Operational Meeting

Ian Durkacz

Accepted.

• Stephen has completed his action to disseminate autoreboot details.
• Alisons Smartboard action is ongoing - got software and user to test.

The following items have been added to the actions list:

• Iain to move beowulf profiles to normal LCFG servers and decommission illustrious.
• MPU to change root password for DIY DICE machines.

Report from Computing Executive Group

Frontline Support Issues

1. Help desks

   The CSO will be part of the unit for this time (as a secondment).

Reports from units.

• Infrastructure.

  3. Appleton Tower decant

     Router discovery is broken for machines with more than one interface on the same wire. Need to use an OSPF header instead (which is only supported on the Forum server wire).

     The ATA Erase command should be sufficient to wipe disks.

     Reminder - all "live" kit to be out by Friday 6th February at latest. May be space in Forum basement to store junk kit, which can be left in AT for now but we will need to move out before the contractors start.

  5. EdLAN-BP connection

     NFS and AuthSMTP services will be affected. Some compute servers as well. Also the ICCS web server. Alison will collate and make announcement.

  6. Linux router crashes

     SL5.3 will fix this.

• Managed Platform.

  1. Linux filesystem ACLs

     Some commands need to be explicity told to preserve ACLs (eg. rsync, tar). ACLs are not portable across platforms.

  2. Default mount options

     Being done to improve security by preventing some common attacks.

  3. Default disk sizes

     Re-partitioning will be necessary in the future (probably on upgrade to SL6) or users could opt-out of having additional research and teaching packages on their own machine.

  • PLEASE REBOOT REMAINING PUBLIC FACING MACHINES TO PICK UP NEW KERNEL ASAP

• Research and Teaching.

  1. Bioinformatics Spam Hack

     Can use Apache config to turn off PHP allow_url_fopen option, this does not have to be done in php.ini. We should change DIY DICE machines to use a non-default root password.

• Services.

  2. Failed disk news

     May take opportunity to reboot this weekend with BP work.

  • Staffmail - Alison speaking to Keith Nicol to formalise the support arrangements
  • Staff/Student SSH login problems at the weekend - home directory issues.

• User Support.

  1. RT Statistics Report

     Pictures please.

  • We should be doing post-event notification, perhaps a systems blog like ECDF have, even when things are fixed quickly. Anyone can (and should) edit the support page alert message if they notice something broken (particularly out of hours). It would be possible to aggregate/collate Nagios information onto the support page. This could be helpful, although we would need to have proper service names rather than hostnames.

Topics for discussion

None.

AOCB

None.

The next meeting will be on Wednesday 11th February in IF4.31 and chaired by Craig Strachan.