Installation on Irix: Installs Fine, Behaves Badly

I've just gotten the latest mailman installed on my irix system (an O2). I set up the test list and it, cosmetically, seemed to be fine; I added myself as a subscriber, poked around at the web interface, and so on. Seemed to like it.
Then I tried posting a message to the list. Nothing. Again. Nothing. Checked syslog, and found this spew came up whenever I tried to send a message;
Dec 20 10:37:42 6C:iago sendmail[18714]: KAA18714: from=iago, size=5, class=0, pri=30005, nrcpts=1, msgid=199912201037.KAA18714@iago.net, relay=iago@localhost Dec 20 10:37:42 6C:iago sendmail[18714]: KAB18714: clone KAA18714, owner=testlist-admin Dec 20 10:37:42 3C:iago Mailman mail-wrapper: Failure to exec script. WANTED gid 2147430212, GOT gid 0. (Reconfigure to take 0?) Dec 20 10:37:42 6C:iago sendmail[18716]: KAB18714: to="|/local/mailman/mail/wrapper post testlist", delay=00:00:04, xdelay=00:00:00, mailer=prog, stat=unknown mailer error 2 Dec 20 10:37:42 6C:iago sendmail[18716]: KAB18714: KAA18716: DSN: unknown mailer error 2 Dec 20 10:37:42 3C:iago Mailman mail-wrapper: Failure to exec script. WANTED gid 2147430212, GOT gid 0. (Reconfigure to take 0?) Dec 20 10:37:42 6C:iago sendmail[18716]: KAA18716: to="|/local/mailman/mail/wrapper mailowner testlist", delay=00:00:00, xdelay=00:00:00, mailer=prog, stat=unknown mailer error 2 Dec 20 10:37:42 6C:iago sendmail[18716]: KAA18716: KAB18716: return to sender: unknown mailer error 2 Dec 20 10:37:43 6C:iago sendmail[18716]: KAB18716: to=root, delay=00:00:01, xdelay=00:00:00, mailer=local, stat=Sent
Now, this doesn't seem to be as clear as 'reconfigure mailman to take 0', because that's, well, what I did on the second or third recompile-with-a-different-mail-delivery-gid attempt. I'm also not sure why it wants gid 2147430212, since that's, well, ludicrous.
I feel as though I've tried FAQ-based solutions, and I've ended up stuck. Am I encountering a solved problem here, and I'm just not applying the solution? I sure hope so.
Thanks for any help,
Fred

As a small embellishment, root is getting mail when these bounces happen, reporting the error:
554 "|/local/mailman/mail/wrapper mailowner testlist"... unknown mailer error 2
/local/mailman/mail/wrapper is most certainly there...

Quoting Fred Hicks (iago@iago.net):
As a small embellishment, root is getting mail when these bounces happen, reporting the error:
554 "|/local/mailman/mail/wrapper mailowner testlist"... unknown mailer error 2
/local/mailman/mail/wrapper is most certainly there...
That's a smrsh problem. Depending on how smrsh is configured, you have to place a symlink to /local/mailman/mail/wrapper in either /etc/smrsh or somewhere else.

I don't appear to be running smrsh, being the major complication to that.
On Mon, 20 Dec 1999, Paul Tomblin wrote:
Quoting Fred Hicks (iago@iago.net):
As a small embellishment, root is getting mail when these bounces happen, reporting the error:
554 "|/local/mailman/mail/wrapper mailowner testlist"... unknown mailer error 2
/local/mailman/mail/wrapper is most certainly there...
That's a smrsh problem. Depending on how smrsh is configured, you have to place a symlink to /local/mailman/mail/wrapper in either /etc/smrsh or somewhere else.

Quoting Fred Hicks (iago@iago.net):
recompile-with-a-different-mail-delivery-gid attempt. I'm also not sure why it wants gid 2147430212, since that's, well, ludicrous.
That gid number is probably "nobody", since "nobody"'s gid is almost always -1, and 21477430212 is what you see when yo specify a negative number to a program that takes an unsigned int.

It's not smrsh; I've recompiled and updated sendmail to the latest stable version (8.9.3) and made explicitly certain that smrsh is not running.
It's not that I've compiled it to expect the gid -1 -- there is no gid -1 on my system. The 'nobody' group -- which I only specified for the cgi gid, not the mail gid -- is 60000. I specified '0' or 'sys' or etc etc etc -- i.e., i specified what it says I should specify -- and I CONTINUE to get the error:
Dec 20 14:36:50 6C:iago sendmail[29091]: OAB29089: OAA29091: DSN: unknown mailer error 2 Dec 20 14:36:50 3C:iago Mailman mail-wrapper: Failure to exec script. WANTED gid 2147430212, GOT gid 0. (Reconfigure to take 0?)
It *has* been reconfigured to take zero and has *never* been configured to 'want' 2147430212, or -1, whatever you want to consider that to be.
Is there some barrage of tests I can run to determine how it is that mailman has decided to become compiled incorrectly? That's the only plausible explanation I can arrive at to explain why it would think it wants something it was never configured to want.
On Mon, 20 Dec 1999, Paul Tomblin wrote:
Quoting Fred Hicks (iago@iago.net):
recompile-with-a-different-mail-delivery-gid attempt. I'm also not sure why it wants gid 2147430212, since that's, well, ludicrous.
That gid number is probably "nobody", since "nobody"'s gid is almost always -1, and 21477430212 is what you see when yo specify a negative number to a program that takes an unsigned int.
-- Paul Tomblin, not speaking for anybody. SETI@Home: Finally a *good* way to impress Jodie Foster http://www.setiathome.ssl.berkeley.edu/

"FH" == Fred Hicks iago@iago.net writes:
FH> Is there some barrage of tests I can run to determine how it
FH> is that mailman has decided to become compiled incorrectly?
FH> That's the only plausible explanation I can arrive at to
FH> explain why it would think it wants something it was never
FH> configured to want.
Looks like you're going to have to resort to your favorite C debugger.
-Barry

On Tue, 21 Dec 1999, Barry A. Warsaw wrote:
"FH" == Fred Hicks iago@iago.net writes:
FH> Is there some barrage of tests I can run to determine how it FH> is that mailman has decided to become compiled incorrectly? FH> That's the only plausible explanation I can arrive at to FH> explain why it would think it wants something it was never FH> configured to want.
Looks like you're going to have to resort to your favorite C debugger.
Alas, I have no favorite, as I don't do C. I'm considering the possibility that the answer here is 'don't try to do this under irix'. I am looking into acquiring other machines with other OSes, so I can compare and contrast. Stay tuned.

On Mon, Dec 20, 1999 at 10:32:20PM -0800, Fred Hicks wrote:
On Tue, 21 Dec 1999, Barry A. Warsaw wrote:
FH> Is there some barrage of tests I can run to determine how it FH> is that mailman has decided to become compiled incorrectly? FH> That's the only plausible explanation I can arrive at to FH> explain why it would think it wants something it was never FH> configured to want.
Looks like you're going to have to resort to your favorite C debugger.
Alas, I have no favorite, as I don't do C. I'm considering the possibility that the answer here is 'don't try to do this under irix'.
I'm going to have to install mailman on an Irix box in the near future (thats what i get for promoting mailman at work :) )... and i'm going to do my best to get it to work... btw did you use the SGI compiler? maybe that's the one doing weird stuff with the C-wrappers... I prefer to use GCC on irix (installing it is a pain in the butt, but once it's installed it's a blessing)
Ricardo.
--

On Tue, 21 Dec 1999, Ricardo Kustner wrote:
I'm going to have to install mailman on an Irix box in the near future (thats what i get for promoting mailman at work :) )... and i'm going to do my best to get it to work...
Good luck. :)
btw did you use the SGI compiler? maybe that's the one doing weird stuff with the C-wrappers... I prefer to use GCC on irix (installing it is a pain in the butt, but once it's installed it's a blessing)
Yes, I did use the SGI compiler. I recognize that's a possible problem -- but I'm wary about tossing gcc onto the box. I don't understand enough about compilers to be confident I won't screw up the both of them in the install. Still, it's a possible answer, yes.
If I am feeling brave, I'll try it out. Let me know how your 'go' goes.

Ricardo wins the scooby snack. :)
I grabbed gcc off of freeware.sgi.com and installed the package, then recompiled and reinstalled python with gcc.
Then I recompiled and reinstalled mailman with gcc.
Redid my steps, sent a message. It failed.
*THIS* time, however, it told me it got gid 998, not 0, and when I re./configured --with-mail-gid=guest (998), the sonofabitch WORKED.
There's your tip, Ricardo. Use gcc when you're doing this. :)
On Mon, 20 Dec 1999, Fred Hicks wrote:
On Tue, 21 Dec 1999, Ricardo Kustner wrote:
I'm going to have to install mailman on an Irix box in the near future (thats what i get for promoting mailman at work :) )... and i'm going to do my best to get it to work...
Good luck. :)
btw did you use the SGI compiler? maybe that's the one doing weird stuff with the C-wrappers... I prefer to use GCC on irix (installing it is a pain in the butt, but once it's installed it's a blessing)
Yes, I did use the SGI compiler. I recognize that's a possible problem -- but I'm wary about tossing gcc onto the box. I don't understand enough about compilers to be confident I won't screw up the both of them in the install. Still, it's a possible answer, yes.
If I am feeling brave, I'll try it out. Let me know how your 'go' goes.
Mailman-Users maillist - Mailman-Users@python.org http://www.python.org/mailman/listinfo/mailman-users
participants (4)
-
Barry A. Warsaw
-
Fred Hicks
-
Paul Tomblin
-
Ricardo Kustner