RPM Community Forums

Mailing List Message of <rpm-devel>

Re: [PATCH]: logging of package install/remove to syslog

From: Jeff Johnson <n3npq@mac.com>
Date: Sat 23 Feb 2008 - 04:28:09 CET
Message-Id: <E1F8EB95-03F0-4A0F-944E-4A940B6E7BA3@mac.com>

On Feb 22, 2008, at 11:20 AM, devzero2000 wrote:

> yum, on top of rpm, already have this capability. I found this very  
> useful , as i use a central syslog. So it
> is possible to query the central syslog for patch management  
> information ex-post (e.g, if all the system have the same version  
> applied).
>

So use yum. No one is stopping you. What is your point?

It's equally possible to query an rpmdb, which is as centrally located
as local syslog files usually are, with the added benefit that all,  
not just NVRA and
a time stamp, information regarding the installation state of packages  
is available.

And if you're serious about package management on distributed
clients then the contents of the package, not just the "version", need
to be verified. There's too many side effects, with multilib and selinux
in particular, for the NVRA & time stamp to be useful for much.

I'm not saying that there isn't use in logging, but I see little reason
(other as a historical archive of erasures, which --repackage already
does, arguably at perhaps too great a cost) for duplicating information
in Yet Another /var/log/rubbish file.

No matter what, if one wants rpm logging, the implementation needs to  
be done
more flexibly (by permitting choice of format and content and not only  
syslog
as output transport), and done in rpmdb/rpmdb.c, not in lib/psm.c,  
because
there is already an implementation there.

73 de Jeff

> But, it is only an opinion.
>
> On Thu, Feb 21, 2008 at 10:48 PM, Jeff Johnson <n3npq@mac.com> wrote:
>
> On Feb 21, 2008, at 3:44 PM, Per Øyvind Karlsen wrote:
>
> > Here's an improved and more upstream friendly version of syslog
> > patch from
> > Mandriva which adds logging of install/remove of packages to syslog.
> >
> > It will be disabled by default, but can be activated by
> > %_log_to_syslog macro.
> >
> > Any objections of commiting or ideas of improvement?
>
> Yes.
>
> What needs doing is to add a "append-only" mode to log to file,
> and a "use-syslog" flag to accomodate those that wish,
> that delivers the results of a headerSprintf() queryformat to a log
> file.
>
> Follow the (esentially) logging that already exists to populate /var/
> cache/hrmib
> for use by the net-snmp HR-MIB in rpmdbAdd/rpmdbRemove to see
> where the configurable logging needs to be done.
>
> And imho, logging +foo/-bar is largely pointless for anything other  
> than
> filling up a disk with rubbish. There's too little useful information
> with
> NVRA to be useful for debugging, and rpm already has a more powerful
> database with a query mechanism for storing the useful information,  
> and
> a more powerful verification mechanism to verify that the data is
> indeed,
> reliable.
>
>  A historical, erase-only log might be useful, but that's not what
> your patch
> is aiming to do.
>
> 73 de  
> Jeff______________________________________________________________________
> RPM Package Manager                                    http://rpm5.org
> Developer Communication List                        rpm-devel@rpm5.org
>
Received on Sat Feb 23 04:28:30 2008
Driven by Jeff Johnson and the RPM project team.
Hosted by OpenPKG and Ralf S. Engelschall.
Powered by FreeBSD and OpenPKG.