Notes from the February 2016 patching

Changes

  • Migrated to scripted form of patching
  • New version is 6.7.1.1 on all systems
  • Migrated Lustre servers from Centos 5.7 to RHEL 6.7
  • Migrated Lustre servers from Lustre 1.8.9 to Lustre 2.4.3

Problems

  • Problems with force_over_24tb flag on OSTs
    • FIXED with tunefs.lustre and adding 'mountfsoptions='errors=remount-ro,extents,mballoc' to re-add original default options and remove the force_over_24tb flag.
  • Node-16 is down, it had problems prior to the upgrade (unreported). Powered off for now
  • The Mark5s changed their interface mapping. Eth0 became eth2, eth2 became eth0.
    • This actually fixed the long standing Lustre mounting problem
    • It broke Mark5fx01 mounting of san01 since eth3 (old 10.1.33.0 interface) became eth0. FIXED by tweaking /opt/bin/netconfig.sh
  • Node-01 doesn't start difx and calcserver properly on boot, seems to be a race condition
  • Mar. 24, 2016 krowe: Control-C still doesn't work. I'm surprised how often I need it.
  • Mar. 24, 2016 krowe: usno-serv-2 still doesn't rsync it's data partition from usno-serv-1

Resolved

  • Mar. 24, 2016 krowe: (V-38643) Turn off virt-who on servers because it makes a world writable file /var/run/virt-who.pid
    chkconfig virt-who off ; /etc/init.d/virt-who stop
  • Mar. 24, 2016 krowe: (V-38643) chmod files on data partition so that they are not world writable
    chmod 755 /opt/services/ganglia-web/ganglia/dwoo/compiled/templates/default/*
    chmod go-w /opt/services/tftpboot/*
    chmod -R o-w /opt/services/ganglia-local/var/rrds
  • Mar. 28, 2016 krowe : (V-38580) Does not apply to us. All our machines pass.
  • Mar. 28, 2016 jrobnett: applied the following changes (note the 3 kernel RPMs commands were not run on the Lustre servers:
    rpm -Uvh /home/usno-serv/updates/yum/RHEL-6.7.1.2/kernel/kernel-abi-whitelists-2.6.32-573.18.1.el6.noarch.rpm 
    rpm -e kernel-debug-devel-2.6.32-573.el6.i686
    rpm -e kernel-debug-devel-2.6.32-573.el6.x86_64
    sed -i -e 's/PASS_MIN_LEN *14/PASS_MIN_LEN    15/g' /etc/login.defs
    chmod 600 /var/log/boot.log
    chmod 644 /usr/lib64/librpfits.a
    

-- JamesRobnett - 2016-02-03
Topic revision: r7 - 2016-03-29, JamesRobnett
This site is powered by FoswikiCopyright © by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding NRAO Public Wiki? Send feedback