[3996] in testers
Re: nsd caching failure?
daemon@ATHENA.MIT.EDU (Robert A Basch)
Wed May 5 17:50:49 1999
Message-Id: <199905052144.VAA25360@boom.mit.edu>
To: testers@MIT.EDU
In-Reply-To: Your message of "Tue, 04 May 1999 18:01:47 EDT."
<199905042201.SAA26813@kiko.mit.edu>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Date: Wed, 05 May 1999 17:44:40 -0400
From: Robert A Basch <rbasch@MIT.EDU>
I looked into this a bit more; it is indeed an nsd caching problem
(gee, what a surprise). There appears to be a race condition when
passwd is modified twice or more in succession; in an rlogin, the
user entry is added temporarily, removed, then added again. I have
verified that nsd caching can lose in this sequence; also, if an
entry is added then removed immediately, nsd can report the entry
as present.
Fwiw, sleeping for one second after updating passwd seems to avoid
these caching problems.
We have sent another plea for relief to sgi.
Bob
