[2865] in Release_7.7_team
miscellaneous linux install news
daemon@ATHENA.MIT.EDU (andrew m. boardman)
Tue Jul 24 23:12:35 2001
Date: Tue, 24 Jul 2001 23:11:18 -0400
Message-Id: <200107250311.XAA30756@pothole.mit.edu>
From: "andrew m. boardman" <amb@MIT.EDU>
To: release-team@MIT.EDU
Athena upgrades will indeed fail if insufficient disk space exists to
take the update. (More specifically, rpmupdate will die without updating
anything at all. (Even more specifically, rpmRunTransactions will by
default check for space for everything on its transaction list before
doing anything.))
Everything in /mit/install/miniroot is current and can be used to build a
current miniroot, which I have in fact done today because AFS servers
have been moving a lot lately and there's a CellServDB hardcoded in the
miniroot.
Everything in /mit/install/installer is *not* current; I'll have an
installer which installs both 8.4 and 9.0 as soon as I go back to W92 and
look at whatever's on the console of my test machine so I can see what
stupid typo I made. (The linux install process in a nutshell: the
etherboot floppy fetches a miniroot image over the network, which then
boots and starts afs, from which the phase2 script in the installer
volume (/afs/athena/system/rhlinux/installer) actually formats the disk
and installs everything in the package list and so forth.) The design of
this makes it relatively easy to add more functionality to the phase2
script for dealing with SCSI, dual-boot, or whatever else, so long as
it's within the "custom install" logic of course.
