[5236] in testers
Different wrong behviors when trying to update to 9.1 when you're
daemon@ATHENA.MIT.EDU (Bill Cattey)
Tue Jun 18 18:59:17 2002
From: Bill Cattey <wdc@MIT.EDU>
To: testers@mit.edu
Content-Type: text/plain
Content-Transfer-Encoding: 7bit
Date: 18 Jun 2002 18:59:14 -0400
Message-Id: <1024441154.17156.75.camel@tokata.mit.edu>
Mime-Version: 1.0
Tom Thornton, on his Linux box, gets the message that he has no cluster
info. This is incorrect: He has cluster info, but it is not valid for
the current release. He instead should have gotten the message that he
does not have appropriate cluster information to take the release he's
trying to take. The cause of this badness is probably cuz it tests for
the null string when getting version-specific cluster info.
The test needs to be enhanced to confirm there is any cluster info at
all, and then the error message needs to be enhanced so that it says,
"You are trying to take an update for a release inapproprate to the
cluster your system is regitered to."
----
Gary Dryfoos, on his Sun box with public, not early, cluster membership,
said update_ws 9.1 instead of update_ws.
The update_ws code dutifully allowed itself to be forced-updated to 9.1,
and then it dutifully attached the 9.0 system packs appropriate for its
cluster, and began behaving VERY strangely.
I think it's unreasonable to expect people to just know that if you do
the linux thing "update_ws 9.1" on the Sun by mistake, terrible flaming
red rocks will fall on your head.
I think update_ws should be enhanced on the Sun side to guard against
this situation.
-wdc
