Return-Path: linux-nfs-owner@vger.kernel.org Received: from mx1.redhat.com ([209.132.183.28]:6646 "EHLO mx1.redhat.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1755873AbaIWQBB (ORCPT ); Tue, 23 Sep 2014 12:01:01 -0400 Date: Tue, 23 Sep 2014 12:00:54 -0400 From: Simo Sorce To: "J. Bruce Fields" Cc: NeilBrown , Steve Dickson , Linux NFS Mailing list Subject: Re: [PATCH 1/2] nfs-service: Added the starting of gssproxy Message-ID: <20140923120054.7dc8764a@willson.usersys.redhat.com> In-Reply-To: <20140923152000.GC29932@fieldses.org> References: <20140922152603.75005941@willson.usersys.redhat.com> <54207BCD.70101@RedHat.com> <20140922204401.GI26763@fieldses.org> <5420911D.6080506@RedHat.com> <20140922223423.GA29932@fieldses.org> <5420B78D.6040704@RedHat.com> <20140922202655.5e308e58@willson.usersys.redhat.com> <20140923015549.GB32712@fieldses.org> <20140923120804.51dbcc2e@notabene.brown> <20140923084854.6c67d401@willson.usersys.redhat.com> <20140923152000.GC29932@fieldses.org> MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Sender: linux-nfs-owner@vger.kernel.org List-ID: On Tue, 23 Sep 2014 11:20:00 -0400 "J. Bruce Fields" wrote: > On Tue, Sep 23, 2014 at 08:48:54AM -0400, Simo Sorce wrote: > > On Tue, 23 Sep 2014 12:08:04 +1000 > > NeilBrown wrote: > > > I don't think you want an install section. That means the service > > > has to be explicitly enabled, which is a pain. > > > I think nfs-server.service should Want= this. > > > I also think > > > > > > ConditionPathExists=/etc/krb5.keytab > > > > > > would be appropriate. > > > > If GSS-Proxy is in use the administrator may choose to use a keytab > > in a different location, so I am not entirely sure we should depend > > on /etc/krb5.keytab, however it is also ok to decide that if the > > admin wants to use a different place that they create a custom unit > > file. Up to you. > > Note we're already using the same line in rpc-gssd.service and > rpc-svcgssd.service. > > Can you suggest a better "does this host have krb5 configured?" test? > > I think false positives are OK, but not false negatives. > > (So, if we run those daemons unnecessarily it may annoy some people, > but if we fail to run them when they're needed then things really > don't work.) I would simply not test for presence of a keytab if it were my call. If the admin decided to start nfs-secure I assume he already got the proper key material, ie I am not so sure that double-checking the admin in the unit files is right for gssproxy, because gssproxy has directives that allow the admin to put the keytab elsewhere. Simo. -- Simo Sorce * Red Hat, Inc * New York