From: <stingay_at_email.protected>
Date: Wed, 9 Aug 2006 15:41:35 +1000 (EST)

Hi Jamie,

I think your idea is a good one. The serials numbers are a bit of a
nuisance to deal with as they are usually in fine print on the bottom of
the disk I think. They are also relatively long and complicated strings
that are not easy to digest at a glance. We have started placing large
labels on the disks, for example on the xraid disk handles e.g. ATNF V001
1 (for ATNF VLBI disk set 1, disk 1) or similar. These allow us to easily
identify the disk at the correlator when searching through piles of them
to find the data to correlate.

But I see no reason why this identifying label could not be incorporated
into a field of your database, mapping to the serial number, perhaps
along with other information.

During the last run, only one or two stations provided information on
which disks were used for which experiments. This slowed us down
considerably at the correlator. Things will only get worse when
(hopefully) we get funded for 200 TB worth of disks. We will need to be
very careful keeping track of them all.

I think the wiki is a good idea to track this information during the
observing session and I think a broader database on the web that also
tracks the location of the disks and their status (observed, experiments
contained on that disk, correlating pending/underway, processed, released
etc) would be a great idea. Craig has a system like this started that you
might want to look at.

I'd be happy for you to coordinate such a system for tracking disks for
the next session. By the next session, do you mean the August 23
observations, or the next full session in November?

Maybe others could give their opinions too.

Cheers, Steven

##=====================================================##
Dr Steven Tingay stingay_at_astro.<!--nospam-->swin.edu.au
Swinburne SKA Project Leader
Associate Professor

Centre for Astrophysics and Supercomputing
Swinburne University of Technology
Mail No H39
P.O. Box 218, Hawthorn, Vic. 3122, Australia

ph: +61 3 9214 8758
fax: +61 3 9214 8797

http://astronomy.swin.edu.au/ska
##=====================================================##

On Wed, 9 Aug 2006, Chris Phillips wrote:

>Message from Jamie:
>
>---------------------------------
>
>
>From: Jamie Stevens <Jamie.Stevens_at_utas.<!--nospam-->edu.au>
>To: VLBI observers <vlbiobs_at_atnf.<!--nospam-->csiro.au>
>
>Hi all,
>
>Last teleconference it was noted that we need a way of knowing which
>experiments are recorded to which disks. Jim suggested that we use the
>disk serial numbers that can be obtained from hdparm, and after some
>looking around on the web, it appears this is the only really standard
>way of identifying disks in software, ie. you can't label a disk with a
>more user-friendly moniker, because that label is only attached to the
>file system, not the disk.
>
>I suggest that before the next run, when the observatories get their
>disks, that we note all the serial numbers and label each disk/caddy
>with its serial number. We then need some way of logging this
>information. The wiki would seem the sensible option - we would need a
>table of disk serial # vs other disk ID information, ie.
>Hobart internal RAID disk 1, serial # KRFS02RAH4D4LC
>
>Logging experiment vs disk would then need to be done at each
>observatory. I have written a little program which takes the disk path
>and outputs the serial numbers of all the disks in that RAID set in
>order. I've already made my server use this program for logging the
>serials, and disko could be easily altered to do so as well if need be.
>
>I'm willing to coordinate the effort to get a system in place for
>logging before next run, however I'd like to hear from people whether
>this scheme is fine, or if it's too complicated, either for observers or
>for correlators.
>
>cheers
>Jamie
>
>--
>-----------------------------------------
>Jamie Stevens Postdoctoral Researcher
>School of Physics, University of Tasmania
>Room 461 Ph: (03) 6226 7529
>Jamie.Stevens_at_utas.<!--nospam-->edu.au
>-----------------------------------------
>
>
Received on 2006-08-09 15:41:58