MPU Meeting Thursday 28th November 2013

Inventory

Alastair has been rewriting the work flow document. This and other details can also be found on the inventory project's home page.

Virtual DICE

The main thing Chris got from the student feedback was that people were getting confused between Informatics' network and that of the rest of the university. It would simplify things for the students if we could make it possible for them to login to Virtual DICE using their DICE credentials and get their AFS home directory as their home directory, with their computer connected to anywhere on the university network. This is currently possible just from the Informatics network. The current stumbling block is that we don't export our LDAP information beyond the Informatics firewall - at least not without authentication, which isn't possible at the login stage. Following discussion at the operational meeting it was agreed that Toby would try logging in from Virtual DICE in this fashion while using a test "log everything" LDAP server, to see exactly what information would need to be exposed to the rest of EdLAN.

Client Refactoring

The v3 client is now in the develop release. Toby found a bug (now fixed) in Secure Mode which was exposed when using the v3 client. A related bug (227) fixed the LogServer's incorrect assumption that an LCFG node name would always be the same as the DNS name. The v3 client works in installs, although you can't yet install using the new generic profile support.

Misc Devel

  • Chris added a resource to the sleep component to make it optionally retain the kernel wake alarm when shutting down. This will help with physically inaccessible machines which should always be up, in this case the meeting room DICE machines inside the speakers' podiums. Users of these machines sometimes shut them down instead of logging out. The machines will now turn themselves on again after being shut down.
  • Stephen has pretty much finished rewriting the audit reports in python. He's been able to add whitelisting support, which should make the reports from the NX servers far more usable. He needs to document the API and code, how to use it, and how to make a report.
  • There was a request for a newer version of Unison on the ssh servers. This led Stephen to try the "alternatives" component. He quickly added extra features to it which he needed, mainly the ability to add new alternatives which have not already been registered.
  • Following the power failure debriefing Stephen has changed how the critical shutdown script initiates a shutdown. Python has a wonderful multiprocessing module which lets you set up a pool of worker processes, give them a task to do and a list of things to do it to, then it's all sorted out for you. All you need to do then is keep reading result reports until the job's finished.
  • Stephen wrote a script to cull orphan processes on the NX servers. This helps somewhat but the utmp data is not always right: it sometimes reports processes which in reality no longer exist.
  • GVFS no longer tries to mount on top of AFS.

Operational

  • To make space for the KB-based RO copies of the RPM repositories, Stephen deleted the SL5 and EPEL5 site mirrors.
    • We need to archive all the SL5 packages.
    • This removal prompted a change to the PkgSearch config to remove SL5 searches.
    • Stephen will make a list of the changes required when decommissioning an OS.
  • DIY DICE has gone to SL 6.4.
  • Chris downloaded firmware updates for Dell R710 and R720 servers.
    • Chris noted that it no longer seems sufficient to concentrate only on "Urgent" firmware updates from Dell: "Recommended" ones can be necessary to fix serious problems. In future he shall check both categories when looking for new updates.
  • Alastair has a new Drupal version 7.2.4 with security fixes. He'll apply it to computing.help on Sunday night. He aims to show other people how to upgrade a Drupal site to a new version but in this case it's not possible to schedule that.
  • Chris had problems printing from computing.help. Alastair pointed out the little printer icon at the bottom of each page, which applies a print-friendly style sheet - a great improvement. Alastair will look into having this style sheet applied automatically when a page is printed (RT:64798).
  • Chris made an inf level test host.
  • Alastair will make next week's stable release.
  • Stephen hopes to reinstall bakerloo with de-RAIDed disks (and thus with far bigger /tmp and /var/nx partitions, like central) at the end of the semester.

This Week

  • Alastair
    • Start Inventory project diary
    • Inventory project
      • Submit bug/enh to App::Cmd author wrt option to die on unspecified options
      • Pester George about location API
      • Publish revised schema (and document)
      • Continue fleshing out processes
    • Order a spare 600GB disk for waterloo (hot spare)
    • Ask George - what does the TXretransmit value mean for switch connections? (Can't see this using rfe -xl )
    • Consider how to make metropolitan usable by users
      • Decide whether we do this now or not
      • ISOs
      • minimal docs (mostly manual)
      • they'll use virt-manager, but not create machines or change config
    • circulate table of LCFG bugs
    • Consider activities list
    • Ask Ian D to present his dhclient proposals to the next LCFG deployers meeting - schedule for January (or next meeting) Scheduled for Feb meeting
    • Look at gnome issue with NX - try inftest account
    • Tidy up RT tickets
    • Updaterpms - are the timeouts configurable - can we reduce for machines on Inf machines - nope, CPP constants. LCFG bug #699
    • Look at more acceptable pool allocation for kvmtool create (people are too lazy to specify a pool, and we have to live with that!) - could kvmcreate just put in the pool with the most space, if a pool isn't specified non trivial. client side currently has no knowledge of which pools on KVM server exist - would need to add support for this so that the client could query details of each pool and choose the emptiest. LCFG bug #700
    • LCFG annual day - timescales for RHEL7

  • Chris
    • Chase Nigel wrt Central
    • Book GIT tutorial
    • Work on GIT project

  • Stephen
    • LCFG client refactor
      • report
    • Consider dhcpd component changes
    • Consider activities list
    • Start project LCFG client - remove obsolete module dependencies - project 274
    • Produce wiki page listing what needs changing when archiving an OS platform
    • LCFG annual day

  • Carol

-- AlastairScobie - 28 Nov 2013

Topic revision: r9 - 23 Sep 2019 - 13:33:37 - AlastairScobie
 
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