Ubu

From CHG-Wiki
Revision as of 12:01, 20 December 2016 by Libby (talk | contribs) (Added chg-rcmrd and IP for chg-ewxtest.)
Jump to navigationJump to search

Ubu is CHG's main workhorse server and is also home to the chg-ewx VM.

ubu.chg.ucsb.edu

General Info

System Name: ubu.chg.ucsb.edu
IP: 128.111.236.129
IPMI IP: (need info)
Location: EH ERI server room
UCID #: 148000220
Grant #: 375FC
Serial #: AH1002841 (Mercury AH-RM424)
Custody Code/Description: 8-5092-22554 USDC NA11OAR4310151 FUNK/HUSAK
General Purpose: Workhorse server
Vendor: Advanced HPC
Contract #: (need info)
Purchase Date: April 2014
Delivery Date: May 2014
Support Expires: Three-Year Standard Warranty on Parts and Labor.

System Configuration

  • Model Number: Mercury RM424 4U 4x3.3GHZ
  • OS Type: Unix
  • OS Version: CentOS 6.3
  • CPU info: Four Intel Xeon E5-4627v2 3.3 GHz Eight Core CPUs
  • Motherboard Specifications:
    • Quad Socket R (LGA 2011) Supports Intel® Xeon® Processor E5-4600 Intel® C602 chipset
    • Up to 1TB* DDR3 1600MHz ECC Registered DIMM; Thirty-Two DIMM Sockets (Depends on Memory Configuration)
    • Expansion Slots:
      • Two PCI-E 3.0 x16 and Two PCI-E 3.0 x8
      • Intel® X540 Dual Port 10GBase-T
      • Two SATA3 Ports + Four SATA2 Ports
      • Eight SAS2 Ports via LSI 2208
      • Integrated IPMI 2.0 + KVM over LAN
      • Eight USB 2.0 Ports (Four Rear, Three Front + One Type A)
  • Chassis Specifications:
    • High Computing Performance and High Storage Capacity 4U Storage Chassis - 29" Depth
    • 4-Way Processor and High Storage Capacity Applications
    • Twenty-Four 3.5" SAS / SATA Hot-Swap Drive Bays with SES2, Mini-iPass (SFF8087) Connectivity, 6 Gb/s SAS2 Expander
    • Eleven (7+4)x Full-Height, Full-Length PCI Expansion Slots
    • 1620W (1+1 redundant) Platinum Level High-efficiency Power Supply (Optional 2+1 Redundant)
    • Four 92x38mm and Two 80x38mm Hot-Swap Cooling Fans
    • Power Switch, Reset Switch & Six LED Indicators
    • Full SES2 Support is Available with SAS Motherboards and Other Compatible Components.

Network

  • NIC speed: Gigabit
  • MAC Address - eth0: 00:25:90:5B:5B:C0
  • MAC Address - eth1: 00:25:90:5B:5B:C1
  • IPMI notes: 128.111.234.188, ACLs confirmed in place, not sure if IPMI is still "borrowing" eth0 or not.

Storage

  • Server Rackmount, 4U
  • Controller: LSI 9211-4i SAS HBA
  • OS SSDs: Two 160GB Intel DC S3500 SSDs
  • HDDs: Eleven 4TB Hitachi Ultra SAS 7200RPM 64MB Cache Disk Drives
  • Memory: 512Gb DDR3-1600 Registered ECC Memory (32x16GB)

Other Hardware

  • GPU: NVIDIA Quadro K2000 Graphics Card (failed April 6th, 2015)

Services

VMs

System Name: IP: Notes:
chg-ewx 128.111.236.142 EWX Development VM, temporarily moved to zippy after some issues Libby 11:23, 14 September 2015 (PDT)

Moved back to ubu after ubu stabilized.

chg-ewxtest IP: 128.111.236.143 Test VM for EWX, not usually on
chg-rcmrd IP: 128.111.236.135 Windows Server 2012 VM for testing and documentation to deploy to RCMRD.

Notes

  • Ubu was moved from EH 1609 to the sixth floor EH ERI server room on November 23rd, 2015. It's domain was also changed from geog.ucsb.edu to chg.ucsb.edu at that time.
  • Used ubu shelf spare on zippy, July 7th, 2014.
  • Graphics card failure on April 6th, 2015. All systems wedged and during reboot ubu got stuck (at BIOS stage 91) until graphics card was pulled ( mtc tried pulling RAM and a few other things first to no effect). mtc will handle the RMA for the graphics card.
    • Update on GPU: Same BIOS issue with new graphics card. Libby 12:37, 4 May 2015 (PDT)
  • 32 TB /raidb added on September 10th, 2015 using ZFS filesystem. Libby 11:23, 14 September 2015 (PDT)
  • A number of CPUs wedged over the weekend, rebooted by Aaron via IPMI, some load strangeness but otherwise operating. Libby 11:26, 14 September 2015 (PDT)
    • On September 15th, OIT notified CHG that a Bitcoin miner (minerd) was detected on ubu, thus explaining the intense load issue that was not showing up via the top command.
      • Ubu was infected via an IPMI exploit - though ACLs had been requested on August 16th, 2015, they had not been put in place due to a communication failure. ACLs in place as of mid-September, 2015.
      • Once it was determined that ubu was indeed rooted, ERI moved to take ubu down and do a complete reinstall, taking the opportunity to upgrade ubu to CentOS7 and gnisv2.