Operational Meeting

These were accepted as a reasonable account of the last meeting.

Actions discussed:

• Look into time sync problem on VMs

  Script has been created and will be in the stable release w/c 27th March. There have been 16 reported so far with the largest time difference being 5 seconds. Kerberos will cope with a difference up to 2 minutes. More responses expected once in stable release.

• Speak to Jon about request for web address

  Neil spoke to Jon and he's keen to have a chat. He would like bayes.ed for publicity reasons but but a redirect would be possible. Also might require some 'bayes' email addresses.

• Review pages in computing.help

  All pages with review date prior to August 2015 should be reviewed by the end of April.

• Look at logging all Apache logs centrally

  blog.inf is now firing extra logging and it seems to be working.

• Removal of tape library from KB

  All 500 tapes will be handed over to the new owner but must be wiped first. Neil's estimate is that it will take 2 weeks. We'll see .....

• MacOS Sierra

  Ongoing

Actions completed:

• Create project for 'sorting' roles
• Document Auristor client for self-managed Windows machines

Blog articles completed:

• Mail relay changes
• Domain-squatters
• Small NAS boxes
• User account creation and expiration

Blog articles suggested:

• None

Report from Computing Executive Group

The sickness absence reporting policy was discussed with no objections.

Reports from Units

• Infrastructure

  1. Forum server room cooling
  2. DHCP summaries and warnings

     These are now being sent to all COs so please look at and fix any problems caused by your machines!

  3. Appleton Tower
  4. Forum UPSes
  5. Small-UPS batteries
  6. COtest wire

     Agreed that this is no longer required.

  7. Network ports
  8. DNS reverse delegations
  9. dns-blackhole-uk-squatters.h
  10. RIPE anchor
  11. SL7 upgrades

      We discussed whether multicast DNS was still required. It was suggested that it would only be useful on desktops and perhaps required for labs. It could, however, be removed from network servers.

  12. IPv4 routing

• Managed Platform

  • nagios headers
  • HP EliteDesk G2
  • SL7 initramfs
  • Attempt to access beyond end of device....

    Alastair stressed that this is bogus when seen at boot time but if seen at any other time, it does indicate a problem.

  • Recent meetings

• Research and Teaching

  1. cluster complaints
  2. hardware failures

• Services

  1. IS Relay on RBL
  2. SL7 progress

• User Support

  • exam dates

    US's wiki page is here.

  • SL7 server upgrades

Topics for discussion

• Implementation of remote serial consoles

  Ian noted that both Lantronix and IPMI consoles have historically been fairly reliable - so what, if anything, has recently made IPMI less reliable?

  • Infrastructure use IPMI consoles and have not seen major problems.
  • Alastair asked colleagues at FLOSS - no-one had seen similar problems with IPMI consoles. Others use ipmitool directly, rather than via conserver. Ian noted that our problems persist when a direct connection via ipmitool is used - this has been tested.
  • Ian noted that the Lantronix service is still supported - and should be used as desired. We could revert to Lantronix en masse, but there would be a cost involved since the machines are old and will need to be replaced in due course. (One at AT has recently failed.).
  • The 'crazy console' problem has been seen on a directly attached screen on SL7. How is IPMI implicated in that case?
  • Stephen suggested that KVM servers should be on Lantronix console servers.
  • Once in a 'strange' state, a power cycle (not just reboot) fixes the problem
  • waterloo's BMC has failed in firmware/hardware, and needs to be replaced.
  • It would be useful if some automatic checking could be done to report on the correct functioning of consoles.
  The conclusion reached was that any console problem should be emailed to MPU and also mentioned in chat. KVM servers should be reverted to Lantronix boxes. Inf Unit will investigate console 'liveness' checking.
• Carol's visit to TCU

  The set up at TCU sounded very similar to IS with little/no integration between groups. One point which we might want to consider is their policy of allocating a member of the support team to new staff members for their first week to help answer their IT queries and give a general introduction to the services provided.

AOCB

• AT furniture etc

  Neil is awaiting a response from David Bell regarding furniture and will keep us posted. As far as we know, things are still on schedule but there is a 'return to AT' meeting next week where Martin will update on progress.

• TDM

  Craig would like suggestions for next week's TDM. Failing that, we can all get on with SL7 upgrades.

The next meeting will be on Wednesday 12th April at 10:00, in IF-4.31, chaired by Craig.