[clamav-users] clamav-milter not being built for 0.102.0

G.W. Haywood clamav at jubileegroup.co.uk
Mon Oct 7 10:45:57 EDT 2019

Hi Micah,

On Mon, 7 Oct 2019, Micah Snyder (micasnyd) wrote:
> On 10/6/19, 11:39 AM, G.W. Haywood wrote:
> > On Wed, 2 Oct 2019, Micah Snyder (micasnyd) via clamav-users wrote:
> > >
> > > We will identify the exact issue and include a fix for it in a
> > > 0.102.1 patch release along with one or two other bug fixes.
> >
> > If you can squeeze in a check somewhere that the temporary directory
> > exists and is writeable by clamd that might save some grief.
> Can you be more specific?  I just got up, drinking my first cup of
> tea (ran out of coffee), but issue isn't ringing any bells with me
> at the moment.

I came across this at the weekend, while I was moving a clamd instance
between machines.  Preparation for the move included telling my milter
that it would use TCP instead of a Unix domain socket, and while clamd
was running on the same host as the MTA using everything was
peachy.  But after copying the binary to a different host, it all went
belly-up.  A Wireshark trace showed that after the command (the milter
uses INSTREAM) and the data to be scanned were sent OK, instead of the
expected reply the clamd instance would close the connection and issue
a RST with no explanation.  As it turned out clamd could not write the
temporary files which it has to write for the scan.  There was nothing
in the logs, even with debug set for both clamd and for LibClamAV.  So
I chowned the relevant directory, problem solved.  It took me too long
and if I hadn't also run out of tea I might have spotted it quicker. :(



More information about the clamav-users mailing list