No news yet on the date for this work.
As trailed at the last Operational Meeting, all Forum edge switches have now been flashed with the latest firmware. We intend to let the Forum power-down (whenever that happens ...) take care of the necessary reboots.
One of the Forum core switches (
core2, a 5412zl) running the latest K.14.60
firmware crashed a couple of weeks ago. We've filed a report with HP, and await
comment from them.
Thanks to those who traced connections in order to document the current usage of fibre channel; https://wiki.inf.ed.ac.uk/DICE/ForumFCSwitchPopulation summarises the current situation.
The only obvious anomalies are that:
metropolitancurrently has both of its fibre ports connected to the same fabric,
fc1*. This isn't optimal: for best multipathing redundancy, one port should be connected to
fc0*, and the other to
fc1*. Can the owners of
metropolitanrearrange this sometime please?
bpbeastcurrently has only one of its two fibre ports connected to any switch. Is this the intention?
In the Forum, we currently we have 8 free ports on the
fc0* fabric, and 5 free ports on the
fc1* fabric. When one of
metropolitan's two connections is moved from
fc0* these numbers will change to 7 and 6 respectively.
This is the best we can do, given that machines fitted with single fibre ports only (namely:
stockholm) are currently
evenly arranged amongst the two fabrics (probably, just by good fortune).
Some questions arising are:
Can everybody using fibre connections please consider these questions so that we can try to make definite plans for future use.
The first stage of Prometheus deployment is scheduled for week beginning July 19th. More details in notes from 13/7 meeting with US unit.
Please contact us with any comments or corrections.
Unless explicitly stated otherwise, all material is copyright The University of Edinburgh