dyer@arktouros.mit.edu (Steve Dyer) (08/21/90)
I really had to put my 2 cents in about the format of the latest
AIX PS/2 1.2 update, 1005. It comprises 19 disks in AIX backup/restore
format. There are 14 or 15 LPPs which can be updated. If your
AIX system (as many do) consists of more than a few of these LPP
(read: all of 'em), you end up inserting up to 19 disks up to
15 times. At the very least, the LPPs which reside on the
later diskettes require you to insert and have scanned the
previous diskettes. Each of the LPP updates is considered
separately, which is why the whole operation approaches O(N^2).
It is now 3:30PM, and I have been swapping diskettes since
11AM this morning. Not only is the procedure long, it requires
constant supervision, since a disk swap has to occur approximately
every three minutes. What a waste of time!
Could someone at IBM Palo Alto or Austin please give me
a charge number so I can bill them for the hours I've wasted.
(No :-)).
---
Steve Dyer
dyer@ursa-major.spdcc.com aka ...!{harvard,linus,ima,m2c,rayssd}!spdcc!dyer
dyer@arktouros.mit.edu, dyer@hstbme.mit.edubuck@siswat.UUCP (A. Lester Buck) (08/21/90)
In article <1990Aug20.194756.16980@athena.mit.edu>, dyer@arktouros.mit.edu (Steve Dyer) writes: < I really had to put my 2 cents in about the format of the latest < AIX PS/2 1.2 update, 1005. It comprises 19 disks in AIX backup/restore < format. There are 14 or 15 LPPs which can be updated. If your < AIX system (as many do) consists of more than a few of these LPP < (read: all of 'em), you end up inserting up to 19 disks up to < 15 times. At the very least, the LPPs which reside on the < later diskettes require you to insert and have scanned the < previous diskettes. Each of the LPP updates is considered < separately, which is why the whole operation approaches O(N^2). < < It is now 3:30PM, and I have been swapping diskettes since < 11AM this morning. Not only is the procedure long, it requires < constant supervision, since a disk swap has to occur approximately < every three minutes. What a waste of time! < < --- < Steve Dyer < dyer@ursa-major.spdcc.com aka ...!{harvard,linus,ima,m2c,rayssd}!spdcc!dyer < dyer@arktouros.mit.edu, dyer@hstbme.mit.edu [Rumages around in his comp.unix.aix archives and finds...] <From: robin@sabre.austin.ibm.com (Robin D. Wilson/1000000) <Newsgroups: comp.unix.aix <Subject: Re: AIX 1.2 Diskette Overload <Message-ID: <2726@awdprime.UUCP> <Date: 11 Jul 90 14:36:23 GMT <References: <1331@msa3b.UUCP> <Sender: news@awdprime.UUCP <Reply-To: robin@reed.UUCP (Robin D. Wilson/1000000) <Organization: IBM AWD, Austin, TX <Lines: 52 < <In article <1331@msa3b.UUCP> kevin@msa3b.UUCP (Kevin P. Kleinfelter) writes: <>AIX PS/2 V1.2 is even more diskettes than V1.1! <>TOTAL ..........85 <> <>Now don't get me wrong -- I LIKE having all these things. <>It just seems a little too much to expect me to shuffle (or even just to <>store) 85 diskettes. Oh yes, speaking of shuffling, don't forget that <>the applying the update diskettes means: <> insert disk 1 <> insert disk 1, insert disk 2 <> insert disk 1, insert disk 2, insert disk 3 <> ... <> insert disk 1, 2, 3, 4, 5, 6, 7, 8 <> <>I will recommend to management that AIX NOT be purchased for additional <>PS/2 machines, because it COSTS TOO MUCH (personnel cost) to INSTALL!!! <>(Yes, TCF can help ... IF all your machines are close enough to have on <>a LAN; not the case for us.) < <The files on the update can be restored to a sufficiently large filesystem <and then "re - backed - up" to any format you wish (ie. a very large file <on disk, or a streaming tape). Be sure to use the "restore -x" and the <"backup -ivf <file>" commands (and not tar/dd/cpio). Once on another format <the backup can be distributed to another machine in this format (or across <the network as a mounted file). Also, just a suggestion.. try making one <system with the base of operation that you want.. ie. BOS, Xwindows, BNU <or whatever. Then back that system up to a more palletable media like <tapes. Distribute the tapes, and leave the diskettes as a backup, or for <installing machine specific products (assuming all machines don't have <all the products). BACKING UP ONE SYSTEM, AND RESTORING IT TO OTHERS <IS NOT RECOMMENDED FOR THE UPDATES. The updates should be applied using <the updatep command. (But updatep WILL work with the previously described <"re - backed - up" distribution methods. < <These suggestions both work quite well on the RT version of AIX (v.2.2.1). <But my office mate is a PS/2 support person and guarantees that they work <on PS/2's as well... But your mileage may vary based on road conditions. < < <Also... on the RT AIX there is a command called "/etc/bffcreate". This <command will allow you to reformat (to a different media) the updates. <You might check and see if it is available on the PS/2. < < <+-----------------------------------------------------------------------------+ <|The views expressed herein, are the sole responsibility of the typist at hand| <+-----------------------------------------------------------------------------+ <|UUCP: cs.utexas.edu!ibmchs!auschs!sabre.austin.ibm.com!robin | <|USNail: 701 Canyon Bend Dr. | <| Pflugerville, TX 78660 | <| Home: (512)251-6889 Work: (512)823-4526 | <+-----------------------------------------------------------------------------+ -- A. Lester Buck buck@siswat.lonestar.org ...!uunet!nuchat!siswat!buck
igb@cunixa.cc.columbia.edu (Isidore G Bendrihem) (08/21/90)
If you have lots of disk space, you can restore the disks to a single temp file. This will save you lots of time and disk swapping. This is the procedure: 1. Restore the diskettes to an empty directory in a file system that has enough space to hold all the disks: mkdir update cd update restore -x 2. Create a single backup file, let's say /tmp/mpp1005, and delete the backup direcory "update": find . -hidden -print | backup -i -vqf /tmp/mpp1005 cd .. rm -r update 3. Apply the updates (answer any relevant questions and let it run) cd / updatep -ac /tmp/mpp1005 rm /tmp/mpp1005 Again, make sure you have enough disk space to hold the backup directory and the backup (updatep) file that you create. If you don't, this procedure is not for you. Isidore Bendrihem VLSI Laboratory Columbia University
dwight@altair.la.locus.com (Dwight Tovey) (08/22/90)
Two comments about this:
1. When you make the directory to restore the update into, make sure it is
on a replicated file-system (if you are using TCF). Otherwise you will
lose the fstore values for the files.
2. The updatep command should be:
updatep -ac -d /tmp/<filename>
^^
/dwight
Dwight Tovey
dwight@locus.com
The views I express are entirely my own and do not in any way represent
those of Locus Computing.