These were accepted.
Blog articles suggested:
Blog articles completed:
Report from Computing Executive Group
Verbal CCPAG report:
holyroodservice is to be replaced
Reports from units
The suggestion was to try it on a VM host first, and then
if that's OK to do servers generally next week.
huxley so would be a good one to try, and we can see if
it affects the loop-statistics graphs.
There was the question of VM clients' time drifting. This shouldn't be happening any more, as there's a cron-job on them all by default.
Graham noted that he also had kit on wire E. He's on the ticket now.
There was a suggestion that it might affect 790s too.
Should the lab machines be left to sleep or opted-out? Chris to consider.
There's a new kernel out now with specific KVM fixes. It'll be put on the KVM hosts, though not rolled out generally.
This still isn't understood. The recommedation is to do the updaterpms by hand first ("-- -f") and then reboot as soon as possible afterwards, though this does bring a very slight chance of unclean shutdowns. MP-unit will continue to investigate.
The relative costing of this as against just adding more licences is still being evaluated. We could still run our own licence servers, though, rather than being dependent on IS's.
Topics for discussion
"Lessons from the Forum power-down" was the only topic. The following points were made on the wiki prior to the meeting:
Points from the discussion at the meeting:
minvshow something? Yes (and the tool should really be renamed to something that more accurately reflects what it does!). MP-unit to see what's possible.
The next meeting will be on Wednesday 26th June at 10am, again in IF-2.33, chaired by George.
Please contact us with any comments or corrections.
Unless explicitly stated otherwise, all material is copyright The University of Edinburgh