Returned mai03

[RICH ROGERS]

--- Received from DIST.ROGERR  217-2584             97-06-17 14.12
  -> IN=ssa-managers@Eng.Auburn.EDU
returned mail




On Sunday, June 15, 1997 11:44 AM, Doug Hughes
[SMTP:Doug.Hughes@Eng.Auburn.EDU] wrote:
> VM2.3 -
>
> We had a disk failure last night. And it sent out email about the failure
> and recovery of all affected volumes, which worked flawlessly. The
interesting
> thing was that the relocation kind of spread data to the wind because there
> was so little space left on any of the disks. Anyway, no data is lost, so
> I can't complain (much ;), but it sure is going to be fun to reconstruct
> that disk once we put it back in.. Yeah, we don't 'have' to put it back
> just the way it was, but we like to keep things balanced, so we will anyway.
> (Easier for maintenance, better disk access balance)
>
> For those interested in a picture of what it did see
> http://www.eng.auburn.edu/users/doug/ssa.gif
>
> It's about 32K.
>
> Notice, huey09 is missing (The failed disk)
>
> u2, the monster volume, got stuck on the end of huey07, huey10, huey12,
> huey03, huey02 (which you can barely see since it's only 5mb) and huey17
>
> The other interesting one is ce_h2. In fairness, it had already
> been grown down to the stripe at the bottom of the screen
> (huey19-24 disks), the stuff from huey09 got put on the end (little
> thin strips you can barely see) of huey22, huey22, huey23, huey21, huey24
> and in the middle of huey13 where a blank space existed. There is also
> a little strip at the end of huey01.
>
> I must commend its relentless seeking of space... ;)

Thanks... that's exactly the idea.

I'm going to suggest that, in some future release, we consider:
- a flag to prohibit fragmentation
- a flag to not use fragments below a certain size
- auto-generating a script or command to move data back