Re: Stopping file servers

[Ronald Oussoren <roussor@cs.vu.nl> writes:]

>Is there a way to cleanly stop a file server (that is stop the file server,
>and remove its name from the namer)?

Namer currently does not track server deaths. All VSTa servers to this
point are written so you can just kill them; I'd like to keep this semantic
as UNIX's shutdown/fsck/lost+found tedium has caused a lot of grief over the
years.

I've looked at making namer track registered server deaths; so far I haven't
been able to convince myself that a kernel mechanism is called for. As it
stands now you try to mount the server name and you get a failure. With
server death tracking you still get a failure. The only difference is that
an "ls" of the namer will show the entry. Is it worth more ring 0 code just
to cover a relatively minor exception condition?

                                                Andy

FromFrom vandys@glare.cisco.com Sat Nov 26 07:40:17 1994
Received: from hubbub.cisco.com (hubbub.cisco.com [198.92.30.32]) by amri.cisco.com (8.3/8.3) with ESMTP id HAA02164; Sat, vandys@glare.cisco.com Sat Nov 26 07:48:10 1994
Received: from glare.cisco.com (glare.cisco.com [171.69.3.78]) by amri.cisco.com (8.3/8.3) with ESMTP id HAA02167; Sat, 26 Nov 1994 07:48:09 -0800
Received: from localhost.cisco.com (localhost.cisco.com [127.0.0.1]) by glare.cisco.com (8.6.8+c/CISCO.SERVER.1.1) with SMTP id IAA17764; Sat, 26 Nov 1994 08:11:51 -0800
Message-Id: <199411261611.IAA17764@glare.cisco.com>
X-Authentication-Warning: glare.cisco.com: Host localhost.cisco.com didn't use HELO protocol
To: Robert Mayer - S 26 Nov 1994 07:40:16 -0800
Received: from dolly.par.univie.ac.at (dolly.par.univie.ac.at [131.130.70.42]) by hubbub.cisco.com (8.6.8+c/CISCO.GATE.1.1) with ESMTP id IAA08113 for <vsta@cisco.com>; Sat, 26 Nov 1994 08:03:56 -0800
Received: from doris.par.univie.ac.at (doris [131.130.70.22]) by dolly.par.univie.ac.at (8.6.9/8.6.9) with ESMTP id RAA05733 for <vsta@cisco.com>; Sat, 26 Nov 1994 17:03:51 +0100
From: Robert Mayer - Student <robert@par.univie.ac.at>
Received: (robert@localhost) by doris.par.univie.ac.at (8.6.9/8.6.9) tudent <robert@par.univie.ac.at>
Cc: vsta@cisco.com
Subject: Re: Stopping file servers
In-Reply-To: Your message of "Sat, 26 Nov 1994 17:03:47 +0100."
             <199411261603.RAA01327@doris.par.univie.ac.at>
Date: Sat, 26 Nov 1994 08:11:51 -0800
From: Andrew Valencia <vandys@cisco.com>

[Robert Mayer - Student <robert@par.univie.ac.at> writes:]

>> ... All VSTa servers to this
>> point are written so you can just kill them; I'd like to keep this semantic
>> as UNIX's shutdown/fsck/lost+found tedium has caused a lot of grief id RAA01327 for vsta@cisco.com; Sat, 26 Nov 1994 17:03:47 +0100
Date: Sat, 26 Nov 1994 17:03:47 +0100
Message-Id: <199411261603.RAA01327@doris.par.univie.ac.at>
To: vsta@cisco.com
Subject: Re: Stopping file servers

[Andrew Valencia <vandys@cisco.com> writes:]
> Namer currently does not track server deaths. All VSTa servers to this
> point are written so you can just kill them; I'd like to keep this semantic
> as UNIX's shutdown/fsck/lost+found tedium has caused a lot of grief over the
> years.

Does this mean that all seover the
>> years.
>Does this mean that all servers should be synchronous or am I misunderstanding
>this ?

There are several ways to deal with this. Write ordering, synchronous
operation, and database-style journalling all work. Actually, even sync-on-
last-close is sufficient; I'm not looking for Tandem-style guarantees here!

                                                Andy

rvers should be synchronous or am I misunderstanding
this ?

Regards,
Robert.
Received on Fri Nov 25 08:00:18 1994