These were accepted.
Report from Computing Executive Group
Reports from units.
Note that the Infrastructure Unit propose making the default outlet state for the Forum power bars be OFF. You have until the next Operational Meeting (10th February) to make sure that the outlets for all your servers are properly configured.
It was noted that we might turn the classifier back on at some point.
As mentioned at the meeting, Craig sent out an e-mail with an explanation of the recent problems. A more detailed explanation for those who are interested will be posted on blob.inf shortly.
We discussed whether all file servers should be ext2. It was agreed that this would cause problems should the server go down as the subsequent fsck would take hours. The possibilities of moving to ext4 and replacing raid5 with raid10 were also mentioned along with removing journalling from all rw volumes.
Topics for discussion
The question was whether there is any reason not to run om file configure nightly, to repair (for example) badly behaved RPMs ? It was felt that this could just 'patch over' bad packages and potentially mask updaterpm problems. So, it was suggested that the MPU unit should give this some further consideration.
AOCBIt was noted that the ex-mars machines have a very small disk and one larger disk. The root partition is currently on the small disk. Support will reconfigure one of them by using the larger disk for the root partition.
The date and time of the next operational meeting will be announced shortly.
Please contact us with any comments or corrections.
Unless explicitly stated otherwise, all material is copyright The University of Edinburgh