MPU Meeting Tuesday 30th October 2012

Server Upgrades

Nothing to report.

Server Hardware

Chris has documented the firmware reports for COs (Firmware Updates for Servers) and for the MPU (The Server Firmware Report).

Some varieties of hardware are not yet covered by the firmware reports, in particular BMCs, DRACs and network interfaces. Nevertheless we will list the available updates for these types of unit in the goodfirmware map along with the other firmware and BIOS updates, as it seems the best place, and will help prepare for the day when these types of unit are included in the report system.

Stephen suggested that it would be good to extend the system slightly so that:

  • applicable upgrades were automatically copied to a server's local disk every night
  • ... together with some instructions on applying them
  • ... and that in time the upgrade process could be somewhat automated, for instance to take care of the /etc/redhat-release dance.

Security Enhancements

Not much to report.

Inventory

Nothing happened.

Miscellaneous Development

kernel component
Alastair has tweaked it to require RPM2 where needed (instead of use right at the start); the change will be in this week's stable release (2012102901). As part of the same change Alastair will now add hooks into multipath and lvm components.
T5600 problem
The Dell Precision T5600 was showing an install problem: the Linux kernel wouldn't even start up. It turns out that it had a limit of 32MB on its initrd size but ours is 40MB. Alastair reduced its size by pulling out some unwanted bits of modules. He wants to go back and do this properly when necessary by including what's necessary rather than excluding bits which aren't. We could also do with reporting its size to the console as a check. For the new kernel Alastair will update initramfs with the current bug fix.
Default LCFG
Stephen has put the default lcfg level SL6 version to 6.3 in this week's release.

Operational

Metropolitan
Alastair has changed its default bridge to br0. Metropolitan is now ready for use as a KVM server.
Jubilee 6.3 upgrade
Chris will upgrade jubilee to SL 6.3. This will be our first KVM server on 6.3 so we'll leave it a week or so after the upgrade before doing any of the other KVM servers.
SAN-based storage pool on hammersmith
hammersmith needs to have a SAN-based LVM storage pool added for KVM use. We'll do this when we do its SL 6.3 upgrade, which will be done a week or so after rebooting jubilee.
External LCFG to Bakerloo
Now that bakerloo is free we can prepare to move the external LCFG functions to it from budapest, except for LCFG Bugzilla which will move to a VM.
Dropping of OpenID
Alastair has checked with the known users of our OpenID not-a-service and they haven't used it in ages and wouldn't miss it, so it looks safe to drop it. Alastair will blog and mail.
Live Defaults Tidy-up
Stephen has tidied up the live defaults file, removing some outdated old cruft.
Hogwood and the Wallet
Stephen reinstalled hogwood for its SL 6.3 update. The Wallet system restored the host's saved key, but unfortunately it hadn't worked the previous time hogwood was installed - apparently because it wasn't quite functional on SL6 at that time - so this upgrade restored the host key from the installation before last, so in the end we had a change of key after all.
Hogwood audit logs preserved
Stephen preserved the /var/log/audit partition when reinstalling hogwood.
Replacement for dunlin
Stephen would like us to replace dunlin with a 1950 so that its hardware matches that of hogwood. Alastair will add this claim to the list.
Spare machines
mousa, trondra and split are in a spare rack in IF-B.02. Once it's free Stephen will add tobermory to that rack too. Chris will correct the inventory details for mousa to reflect his having moved it to that rack. Alastair has a wiki page for spare machines is called ParkedKit.
BMC Update
Stephen and Chris dealt with an emergency BMC firmware update this week. Version 2.50 of the BMC firmware provided a critical security update for Dell PowerEdge 1950, 2900, 2950 and R805 servers (and other models we have none of). Our instructions (in /afs/inf.ed.ac.uk/group/mp-unit/firmware/dell/BMC-instructions) worked and came in useful when writing the ServerFirmwareUpdates page.
Notice of Critical Firmware/BIOS Updates
Alastair will look into the possibility of our getting notice of important firmware updates from Dell (and HP?).
Office Desktops to SL 6.3
Stephen intends to upgrade the office desktops to 6.3 this week after the stable release comes out, as this week's release includes a new kernel so will cause reboots anyway.
Moving the LCFG Master
Stephen has written a Plan for Moving the LCFG Master to a New Server in preparation for today's move from tobermory to schiff.
VMware Deprecation Progress
Most of the remaining VMware VMs are almost finished with and many could even be gone within a week. Chris will carry on tracking this.
Conserver KVM Support
The last bit of conserver support for KVM consoles is in this week's stable release. Once it's there Alastair will try it out on some of our test VMs.
SL 6.3 Upgrades on MPU Servers
We need to do them. Chris will allocate the upgrades around the group.
Useful Links
Could everyone please contribute their most useful tech news links to MPUusefulLinks?
Old Server Names
We're startng a rolling programme of eliminating old server names (mousa, trondra, split, etc.) from instructions in the wiki.
Multipath Instructions
We need to update the multipath instructions on the LCFG wiki FibreAttach page.
User Login Logs project
Stephen will flesh out this project's proposal in time for next week's development meeting.
Star Gazing Page
We will all update the MPUStarGazing page with ideas for what we should be looking at next.

This Week

  • Alastair
    • Add calls to kernel.rebuilddracut to lvm and multipath components - waiting on Matthew to accept and apply diff
    • Update initramfs for new kernel (and T5600 bug fix) Will now just filter through ...
    • Mail announcing openid service being killed off (and put on blog)
    • Ask Esteem/Dell if there is a mailing list/RSS feed for BIOS/firmware updatesIf have an Dell account, can subscribe to email. Not sure can subscribe for more than one device though.
    • Look at conserver after 31/10 - test migration, rebooting, interaction with kvmtool
      • short suspend/resume
      • long suspend/resume
      • install
      • reboot of KVM server (will need to use district for this)
      • poweroff and restart of guest
      • migrationBut src-host libvirt died - may have been because migrated from AMD -> Intel
    • Systems blog article on the KVM Service
    • do something about KVM server for end-users
    • Report libvirt empty LVM group issue to Redhat, unless fixed in 6.3
    • create an MPU KVM server header
    • Document ssh keys mgmt - home page and windows
    • Stargaze

  • Chris
    • Add BMC/Drac/NIC supported versions to goodfirmware map (not code detection yet)
    • Add Stephen's thoughts to server hardware final report and to wee projects.
    • modify mousa invedit
    • Document ssh keys mgmt - macos
    • Lead 6.3 upgrade
    • Stargaze

  • Stephen
    • upgrade LCFG master
    • Flesh out User Login logs project
    • Stargaze
    • Document ssh keys mgmt - linux
    • Security enhancements project
    • AFS Tutorial
    • Re-enable DR mirroring of lcfg-master
    • SL6.3 roll out to office desktops

-- AlastairScobie, ChrisCooke - 30 Oct 2012

Topic revision: r15 - 12 Nov 2012 - 17:10:16 - ChrisCooke
 
This site is powered by the TWiki collaboration platformCopyright © by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding TWiki? Send feedback
This Wiki uses Cookies