IPMI IP Address Remapping

[leeroy]

TrueNAS-SCALE-22.02.0.1
Supermicro; X12SPi-TF
Network: bonded/Lagg
Console: dedicated network port

Hi, when testing some startup scripts to make some modifications via IPMITOOL, I noticed I couldn't resolve the IP. I tested this on my local terminal and got the following (Which worked):

Code:

ping 172.17.17.20
PING 172.17.17.20 (172.17.17.20) 56(84) bytes of data.
64 bytes from 172.17.17.20: icmp_seq=1 ttl=62 time=0.783 ms
64 bytes from 172.17.17.20: icmp_seq=2 ttl=62 time=0.783 ms

However, I try this from a terminal on my TrueNAS Scale box, I get the following:

Code:

ping 172.17.17.20
PING 172.17.17.20(172.17.17.20) 56(84) bytes of data.
From 172.17.0.1 icmp_seq=1 Destination Host Unreachable
From 172.17.0.1 icmp_seq=2 Destination Host Unreachable

Why would it look to `172.17.0.1`? This would explain why it can't reach that host/ipmi. I don't know what would cause this to happen? Even the GUI see's the IPMI is available and clicking manage, drops me into a browser at the correct IP. TrueNAS Scale and IPMI are on different VLANs. Not sure if this is an issue for TrueNAS?
Any help is much appreciated.

 

[leeroy]

Oddly this is not presenting itself on TrueNAS Core.

 

[Etorix]

You'd need to better explain your network settings (net/mask, gateway, what's plugged and how, addresses and how they're set,…).

 

[leeroy]

Thanks Etorix, hopefully this is enough:

IPMI Lan: 172.17.17.20/24
Edge port on switch: VLAN 100

NAS:
IP Address: 172.18.18.20/24
NameServer1: 172.18.18.1
Default Route: 172.18.18.1
Outbound Network: Allow All
Edge ports on switch LACP LAGG: VLAN 200
Network Card: Dual 10GBase-T with Intel® X550
Media Type: Ethernet
Media Subtype: 20000Mb/s Other
Interface Settings: bond0
Description: Link Aggregation of eno1 and eno2
DHCP: True
Autoconfigure IPv6: False
Link Aggregation Protocol: LACP
Transmit Hash Policy: LAYER2+3
LACPDU Rate: SLOW
Link Aggregation Interfaces: [eno1,eno2]
Disable Hardware Offloading: False
MTU: 1500
Aliases: None

There is no port isolation on the switch or firewall rules that would limit these two subnets from communicating. It seems TrueNAS Scale is the only item on the network that has access to this console address that doesn't seem to resolve.

 

[leeroy]

Also using `arp -e` I think I'm closer to the issue:

Code:

Address                  HWtype  HWaddress           Flags Mask            Iface
172.18.18.1              ether   aa:aa:aa:aa:aa:aa   C                     bond0
172.17.17.1                      (incomplete)                              kube-bridge
172.17.0.6               ether   bb:bb:bb:bb:bb:bb   C                     kube-bridge
172.17.0.7               ether   cc:cc:cc:cc:cc:cc   C                     kube-bridge

 

[leeroy]

I should also add that there is still NO issue hitting the BMC directly without the IP and all other remote management NOT on this machine works fine hitting that IP. Looking at my other startup scripts (which I haven't touched in years) all do this without an ip.

I think I was wrong about this NOT being an issue before. I think it was just different setups. I checked a couple other servers that have what I believe is a shared network controller sideband interface and seeing the same results and also explains the arp entries. Likely also explains why all hell broke loose when I blew away my arp table debugging this.