(r5) MPunitMeeting20150818 < DICE

DICE Web>ManagedPlatformUnit>MPUnitMeetings>MPunitMeeting20150818 (revision 5) (raw view)~~EditAttach~~
---+ MPU Meeting Tuesday 18th August 2015

---++ Inventory

Chris has reviewed the parser code, he found a few issues with the regular expressions which have been fixed. He also commented on the notes from the meeting with Graham. 

Alastair has been working on a replacement for the clientreport script. The aim is to modernise how the data is collected and support secure upload. He has been looking at using the [[https://metacpan.org/pod/Parse::DMIDecode][Parse::DMIDecode]] Perl module to query the data from dmidecode. This makes the coding much simpler, particularly when handling items which appear multiple times (e.g. RAM).

Alastair has also been looking at how to gather information on the monitors attached to a machine. For graphics cards with drivers in the standard kernel there are edid files in the =/sys= filesystem which can be used. This should allow us to get the information for all monitors attached to a system. This doesn't work for all graphics cards, in particular the nvidia closed-source driver does not expose any useful information about the monitor, it also doesn't give us useful data in the Xorg log file.

---++ LCFG Client Refactoring

Nothing happened but hopefully this will be revived soon.

---++ Miscellaneous Development

   $ pkgsearch : Chris has been looking at pkgsearch, it appears that the code for fetching the yum repodata will need to be completely rewritten. Hopefully the rest will not need such radical surgery.

   $ diydice : Chris has found install problems with diydice, at the least the necessary installbase profiles are missing for SL7.1.

   $ bugzilla : The version of bugzilla we are currently using will reach end-of-life some time in November so we need to upgrade. The newer versions available are 4.4 and 5.0. We don't expect that everything we need for the SL7 server platform will be ready in time so it looks like we will need to upgrade the service on SL6.

   $ kvmtool : The new kvmtool has shipped with support for INI-style configuration files.

   $ SL6 initscripts : Stephen has updated the SL6 initscripts package and reworked the contents of the associated lcfg-upstarthooks to match. Mostly this involved importing some missing files and removing the old =rcX= files. There is now a single =rc= file which does most of the work.

   $ LCFG kernel component : The =initramfstrigger= script has been imported into the kernel component package as it is generally useful. On SL7 the initramfs file will now be rebuilt when either systemd or dracut changes. On SL6 it will be rebuilt if the initscripts, lcfg-upstarthooks or dracut packages change.

   $ fail2ban : Stephen fixed [[https://bugs.lcfg.org/show_bug.cgi?id=705][bug#705]] so that the fail2ban component will work on SL7. He also corrected the dependencies on SL7 to resolve [[https://bugs.lcfg.org/show_bug.cgi?id=887][bug#887]]

   $ xinetd : Stephen fixed [[https://bugs.lcfg.org/show_bug.cgi?id=723][bug#723]] so that the xinetd component will work on SL7. 

   $ rkhunter : Stephen updated the rkhunter packages on SL6 and SL7 to fix [[https://bugs.lcfg.org/show_bug.cgi?id=891][bug#891]]. The template in the component needed some tweaking for the latest version.

   $ ipv6 on SL6 : It turns out that the way we had disabled ipv6 support on SL6 by blacklisting the kernel module was not the correct approach. Both bridging and bonding need the ipv6 module to be loaded (and possibly others). The correct approach is to set the ipv6 module options to =disable_ipv6=1=. Hopefully this change will not make java slow again, it's worth noting that we have not had any complaints about slowness with java on the compute servers where ipv6 was never disabled due to the need for network bonding.

---++ Operational

   $ _oyster_ : We need to reschedule the reboot of the KVM server _oyster_ for SL6.6

   $ _waterloo_ : The suspend space on the KVM server _waterloo_ has been increased.

   $ _azul_ : The KVM server _azul_ has been moved to one of the "fibre" racks and an FC card has been added. To avoid wasting spare ports on the FC switch it has not been cabled.

   $ _district_ : The old test server _district_ has been junked.

   $ Other reboots : Reboots for SL6.6 have been done for _piccadilly_ (nx.inf), _bruegel_ (package master) and _brendel_ (staff.ssh). The other ssh server _schiff_ will be done on Thursday morning.

   $ HP 800 G1 : The recent batch of HP 800 G1 desktops have a problem which means that reboots will sometimes fail. Alastair investigated and the solution appears to be to disable the BIOS "fastboot" option. See [[https://rt4.inf.ed.ac.uk/Ticket/Display.html?id=72904][RT#72904]] for details.

   $ Activities List : The various lists of activities and "wee projects" have been reviewed and merged. There is now a single MPUActivitiesList page.

   $ IBM storage array : We will be switching to a 3rd party maintenance contract for the IBM storage array. Before the current IBM contract expires we should download all available firmware and software, including the version of the management application for Windows.

   $ pkgforge builders : There is a new pkgforge builder, named _badger_, this is a VM on _amarela_ at KB. It is just building packages for SL6 and SL6_64. Hopefully this will reduce the load on the other two builders. The current two are getting quite old and are due to be replaced soon.

   $ bash for SL7.1 : The local version of bash on SL7.1 needs to be updated.

   $ mate window manager : mate on SL7.1 has been updated to version 1.10.

   $ KDE screensaver : The PAM configuration for the KDE screensaver has been fixed so that it is now possible to unlock.

   $ _telford_ : The old AFS package volumes have been deleted and the SAN volumes unmounted. The machine is still running and needs to be junked.

---++ This Week

   * Alastair
      * Inventory project
         * continue working through InvProjectWorkFlow
         * finish deploying new order file processing code
         * consider what next can be integrated into existing system, if anything
         * Consider sublocation, macaddr discovery tool
         * Blog articles
            * edid - DRM / nvidia
            * <strike>json stuff - HTTP problem</strike>
         * Continue with clientreport
            * look at JSON -> postgresql
      * @home -  look at using rsync from site.pkgs instead of mirroring from upstream
      * Remove default pool if ops meeting agrees
      * Schedule firmware upgrade for DS3254
      * Add instructions on adding extra disks to an LVM/KVM pool - to KVM host guide
      * Talk to Phil Wadler about machine upgrade - ask for a list of software he requires built 
      * Experiment with different window managers under VNC (making the assumption that performance under NX will be similar)
      * Schedule reboot of oyster with new memory install - Expand suspend space-  needs a reboot to pick up /dev/sda4 - move /var/libvirt before KVM clients come up  (need shutdown rather than suspend) -<strike>await new memory</strike> Check using eth0 and eth2. Create plan of action and circulate.
      * Think of a use for 'atom'
      * Understand how NetworkManager works wrt init scripts
      * Flesh out Base SL7 server project
      * DS3524 - look at work for fibrechannel switch switchover
         * <strike>download windows and mac versions of client software (just in case linux version doesn't work under SL6)</strike>

   
   * Chris
      * Inventory project
         * continue working through InvProjectWorkFlow
      * pkgsearch for SL7
      * <strike>Add improved NX culling script to activities list</strike>
      * Flesh out Base SL7 server project
      * Continue work on VirtualDICE for SL7
      * Groundwork for bugzilla upgrade
      * Reboot <strike>vermela and</strike> hammersmith 
      * Liaise with George over iDRAC documentation
      * <strike>Fix up vermelha and amarela pool autostart</strike>


   * Stephen
      * LCFG client refactor stage 1 
         * schedule debrief meeting
      * LCFG client refactor stage 2
         * finish off C code
      * Think about PD - Interested in [[http://zeromq.org/][ZeroMQ]]
      * Finish off window manager selector code
      * <strike>Reboot  schiff</strike>
      * Flesh out Base SL7 server project
      * Junk telford


-- Main.AlastairScobie - 18 Aug 2015
Topic revision: r5 - 25 Aug 2015 - 14:10:33 - AlastairScobie
DICE
This is WebLeftBar
Copyright © 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