On 3/14/12 7:43 AM, G.W. Haywood wrote:
Hello again,

On Wed, 14 Mar 2012, Forrest Aldrich wrote:

What's happening is the clamav installation (make install) creates a
file *.tmp and removes it. This is why the process failed because I
mount the directory read-only on most of the systems to prevent
corruption. This is easily resolved by simply using another NFS mount
from where I keep distribution src.
...
I think you misunderstand. I'm keeping the source tree ...

No, I understand that you're keeping the source tree, I just don't
understand why you are keeping it if you're so worried that it might
become corrupted.  Why not just delete it when the build is finished?
Then it's most unlikely to get corrupted. :)

LOL no, I'm concerned about /usr/local becoming corrupted - the binary NFS mount that we want to use on our production systems. The problem I ran into is doing a "make install" from a read-only NFS mount broke the installation as it needs to create (and remove) a *.tmp file. I had built it from another system onto that export mount.



There you go making life difficult for yourself again.  Why not set up
your own ClamAV database mirror?
I'm not sure how to do this; however, we have only about 4 or 5 machines that poll for virus updates. And the mirror would be private (not publicly accessible).


In our case, it's all the same revision/OS (RHEL 5.x) -- we have both
32- and 64-bit to contend with, therefore 2 separate builds ...

I understand that.  What I don't understand is why you don't just build
your own RPM and have your mail filtering machines install that instead
of doing weird and wonderful things with NFS.
I think creating custom RPMs for this is overkill for what I need.


What's this 'corruption' that you're so afraid of?  If you want some
files to be immutable, then on Solaris look at
Corruption meaning I don't want local users (even privileged) to write to the NFS mount, we want to keep the binaries (and their configs that need to be there) in a consistent state.


I don't understand that.  All you're doing is building in a single
point of failure.  NFS sucks anyway, but when it goes down all your
mail filtering will go down with it.
Believe me, I understand that. What we really need (for all our video and other stuff) is a SAN. But try to convince my boss' boss about that one! ;-) I gave up. I just work with what I have :-)

_______________________________________________
Help us build a comprehensive ClamAV guide: visit http://wiki.clamav.net
http://www.clamav.net/support/ml

Reply via email to