reason is that it's a potentially unstable config since you're using random 2nd hand memory that doesn't match, an engineering sample CPU and a motherboard that is hacked with random BIOSes downloaded off of STH to make it work.

    You do not want to colo hardware that is potentially unreliable as you'll get absolutely destroyed by remote hands fees trying to get things working. In my opinion, get one of those ASRock 1U Ryzen servers if you want something to colo and do it right with new hardware that you burn in.

    Also regarding setting up a software router within Proxmox, I made a post about this previously and would suggest you give it a read: https://lowendspirit.com/discussion/comment/203907#Comment_203907

    You do not want to have Proxmox exposed to the Internet, while there are no disclosed vulnerabilities right now, it's just never going to be a secure setup.

    Anyway good luck with the hardware.

    Thanked by (3)quicksilver03 Freek Andre
  • FreekFreek Senpai

    Every day I work from home, this thing lies on my desk, staring at me. I want to finish it, but I’m stuck. I don’t know which tool can reliably monitor the values of PPT, EDC, and TDC on Ubuntu Server.
    Does anyone know of a tool or utility that can monitor the set values of PPT, EDC, and TDC?

    Thank you.

    Thanked by (1)Not_Oles

    LinuxFreek.com — Hosted on 🇪🇺 Scaleway Stardust with Native IPv6 | IPv4 Proxy, WAF & DNS powered by 🇳🇱 DutchIS

  • @Freek said:
    Every day I work from home, this thing lies on my desk, staring at me. I want to finish it, but I’m stuck. I don’t know which tool can reliably monitor the values of PPT, EDC, and TDC on Ubuntu Server.
    Does anyone know of a tool or utility that can monitor the set values of PPT, EDC, and TDC?

    Thank you.

    Maybe this? https://github.com/ocerman/zenpower.git

    It should show something like:

    zenpower-pci-00c3
    Adapter: PCI adapter
    Tdie:             +54.5°C
    Tctl:             +54.5°C
    PPT:              142.00 W
    TDC:               95.00 A
    EDC:              140.00 A
    CPU Core Voltage:  1.35 V
    
    Thanked by (2)Not_Oles Freek

    If you want information, feign ignorance reply with the wrong answer. Internet people will correct you ASAP!
    It’s OK if you disagree with me. I can’t force you to be right!

  • FreekFreek Senpai

    @somik said:

    @Freek said:
    Every day I work from home, this thing lies on my desk, staring at me. I want to finish it, but I’m stuck. I don’t know which tool can reliably monitor the values of PPT, EDC, and TDC on Ubuntu Server.
    Does anyone know of a tool or utility that can monitor the set values of PPT, EDC, and TDC?

    Thank you.

    Maybe this? https://github.com/ocerman/zenpower.git

    It should show something like:

    zenpower-pci-00c3
    Adapter: PCI adapter
    Tdie:             +54.5°C
    Tctl:             +54.5°C
    PPT:              142.00 W
    TDC:               95.00 A
    EDC:              140.00 A
    CPU Core Voltage:  1.35 V
    

    Thank you. Let me look into this and get back to you :)

    Thanked by (1)Not_Oles

    LinuxFreek.com — Hosted on 🇪🇺 Scaleway Stardust with Native IPv6 | IPv4 Proxy, WAF & DNS powered by 🇳🇱 DutchIS

  • @Freek said:

    @somik said:

    @Freek said:
    Every day I work from home, this thing lies on my desk, staring at me. I want to finish it, but I’m stuck. I don’t know which tool can reliably monitor the values of PPT, EDC, and TDC on Ubuntu Server.
    Does anyone know of a tool or utility that can monitor the set values of PPT, EDC, and TDC?

    Thank you.

    Maybe this? https://github.com/ocerman/zenpower.git

    It should show something like:

    zenpower-pci-00c3
    Adapter: PCI adapter
    Tdie:             +54.5°C
    Tctl:             +54.5°C
    PPT:              142.00 W
    TDC:               95.00 A
    EDC:              140.00 A
    CPU Core Voltage:  1.35 V
    

    Thank you. Let me look into this and get back to you :)

    Sure! I don't have any ryzen Linux machine so can't test it.

    Thanked by (1)Not_Oles

    If you want information, feign ignorance reply with the wrong answer. Internet people will correct you ASAP!
    It’s OK if you disagree with me. I can’t force you to be right!

  • I have good news and bad news. Let’s start with the bad news: this project will be on hold for about six months. The reason is that we’re moving to a new house, and according to the realtor, “having a computer on the ground that is getting open-heart surgery” is not a good selling point for our current place. Here’s a final picture before I put it into the storage unit:

    Now for the good news: in our new house, I’ll finally have space for my own server rack, which has been a wish of mine for a long, long time. This means I can tinker with it more often and no longer need to colocate it. Previously, I was very limited in when I could work on it because it made a ton of noise—too much to run when our baby boy was asleep. And when he was awake, well, he needed attention. So the server was mainly collecting dust and attacking my pinky toe.

    That’s all I have for now.

    LinuxFreek.com — Hosted on 🇪🇺 Scaleway Stardust with Native IPv6 | IPv4 Proxy, WAF & DNS powered by 🇳🇱 DutchIS

Sign In or Register to comment.