Stephen intends updating the list and posting out after the meeting. (Summary: "There are 58 SL6 machines currently needing a kernel update." Full list in Stephen's email.)
Alastair has done part of this, but there's also input needed from Alison.
Amend action to RAT Unit. Delete "ASAP".
Actions added or revived:
These had been brought up to date as of the meeting. Updating will now be an ongoing process.
It's on the MPU list
Blog articles suggested:
Blog articles completed:
Report from Computing Executive Group
Reports from Units
Alastair has observed that the
quite often doesn't start
on SL7. The user-visible symptoms are that the first DNS lookup by an
application will incur a timeout, after which the resolver will fail over
to the next one in the list; and that will be cached so that any subsequent
lookups will go straight through.
He and George have been trying to debug this. It seems to be timing-related, and it looks as though something is sending a SIGTERM to the daemon to tell it to shut down. Things still to try include trying to extract more information from process accounting and auditing, and having systemd start it directly. We'll let you know...
Stephen described the way in which this will now work (see also the Research and Teaching report). There's a small change required in the computing.help page to take account of this.
This should now be added to <develop> so that it can be tested more widely.
Network manager had been getting in the way, so had been being told to shut down after the initial interface configuration. Unfortunately this had interacted badly with systemd, so it's now proposed to leave it running but configured not to do anything to interfaces we don't want it to meddle with.
There's a workaround for a bug on SL7 which should hit this week's <stable>.
Topics for discussion
Iain asks: "We don't currently ship wine on sl7 and we only have 64 bit rpms available from EPEL which won't run 32 bit windows apps. Should we roll our own wine or push people towards virtualbox/campus agreement?"
The conclusion was that it would be a lot of effort and not worth it. We shouldn't install it at all, as having it only partially working would be worse in practice. We should direct people towards a VM solution instead. The computing.help pages should be amended accordingly.
This was a followon from the item in the CEG report, which was taken here but minuted above.
We should start thinking about arrangements for this year's Christmas Lunch (or whatever). Suggestions welcome.
The next meeting will be on 14th October 2015 at 10:00 in Wilkie G.14. Alison will convene.
Please contact us with any comments or corrections.
Unless explicitly stated otherwise, all material is copyright The University of Edinburgh