Table of Contents

mako{01..10}

Summary

Mark had been increasingly hearing that customers wanted details on how Ceph performed on non-Intel systems. Red Hat purchased 10 servers with AMD processors in Q1 2021.

These servers also have some donated Samsung MZ-QLB3T80 3.84TB (Model: PM983) in each. 8 in the first two nodes and 6 in the rest. 64 total.

Purchasing Details

Purchasing ticket: https://redhat.service-now.com/surl.do?n=PNT0915434
Racking ticket: https://redhat.service-now.com/surl.do?n=PNT0915435

Samsung drive carriers purchasing ticket: https://redhat.service-now.com/surl.do?n=PNT1008475
Racking ticket: https://redhat.service-now.com/surl.do?n=PNT1008476

DAC Cables purchasing ticket: https://redhat.service-now.com/surl.do?n=PNT0960669

Hardware Specs

Count Manufacturer Model Capacity Notes
Chassis 1U Dell PowerEdge R6515 N/A
Mainboard N/A Dell 0R4CNN N/A
CPU 1 AMD AMD EPYC 7742 128 cores
RAM 8 DIMMs Micron 18ASF2G72PDZ-3G2E1 16GB 128GB Total
SSD 2 Micron MTFDDAV480TDS 480GB Behind hardware RAID1 for OS
NVMe 1 Dell P4510 1TB For OSD journals?
NIC 2 ports Dell 1Gb On-board. Unused.
NIC 2 ports Broadcom 57416 BaseT 1/10Gb Oops. Won't be using this.
NIC 2 ports Mellanox ConnectX-6 100Gb 1 port as uplink
BMC 1 Quanta N/A N/A Reachable at $host.ipmi.sepia.ceph.com using usual IPMI credentials.

PXE/Reimaging

These PXE using Legacy/BIOS mode and can be provisioned via Cobbler normally.

Network Config

These nodes are connected to the officinalis QFX5200 (s/n WH0218170419 [formerly WH3619030401]) uplinked and managed by Red Hat IT. For an example of how to report an outage, see https://redhat.service-now.com/surl.do?n=INC1201508.

The 100Gb connection is the only uplink for now. The top-of-rack switch in that rack probably has capacity if we need a 1Gb uplink and reserve the 100Gb NIC for backend traffic.