[1537] in SIPB-AFS-requests
Re: Upgrades
daemon@ATHENA.MIT.EDU (Jonathon Weiss)
Tue Oct 18 19:45:08 1994
To: Matt Braun <mhbraun@MIT.EDU>
Cc: star-maintainers@MIT.EDU
In-Reply-To: Your message of "Thu, 13 Oct 1994 17:40:28 EDT."
<199410132140.RAA04036@medic.MIT.EDU>
Date: Tue, 18 Oct 1994 19:41:47 EDT
From: Jonathon Weiss <jweiss@MIT.EDU>
Sketchy meeting minutes:
We started at about 10:20 on 10/17/94
What is a reasonable migration schedule for the SS20's
ghudson: not predictable, we are a volunteer group
beacon: a few things left to do, there will be a bit of
work at the actual migration time: We will need
to rebuild the spool, so someone doesn't lose
rtfm: need to coordinate w/ news.answers to get
things tested and do the actual move
rtfm needs to move soon, p-lust probably won't support
WAIS even tho it is an SS20, too big a DB.
discussions of secure access to p-lust for *.answers
mumble Skey mumble mumble srvtab mumble mumble
remember 10% of the effort gets you 90% of the way there
to migrate bed we need among other things 2 freed gig
disks for the news spool (currently planning on
1 from current beacon and one from current rtfm)
What about moving the disks we have? Can we use the
filesystem we already have? Probably not, but we should
check.
Secondary services
(ie., those waiting for free maxines)
We want to make rosebud go away,
nothing else is really pressing
What's happening to DS3100
it's going back, we may be able
to keep it for testing, but not necessarily
it is on self-maint now (dcns-cluster supplies
parts when they die, this will last till the
summer?)
AFS disk
Should we use the internals?
(on the maxines)
how about we use them for local binaries instead
help, I've fallen and I can't get up
tlyu will look into this
ghudson doesn't trust the RZ25's
if serving them causes the disk to lose
it will cause an outage
decision: don't server the internal disks
Buying Disks
9G?
mhbraun: 1 spool, slow
yandros: ditto
yoav: according to Jeremy someone the 9G disks aren't
as reliable as the 2G or 4G ones
9G: also all eggs in one basket
9G: $3500 no contract 35c-40c/G
4G $2600 " " ~60c/G
AFS has a limit of 2G/partition
limit of 8partitions/device
jweiss: What about getting server hardware from IS
(we haven't gotten any in FY1995)
(FY '95 = 7/94-6/95)
It will take them time to get us the disks
Is the cell that full?
Yes, there are things that would be added now,
but for lack of space
mumble mumble what size disk to ask for
ghudson and mhbraun: will talk to Lorraine about
getting a 4G disk
What else will we need to upgrade in FY95
WWW is hosed
wait till FY96, demand will rise to meet resources
probably need to replace in FY 95
Does IS want us providing WWW services
(will they continue to give us hardware for it)
Yes. The feeling is that IS is handling the
official MIT page, but SIPB has more interesting
stuff for students, etc.
Waht about adding CPU's / RAM to the SS20's
Can we add CPU's under SunOS?
The mailhubs have
yandros: Ask IS for a fast Pentium for a server?
what OS?
Linux or NetBSD
ghudson: Pentium 90 for WWW would be interesting.
pentium would be a cheap server, but would have
to be experimental in some ways
SGI's???
not reasonable server platform right now
Whatever we get has to be able to do LOTS of net and disk I/O
So, what DO we want
definately something to replace or split the load with our
Web server.
maybe some disks for non-AFS (news, beacon, less likely rtfm)
What about another tape drive, there have been conflicts.
Maybe if it falls out of the sky
maybe we can buy it from our budget, if we get the disk from IS
What about FY96?
We sould upgrade the maxines
We can't plan that far ahead
Some other stuff I no doubt missed
