On Sat, 14 Feb 2009 20:57:52 -0800
Dennis Peterson <denni...@inetnw.com> wrote:

> Steve wrote:
> > On Sat, 14 Feb 2009 16:50:44 -0800
> > Dennis Peterson <denni...@inetnw.com> wrote:
> > 
> >> Steve wrote:
> >>> On Sat, 14 Feb 2009 23:21:16 +0100
> >>> aCaB <aca...@digitalfuture.it> wrote:
> >>>
> >>>> Steve wrote:
> >>>>> Unfortunately, no change.
> >>>> That's likely because you didn't update the svn checkout or recompiled,
> >>>> or reinstalled, or restarted the daemons.
> >>>> _______________________________________________
> >>>> Help us build a comprehensive ClamAV guide: visit http://wiki.clamav.net
> >>>> http://www.clamav.net/support/ml
> >>> Did you not check the version number in  the clamd log, or the timestamps?
> >>>
> >> Are all vestiges of previous versions of ClamAV gone? Specifically, 
> >> libraries. 
> >> What do you get when running ldd against the ClamAV binaries? I suggest 
> >> this 
> >> only to eliminate a common and recurring problem with ClamAV installations 
> >> and 
> >> that is left-overs from earlier versions.
> >>
> >> dp
> >> _______________________________________________
> >> Help us build a comprehensive ClamAV guide: visit http://wiki.clamav.net
> >> http://www.clamav.net/support/ml
> > 
> > I shut everything down, ran the uninstall for 0.94.2, then the install from 
> > svn, with no change ):
> 
> 
> Ok -- looks good so far. But... One thing I forgot to mention in the earlier 
> note is to never ever trust the uninstall tool nor the rpm tool dejur to 
> actually completely uninstall anything. They can fail with mysterious results.
I've had no problems with the uninstall/install method when building clamav 
from source so far... and debian doesn't use rpm (:
> 
> The other issue is any tests you do with ldd can be account-sensitive. Some 
> accounts for example may have LD_LIBRARY_PATH defined, others not. Some 
> systems 
> admins set that as a global, some don't. Some systems (Solaris, for example) 
> have global library paths set using crle, others use ldconfig. It's a crazy 
> world. Then there are the hard-coded path dependencies built into the build 
> process of specific applications. You absolutely cannot depend on version 
> x.xx.x 
> to uninstall version x.xx, so if you no longer have the earlier version 
> source 
> to do the uninstall you should expect to manually review the debris left 
> behind. 
> This is especially true of rpm's that come from different sources - the 
> builders 
> don't connect with each other to ensure one builder's package is compatible 
> in 
> any way with that of another builder.
I am the sysadm, installs/startups/tests are all run as root. I never use 
LD_LIBRARY_PATH unless absolutely necessary, it's too much of a security 
liability. This is all running on a 32 bit debian stable VPS. As I said before, 
I uninstalled using 0.94.2 and installed the current subversion install. I can 
find no fault with this, the developers of clamav have been exemplary in this.

All of this is built from source, I have never, ever mentioned rpms.
> 
> What this means is don't trust anything, scan your environment to see if 
> there 
> are legacy bits laying about and get rid of them. You may not find them but 
> if 
> you do you certainly would not be the first.
Look, I'm a systems administrator, so I'm paid to be a pessimist (: 
> 
> dp
> _______________________________________________
> Help us build a comprehensive ClamAV guide: visit http://wiki.clamav.net
> http://www.clamav.net/support/ml
My main frustration is that the only way I can get more information from the 
applications is to rewrite the
 code itself... at least it's written in a real language (runs for cover!). but 
it would be great to be able to change the log level to get more detailed info 
out. Then I would be able to take a more proactive approach in debugging this 
problem.

Cheers,

Steve
-- 
Steve <st...@greengecko.co.nz>
_______________________________________________
Help us build a comprehensive ClamAV guide: visit http://wiki.clamav.net
http://www.clamav.net/support/ml

Reply via email to