kate3
Member
- Joined:
- Jul 2024
- Posts:
- 208
- From:
- Utrecht, NL
NetBox. Full stop.
Virtualization tax on your brain is real when you're guessing which host runs what. Spreadsheet chaos scales linearly; NetBox scales with your ASN. I migrated 60 hosts last year. IPAM, circuits, device roles, all in one. OpenVZ-era tools won't cut it.
Kernel-level detail: if you're not tracking cgroup limits per host, you're flying blind. Documentation is infrastructure. Treat it like one.
virsh list --all | wc -l: 47
steveipw
Member
- Joined:
- Jun 2024
- Posts:
- 163
- From:
- Denver, US
- NetBox for IPAM and rack elevation
- Custom ZFS dataset naming for pool mapping
- Auction server serials in plain text file
- For that money you get a used spreadsheet and regret
Seriously though, 40 servers without asset tracking? You're one disk failure from archaeology. I label physical drives with purchase date and pool. NetBox handles the rest. Hardware without history is just scrap metal.
zfs send | zfs receive. repeat.
kate3
Member
- Joined:
- Jul 2024
- Posts:
- 208
- From:
- Utrecht, NL
Docker Compose. One file. Ten minutes. I have done this migration three times now.
The real cost is not the install. It is the data entry. But you pay that cost once, or you pay it every time a disk dies at 3am and you are grepping dmesg through a KVM lagging like a dreamcast.
virsh list --all | wc -l: 47
steveipw
Member
- Joined:
- Jun 2024
- Posts:
- 163
- From:
- Denver, US
Ansible inventory pulls from it.
This is the real answer for a lot of us. Static inventory from Git, dynamic from NetBox. Hybrid approach. I do not track ZFS pool layouts in NetBox though. That lives in a script that emails me monthly. Some things are too fluid for a database with history you do not want to explain.
zfs send | zfs receive. repeat.