[Greylist-users] perl errors on non-numeric $rowids > 0

[Ken Raeburn]

I just got back from a trip, fixed a mailer that was offline, finally
whitelisted my newest MX server, and ran its queue of delayed mail.
Now, in the (mixed sendmail + relaydelay) logs on the receiving end, I
see a lot of these perl errors reported:

Nov 14 20:19:15 kal-el relaydelay:   Whitelisted Relay.  Skipping checks and passing the mail.
Nov 14 20:19:15 kal-el sm-mta[24752]: hAF0shSC024752: from=<bounce-debian-user=raeburn=xxxx.raeburn.org at lists.debian.org>, size=5451, class=-30, nrcpts=2, msgid=<y4tq_.A.MFC.agCs_ at murphy>, proto=ESMTP, daemon=MTA, relay=splat.raeburn.org [140.239.227.39]
Nov 14 20:19:15 kal-el sm-mta[27855]: hAF0shSA024752: to=raeburn at xxxx-internal.raeburn.org,nobody at xxxx-internal.raeburn.org, delay=00:00:01, xdelay=00:00:01, mailer=esmtp, pri=149008, relay=xxxx-internal.raeburn.org. [10.0.0.6], dsn=2.0.0, stat=Sent (hAF1JFN19957 Message accepted for delivery)
Nov 14 20:19:15 kal-el relaydelay:   IN EOM CALLBACK - PrivData: 1544,1544
Nov 14 20:19:15 kal-el relaydelay: Argument "1544,1544" isn't numeric in numeric gt (>) at /usr/local/sbin/relaydelay.pl line 432.
Nov 14 20:19:15 kal-el relaydelay:   * Mail successfully processed.  Incremented passed count on rowid 1544.
Nov 14 20:19:15 kal-el relaydelay:   * Mail successfully processed.  Incremented passed count on rowid 1544.
Nov 14 20:19:15 kal-el sm-mta[27857]: hAF0shSC024752: to=/dev/null, ctladdr=<nobody at xxx.raeburn.org> (8/0), delay=00:00:00, xdelay=00:00:00, mailer=*file*, pri=149610, dsn=2.0.0, stat=Sent
Nov 14 20:19:15 kal-el relaydelay: 

(Actual recipient addresses have been obscured.)

These messages are coming in with two recipient addresses each, and
record id 1544 in the database is the newly added whitelist entry for
the downstream server those messages are coming in from.

I'm using relaydelay 0.04 modified with the VERP-handling patch I sent
a while back.  (Actually, a minor variation of it.)  Line 432 is here,
in eom_callback:

    # Issue a temporary failure for this message.  Connection may or may not continue
    #   with delivering other mails.
    return SMFIS_TEMPFAIL;
  }

  # Only if we have some rowids, do we update the count of passed messages
  if ($rowids > 0) {
^^^^^^^^^^^^^^^^^^^^
    # split up the rowids and update each in turn
    my @rowids = split(",", $rowids);
    foreach my $rowid (@rowids) {
      $dbh->do("UPDATE relaytofrom SET passed_count = passed_count + 1 WHERE id = $rowid") or goto DB_FAILURE;
      print "  * Mail successfully processed.  Incremented passed count on rowid $rowid.\n" if ($verbose);

Actually, there are a few places where "$rowids > 0" is checked, but
the rowids field of the private data appears to be a comma-separated
list of row ids.  I expect this code should probably be checking for
an empty string, instead, or a string that doesn't match "^0+$", or
something like that.

It doesn't appear to be hurting anything; the counts do get updated,
so at least in these error cases, it looks like the correct branch is
being taken.

Ken

--
We are Grey.  We stand between the Spam and the Light.