Hi Konrad,
you are right, if the network connectivity is lost via the two high speed switches, we need a physical access to the HCI and a direct connection via 1Gbit.
This will allow us to access the node coreOS and the switches.
It is a viable approach to connect a small jumphost to one of the 1Gbit switches and the customer network.
On the 1Gbit switches, we only use IPv6, so there are no duplicate IPs.
I have seen this on internal lab systems, but not yet on a customer installation.
Feel free to call me if you have furhter questions.
------------------------------
Daniel Danner
------------------------------
Original Message:
Sent: Thu June 08, 2023 10:42 AM
From: Konrad Trojok
Subject: Storage Fusion HCI management access
Hello Community,
I am certain some of you might have run in similar situations. We needed to shutdown the Fusion HCI due to a power maintenence. Some little detail was lost on the networking configuration - the portion where DHCP helper information was configured. When powering back up, nothing really happened, since without DHCP nothing will start. We were quite swiflty certain how to fix it, yet we needed someone to go to the rack and locally check the DHCP helper config on the highspeed switches.
My question is not about this issue, but was raised by this incident:
Do you have a little local jump host in the Fusion HCI rack to have a "emergency" access to the hardware management? I was thinking about installing a NUC with two interfaces, where the inward facing interface is connected to the 1G switch, while the other interface connects to the company network. What are your general thoughts about such a device?
kind regards
Konrad
------------------------------
Konrad Trojok
Head of Competence Center IBM Storage
#IBMChampion #SVA
------------------------------