[3355] in testers
Re: sgi 8.1.15: netscape
daemon@ATHENA.MIT.EDU (Ted McCabe)
Thu Nov 13 15:06:07 1997
Date: Thu, 13 Nov 1997 15:04:21 -0500
From: Ted McCabe <ted@MIT.EDU>
To: Matt Braun <matt@MIT.EDU>
Cc: Jonathon Weiss <jweiss@MIT.EDU>, Greg Hudson <ghudson@MIT.EDU>,
Craig Fields <cfields@MIT.EDU>, testers@MIT.EDU, ops@MIT.EDU
In-Reply-To: "[3353] in testers"
Well, we're talking about the same amount of disk space no matter what
we decide. Smaller, more numerous volumes gives us, ops, more
flexibility for placing the volumes.
As to the balancing question, we could just as easily turn around and
put them all on the same server. Point is, we'll have the choice.
One problem that I see as more likely to pop up due to multiple small
volumes is the occassional need to move RO's as servers and disks move
in and out of service. That's only a problem if it's not safe to
re-release the volume due to ongoing work. In my experience, I've yet
to find this to be the case before doing a release (though sometimes
it took some effort to be sure). So I wouldn't worry about this issue.
A second problem that perhaps Jonathon was thinking of is that it
becomes more likely that a problem will occur when the multiple
volumes are released. You can't release them all at once, so how do
you deal with the interval when the various RO's are out of sync?
Could that be a problem with multiple small volumes?
Other than this second problem, I'm all for the small volume option.
--Ted
