Hi All,
Quick question: I installed a six node cluster end of 2023. Consists of two A250 and a C800.
We used the nVidia SN2100 as cluster switches. Installed with the rcf file on Cumulus 5.6.
Reading the documentation, I now only read 5,4 is supported / mentioned.
Sometimes: minimal 5.4
Sources:
https://mysupport.netapp.com/site/info/nvidia-cluster-switch
https://mysupport.netapp.com/site/products/all/details/nvidia-cluster-storage-switch/downloads-tab/download/64386/StorageRCF_1.10/downloads
Having trouble finding the config in IMT.
Does anyone know if we need to downgrade or can we run 5.6 in production?
Thanks!!
#Nvidia SN2100 - supported version
green sun
past star
HWU is pretty clear in that only 5.4 is supported
so if you want to be on the safe side I would downgrade ...
green sun
We needed to upgrade to 5.6 last year to be supported
green sun
Any downgrade procedures / documentation available?
twin island
I don't find any info regarding recommendations for Cumulus 5.6. Also checked Metrocluster configs and even for a C80 MCC-IP which should be the newest released Metrocluster only 5.4 is supported.
Interestingly this KB mentions 5.9 as the recommended version for MCC-IP with 9.15.1... might be a typo?
https://kb.netapp.com/on-prem/ontap/mc/MC-KBs/Recommended_MetroCluster_component_releases_on_the_NetApp_Support_Site
In the NetApp docs I don't find any downgrade guide, in the Nvidia docs only this: https://docs.nvidia.com/networking-ethernet-software/cumulus-linux-56/Installation-Management/Upgrading-Cumulus-Linux/#roll-back-a-cumulus-linux-installation
Regarding support for other Cumulus versions:
"Call NetApp Technical Support if during the course of any NVIDIA engagement Cumulus version or configuration changes are recommended"
I would really involve NetApp support here since breaking your cluster switches is never good
green sun
Thanks @twin island
green sun
Just an update. Cumulus 5.6 has been added to IMt as supported!
clever crypt
Just an update. Cumulus 5.6 has been added to IMt as supported!
Unfortunately, that does not apply for ONTAP cluster switch use-cases. See: https://mysupport.netapp.com/site/info/nvidia-cluster-switch
clever crypt
Cumulus Linux 5.11 support and RCF 1.13 for NVIDIA SN2100 cluster or storage switches are now published on NSS:
https://mysupport.netapp.com/site/info/nvidia-cluster-switch
CL 5.11 requires RCF 1.13 and vice-versa, from a compatibility perspective.
green sun
Also suppported for intercluster
green sun
Just upgraded a customer to Cumulus 5.11 and RCF 1.13. Went very smooth. The switches are used in a 6-node cluster Running 9.15.1P7
Take note of this KB: https://kb.netapp.com/on-prem/Switches/Nvidia-KBs/Cannot_set_eth0_interface_IP_address_in_Nvidia_cluster_switch_due_to_DHCP
I already asked to add this command to the documentation.
obsidian sphinx
Hello
There seems to be a mess concerning SN2100 recommandation for Metrocluster...
HWU indicates :
RCF Generator 1.7a recommends 5.7 that seems to be not supported :
RCF Generator 1.9a recommends 5.11 :
But download page is saying RCF Generator is wrong ???
What is official statement from NetApp ? 😅
lunar linden
RCF 1.10 -> 5.4.0
RCF 1.13 -> 5.11.0, 5.11.1, 5.4.0
Sometimes HWU does not align with the Support Matrix on the Support-switch matrix (above)
obsidian sphinx
This page does not apply to Metrocluster
Metrocluster RCF has to be generated using RCF Generator. And this tool is supposed to be THE reference
lunar linden
The inital post was not about MCC. it was just ONTAP. someone decided to "merge" ONTAP and MCC here.
I posted the link that should be followed