Alastair mentioned USB ID dongles, which might be useful. He'll circulate some details.
There was considerable discussion of this. Two opposing viewpoints emerged: that the errors were legitimate; and that long-term errors made the tactical overview page rather less useful. It was suggested that it would be nice if acked entries didn't appear on this page; Ian didn't think this would be straightforward though.
Tim thought that checksum unloading was probably there for reasons which no longer apply. He'll discuss the particular failure case with Alastair and take it from there.
Toby's changes have been reviewed in gerrit. Craig to revisit the password portal.
Actions added or revived:
Neil to check the current situation
syncactually means on KVM clients
Blog articles suggested:
Blog articles discussed:
Report from Computing Executive Group
Alastair gave a verbal report, including inter alia:
Reports from units
This prompted the question: what does
sync actually mean on
a VM client, due to the caching done by the host. MPU will think about
this, and maybe tweak some settings and/or issue some advice.
Might it make sense for the KVM hosts to be "on" a private subnet rather than one of the site server subnets?
It is useful for hosts to offer several pools, rather than amalgamating, as they have different usage characteristics. It might make sense for the names to be a bit more descriptive, however.
Topics for discussion
dmesgis useful for users, as it can let them see why their code has failed. It's also useful to attackers, as it can give away system details. The question is how to strike a balance.
dmesgfrom multi-user machines, but to leave it for desktops.
Matlab upgrade - Graham
The next meeting will be on 24th June 2015, 10am in 4.31. George will convene.
Please contact us with any comments or corrections.
Unless explicitly stated otherwise, all material is copyright The University of Edinburgh