The current actions were not reviewed at this meeting.
The following items have been added to the actions list:
Report from Computing Executive Group
Alastair reports that he is extremely pleased with the professional looking Forum server area - please lets all try and keep it that way.
Frontline Support Issues
Reports from units.
Not had any response from IS yet as to why ed.ac.uk had outage at the same time (10-12 on the Thursday). Still not clear cause of problem, was it the AT work for instance. Going to start plugging back in links - at risk period may not be just immediately after as problems may take a while to appear (suggestion was to wait until after the AT work this weekend before doing this). Should have a discussion in the future on network topology and the trade off between a simpler less robust model (less redundancy for link/switch failure) which would at the least reduce the complexity such that more people could do fault finding. Should phones have a separate switch in the basement - would this actually help much. We should consider how to inform users - the human touch is important but it is also important to have some kind of information board where people come into the Forum (independent display screen or just a portable whiteboard).
Phones, wireless and VLANs (out of Forum) will all be out for the duration. However we should have a fallback for data traffic. There are now switches in FH that are logically part of the AT network and may be useful to help temporary re-location of machines from AT to FH before work commences in AT. Some discussion on how to maintain www.inf and AFS servers. Shouldn't spend too much effort on a generic solution as in the future top level at least of www.inf will be under Polpoly and IS management anyway. Alastair will announce the outage.
IS are also making this changeover. Simon raised the question of what a University PKI should look like.
Safe from at least 620s. We can move to swap files in the future as the performance is now as good and this will give us more flexibility.
Live headers should only be used for dynamic data and not static configuration, the quality of service will be better as a result.
It would be worth putting in the auto reboot mechanism for at least basic login/compute servers. Stephen will post an example. It would also be nice if the autoreboot component could use remctl to tell nagios that a reboot is going to happen and to suspend checks for 10mins after or so.
Course descriptors have been frozen as this information is now held in EUCLID.
All users should be on AFS by the end of 08/09 academic year. The Linux servers will provide better performance.
MacOS 1.5.6 does not work using Cosign to the Wiki or LCFG Wiki. Either a Cosign server configuration problem or Safari browser problem.
No-one to report. Comments from the minutes from the previous Ops meeting: Can we see a breakdown by Unit queue? Can we have graphs? Other comments to Alison please.
Topics for discussion
The next meeting will be on Wednesday 28th January in IF4.31 and chaired by Tim Colles.
Please contact us with any comments or corrections.
Unless explicitly stated otherwise, all material is copyright The University of Edinburgh