[646] in Release_7.7_team
New Hesiod Information for 8.0
daemon@ATHENA.MIT.EDU (Craig Fields)
Wed Jul 17 16:06:21 1996
Date: Wed, 17 Jul 1996 16:06:19 -0400
From: Craig Fields <cfields@MIT.EDU>
To: ops@MIT.EDU, release-team@MIT.EDU
Following is an explicit layout of how everything should eventually
look for clusters that provide 8.0, according to the plan I sent out a
couple of weeks ago. There are also additions from requirements I
hadn't thought of at the time but picked up from consulting with
people.
All of the filesystems listed below have been created if they didn't
already exist. The work that remains is updating cluster
information. Before I do that (to alpha, beta, and early clusters -
obviously I can't do it to public clusters until their respective
releases), I thought I'd send out the big picture so people can see if
I've forgotten anything.
---
Cluster: public-generic
syslib athena-generic-srvd-77
syslib athena-genericsys-80 8.0
athena-generic-srvd-77 AFS /afs/athena.mit.edu/system/@sys/srvd.77
athena-genericsys-80 MUL athena-generic-os-80 athena-generic-srvd-80
athena-generic-os-80 AFS /afs/athena.mit.edu/system/@sys/os
athena-generic-srvd-80 AFS /afs/athena.mit.edu/system/@sys/srvd.80
Clusters: alpha-sun4, beta-sun4
syslib public-sunsys
syslib dev-sun4sys-80 8.0
public-sunsys AFS /afs/athena.mit.edu/system/sun4m_53/srvd.77
dev-sun4sys-80 MUL dev-sun4-os-80 dev-sun4-srvd-80
dev-sun4-os-80 AFS /afs/dev.mit.edu/system/sun4m_54/os
dev-sun4-srvd-80 AFS /afs/dev.mit.edu/system/sun4m_54/srvd.80
Clusters: public-sun4, early-sun4
syslib public-sunsys
syslib athena-sun4sys-80 8.0
public-sunsys AFS /afs/athena.mit.edu/system/sun4m_53/srvd.77
athena-sun4sys-80 MUL athena-sun4-os-80 athena-sun4-srvd-80
athena-sun4-os-80 AFS /afs/athena.mit.edu/system/sun4m_54/os
athena-sun4-srvd-80 AFS /afs/athena.mit.edu/system/sun4m_54/srvd.80
Clusters: alpha-sgi, beta-sgi
syslib public-sgisys
syslib dev-sgisys-80 8.0
instlib dev-sgi-inst-80 8.0
public-sgisys AFS /afs/athena.mit.edu/system/sgi_52/srvd.77
dev-sgisys-80 MUL dev-sgi-os-80 dev-sgi-srvd-80
dev-sgi-os-80 AFS /afs/dev.mit.edu/system/sgi_53/os
dev-sgi-srvd-80 AFS /afs/dev.mit.edu/system/sgi_53/srvd.80
dev-sgi-inst-80 AFS /afs/dev.mit.edu/system/sgi_53/install
Clusters: public-sgi, early-sgi
syslib public-sgisys
syslib athena-sgisys-80 8.0
instlib athena-sgi-inst-80 8.0
public-sgisys AFS /afs/athena.mit.edu/system/sgi_52/srvd.77
athena-sgisys-80 MUL athena-sgi-os-80 athena-sgi-srvd-80
athena-sgi-os-80 AFS /afs/athena.mit.edu/system/sgi_53/os
athena-sgi-srvd-80 AFS /afs/athena.mit.edu/system/sgi_53/srvd.80
athena-sgi-inst-80 AFS /afs/athena.mit.edu/system/sgi_53/install*
Filesystem aliases:
athena-genericsys athena-genericsys-80*
athena-sun4sys athena-sun4sys-80
athena-sgisys athena-sgisys-80
dev-sun4sys dev-sun4sys-80
dev-sgisys dev-sgisys-80
---
* In Moira, not yet in Hesiod
---
A few things to note:
The filesystem aliases exist for user convenience only, and
should not be used in cluster information. Otherwise, machines
could involuntarily be pointed to new system packs when they
should not be.
syslib information pointing to 7.7 must not be tagged as 7.7.
This is because 7.7 machines that have not been updated to the
latest patch don't parse the tag in the currently correct way,
and so would lose track of their system packs entirely.
All syslib information other than 7.7 must be tagged
appropriately.
I have chosen to leave public-sunsys and public-sgisys as
they are and continue to use them for 7.7 pointers, leaving
well-enough alone.
I'm planning to redirect the 7.7 packs in the test clusters
to point to the Athena cell, in order to remove dependencies
on the 7.7 packs in the dev cell so we can decommission them.
Steps:
Update clusters {alpha,beta,early}-{sun4,sgi} to look
as described above, when convenient. Then eliminate clusters
alpha-sun24, beta-sun24, early-sun24, in favor of the updated
*-sun4 clusters. (The mechanics of this may a bit complicated
as these clusters are a bit messed up. But I plan to take it
carefully, one step at a time.)
Later, eliminate filesystems named {alpha,beta,early}-{sun,sgi}*.
Update clusters public-{generic,sun4,sgi} as those releases go to
field. (Updating generic with sun4 will do the right thing.)
I may think of snags once I actually sit down to do this stuff. Anyone
see anything I've missed, or any errors I've got in the Hesiod that's
out there now?
Thanks,
Craig