.43b/.44 problems (fwd)

[Oliver Xymoron <oxymoron@waste.org>]

I'm having serious troubles with Zmailer this evening. I had been running
.43b with a couple patches for most of December, until earlier this
evening a message on one of our lists (~1000 remote subscribers) seems to
have caused router to explode.

Since I could find no sign of what the trouble was, beyond that it ate
enough system resources to prevent console logins, and I had the archive
for 2.99.44 handy, I did a clean install, including the patch to
functions.c from the 28th. All seemed well, until I tried to expand one of
my majordomo outgoing aliases and discovered this:

Usage: listexpand [ -e error-address ] [ -E errors-to-address ] [-p
privilege] [ -c comment ] [ -N notarystring ] $attribute $localpart
$origaddr < /file/path

which suggested the following fix:

--- aliases.cf~ Sun Jan  5 21:40:36 1997
+++ aliases.cf  Sun Jan  5 22:44:50 1997
@@ -122,8 +122,11 @@
                priv=$(get $attr privilege)
                [ -f "\1" ] &&
                 priv=$(getpriv "644" $priv "\1" include) &&
-                 nattr=($newattribute $attr privilege $priv) &&
+                 nattr=$(newattribute $attr privilege $priv) &&
                   defer=''
                return $(listexpand -E "$user-owner@$mydomain" -p $priv \
                                    -e root -c "\1 file expansion"      \
                                    $nattr "\1" "$user" < "\1")

This fixed, I proceeded to try and send a message to a test list, which
resulted in these lines at the end of router's log:

<Pine.LNX.3.93.970105231611.308F-100000@death.cyberoptics.com>: address:
test-outgoing
s_apply() ran out of svstack[]! (svstop=5)

And router promptly disappears without leaving a core(?!). My experimental
fix:

--- router/shliaise.c~ Sun Jan  5 21:11:06 1997
+++ router/shliaise.c  Mon Jan  6 00:02:32 1997
@@ -11,7 +11,7 @@
 struct conscell **return_valuep = 0;
 struct conscell *s_value = 0;
 int svstop = -1;
-#define SVSTOPMAX 5 /* Is this too low ? */
+#define SVSTOPMAX 20 /* Is this too low ? */
 struct conscell *svstack[SVSTOPMAX] = { 0 };

 extern int l_apply __((u_char *fname, struct conscell *l));

...and it's still running after 20 minutes and a couple mailing list
messages, even though I have no idea what the above module is for.

Anyone seen similar problems with .43b or .44?

--
 "Love the dolphins," she advised him. "Write by W.A.S.T.E.."