[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Freeipmi-devel] minor bug/inconsistencies in bmc-config output
From: |
Albert Chu |
Subject: |
Re: [Freeipmi-devel] minor bug/inconsistencies in bmc-config output |
Date: |
Tue, 22 Jan 2013 16:49:14 -0800 |
Hey Dan,
For the FreeIPMI 1.2.5 release I fixed three things.
1) Fix the --filename option to work with --checkout (broke in a release
long ago).
2) Clarify --filename option in config tool manpages (bmc-config,
ipmi-pef-config, etc.).
3) On non-fatal errors, config tools exit with 1. On fatal errors now
exit with 2.
Admittedly I was a bit wary of doing #3 in a minor errata release. But
given how I know most users use FreeIPMI, I don't think it'll be a big
deal. I hope I'm right :P
If you'd like to try it out, the SVN branch is this one:
svn co
svn://svn.savannah.gnu.org/freeipmi/branches/Release-1_2_0_branch
./autogen.sh
./configure
Al
On Tue, 2013-01-22 at 12:16 -0800, dan farmer wrote:
> Al, thanks for the quick response.
>
> I got another off-list response pointing out my ancient version of freeipmi
> (yum doesn't apparently upgrade much!), but I just checked the latest (1.24)
> and my findings were repeated.
>
> I'm not sure what the right answer is here, but I'll describe a potential use
> case; I'm investigating taking the output and run it through some basic
> security/sanity checks (e.g. is cipher 0 enabled.) I suppose I could use the
> -L option to query for sections, and then iterate the tool on each section.
> Doing that is somewhat interesting:
>
> # for i in `bmc-config -L`; do
> echo -n $i :
> bmc-config --always-prefix -v --section=$i --checkout > /dev/null
> echo $?
> done
>
> User1 :1
> User2 :1
> User3 :1
> User4 :1
> User5 :1
> User6 :1
> User7 :1
> User8 :1
> User9 :1
> User10 :1
> Lan_Channel :0
> Lan_Conf :0
> Lan_Conf_Auth :0
> Lan_Conf_Security_Keys :0
> Lan_Conf_Misc :0
> Rmcpplus_Conf_Privilege :1
> SOL_Conf :0
>
> Your suggestion of having a 1 or a 2 based on fatal or not would be quite
> helpful to understand if there are any serious problems or not.
>
> And yes, I can obviously workaround the --prefix business by redirecting the
> output into a file ;)
>
> dan
>
> ^..^
>
>
>
>
> On Jan 22, 2013, at 10:11 AM, Albert Chu <address@hidden> wrote:
>
> > Hey Dan,
> >
> > On Mon, 2013-01-21 at 12:59 -0800, dan farmer wrote:
> >> First, thanks for the tool, I'd been doing individual grabs of config
> >> values before, bleah!
> >>
> >>
> >> Secondly, bmc-config returns 1 even if it succeeds pulling
> >> configuration details (my sys-details at the bottom.) I suppose it's
> >> possible that it returns non-zero when it gets an "Unable to checkout"
> >> on a field, but that isn't even listed in the non -v output and is a
> >> fraction of the output. I hope this is a bug, or there is an
> >> alternate way to get an exit code, or else it makes running the tool
> >> in an automated fashion really sucky (how can you know what worked?)
> >
> > I'll go back and look at this code some more. The exit code is
> > generally considered to be non-zero if there is any checkout failure
> > (fatal or non-fatal). However, that should probably be tweaked, b/c of
> > how bmc-config has grown, it's probably impossible for all config
> > checkouts to always succeed on any hardware out there. The exit codes
> > should atleast be different (maybe 1 on fatal error, 2 on non-fatal
> > error).
> >
> >> Thirdly, a small bug, or at least a consistency issue, with the
> >> --filename output, which ignores the --always-prefix argument (the
> >> first command prepends the host, the 2nd doesn't); the manpage says
> >> under --always-prefix that the tool will "always prefix output" - is
> >> this a problem with my system (details below), the tool, the man page,
> >> or my understanding, or some combo of the same? :)
> >
> > This is probably a documentation mistake. B/c the output of bmc-config
> > is supposed to be fed back into bmc-config, it's one of the FreeIPMI
> > tools that shouldn't support the always-prefix option. I'll have to
> > look into this more closely.
> >
> > But thanks for the bug report. Somewhere along the line --filename
> > stopped working w/ --checkout. So I need to fix that in the upstream
> > version (it's funny that no one noticed until now. Most probably pipe
> > --checkout stdout to a file).
> >
> > Al
> >
> >> (Three runs of bmc-config, the first to show a valid exit code, then
> >> the two with --filename and simple file redirect, then a diff to show
> >> the lack of prefix)
> >>
> >>
> >> # bmc-config --version
> >> bmc-config - 0.7.16
> >> Copyright (C) 2003-2008 FreeIPMI Core Team
> >> This program is free software; you may redistribute it under the terms
> >> of
> >> the GNU General Public License. This program has absolutely no
> >> warranty.
> >> address@hidden zen]# echo $?
> >> 0
> >>
> >>
> >> # bmc-config --checkout --always-prefix > foo1
> >>
> >>
> >> # echo $?
> >> 1
> >>
> >>
> >> # bmc-config --checkout --always-prefix --filename foo2
> >>
> >>
> >> # echo $?
> >> 1
> >>
> >>
> >>
> >>
> >> # diff foo1 foo2
> >> 1,476c1,476
> >> < localhost: #
> >> < localhost: # Section UserX Comments
> >> < localhost: #
> >> < localhost: # In the following User sections, users should
> >> configure usernames, passwords,
> >> [...]
> >>
> >>
> >> System details it was tested on:
> >>
> >>
> >> # bmc-config -V
> >> bmc-config - 0.7.16
> >> Copyright (C) 2003-2008 FreeIPMI Core Team
> >> This program is free software; you may redistribute it under
> >> the terms of
> >> the GNU General Public License. This program has absolutely
> >> no warranty.
> >>
> >>
> >> # uname -a
> >> Linux blackhol3 3.5.4-3.el6xen.x86_64 #1 SMP Sat Sep 29
> >> 15:16:26 EST 2012 x86_64 x86_64 x86_64 GNU/Linux
> >>
> >>
> >> # cat /etc/redhat-release
> >> CentOS release 6.3 (Final)
> >>
> >>
> >> # bash --version
> >> GNU bash, version 4.1.2(1)-release (x86_64-redhat-linux-gnu)
> >> Copyright (C) 2009 Free Software Foundation, Inc.
> >> License GPLv3+: GNU GPL version 3 or later
> >> <http://gnu.org/licenses/gpl.html>
> >>
> >>
> >> This is free software; you are free to change and redistribute
> >> it.
> >> There is NO WARRANTY, to the extent permitted by law.
> >>
> >>
> >> Happy to answer any q's, etc., and thanks again.
> >>
> >>
> >> dan
> >>
> >> ^..^
> >>
> >>
> >>
> >>
> >>
> >>
> >>
> >> _______________________________________________
> >> Freeipmi-devel mailing list
> >> address@hidden
> >> https://lists.gnu.org/mailman/listinfo/freeipmi-devel
> > --
> > Albert Chu
> > address@hidden
> > Computer Scientist
> > High Performance Systems Division
> > Lawrence Livermore National Laboratory
> >
>
--
Albert Chu
address@hidden
Computer Scientist
High Performance Systems Division
Lawrence Livermore National Laboratory