This is the mail archive of the gcc@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]

Re: [OT, for Craig only] Re: mail sent to <craig@jcb-sc.com> is bouncing, why?


Yes, 24,999 other people are working fine... I agree it's important to keep
this in mind when looking for trouble.

Jonathon
--
Jonathon Alsop   jalsop@world.com
The World: Director of Web Services   http://www.std.com/web
617-739-0202 (main)   617-739-4927 (direct)

----- Original Message -----
From: <craig@jcb-sc.com>
To: <pderbysh@usa.net>
Cc: <egcs@egcs.cygnus.com>; <postmaster@globalserve.net>;
<kiosk@world.std.com>; <postmaster@world.std.com>; <craig@jcb-sc.com>;
<burley@gnu.org>; <burley@world.std.com>
Sent: Sunday, February 14, 1999 11:07 PM
Subject: Re: [OT, for Craig only] Re: mail sent to <craig@jcb-sc.com> is
bouncing, why?


[Again, this doesn't belong on the egcs list, except I feel readers
of that list should know I don't think it's 100% clear world.std.com
is doing anything (seriously) wrong.  Maybe they're just blocking
a known spam site, or think they are, or maybe the SMTP connection
their server is accepting has something wrong being sent down it
that they're rejecting.

Paul, and anyone else, could you switch to mailing me about this topic
my old address, <burley@gnu.org>, in addition to my normal one, to
ensure it'll get through?  I think I've set up the Reply-To header
accordingly, but please check it.

At the same time, it's best to stop mailing the egcs list, except maybe
a final summary, regarding this particular problem.  I'll just pick
my email up from gnu.org more frequently...assuming the systems are
up and running when I try, which is a whole 'nother problem (it
being Sunday night and all).]


[Also, I've tried to reconstruct and quote some of the message after
chopping some out at first and then answering, plus I've tried
to add the original bounce message that was in my previous email
in appropriate '>'-notation context, so the people at the postmaster
addresses getting this will have a reasonably complete record in
this one email.]


>At 03:24 PM 2/13/99 -0000, you wrote:
>>Do you have any bounced email to forward along, so I can see what
>>was actually happening?  That's the first thing my ISP would
>>ask for, probably.
>
>Yeah. Here. (Apologies to list, but since Craig's mail server is still
>bouncing mail intermittently...)
>
>I already tried sending this to postmaster@jcb-sc.com, and got the same
>sort of bounce. I did Whois lookup and saw their technical and admin
>contacts at world.std.com; I mailed them suggesting they fire the
>postmaster and noc at jcb-sc.com and telling them why, and world.std.com
>bounced as well. Is damned weird.
>
>At least if you suddenly find that your subscription to egcs lists has
>evaporated, you'll know why (your server started bouncing incoming mail
>again and triggered the automatic bad-address-unsubscribe) and who to blame
>(your ISP).
>
>
>>The original message was received at Fri, 12 Feb 1999 19:02:26 -0500 (EST)
>>from dialin109.ottawa.globalserve.net [207.176.153.109]
>>
>>   ----- The following addresses had permanent fatal errors -----
>><craig@jcb-sc.com>
>>
>>   ----- Transcript of session follows -----
>>... while talking to world.std.com.:
>>>>> MAIL From:<pderbysh@usa.net>
>><<< 550 Wrong.
>>554 <craig@jcb-sc.com>... Service unavailable
>>Reporting-MTA: dns; smtp2.globalserve.net
>>Received-From-MTA: DNS; dialin109.ottawa.globalserve.net
>>Arrival-Date: Fri, 12 Feb 1999 19:02:26 -0500 (EST)
>>
>>Final-Recipient: RFC822; craig@jcb-sc.com
>>Action: failed
>>Status: 5.0.0
>>Remote-MTA: DNS; world.std.com
>>Diagnostic-Code: SMTP; 550 Wrong.
>>Last-Attempt-Date: Fri, 12 Feb 1999 19:02:40 -0500 (EST)
>>Return-Path: <pderbysh@usa.net>
>>Received: from chaoszone (dialin109.ottawa.globalserve.net
[207.176.153.109])
>> by smtp2.globalserve.net (8.9.1/8.9.1) with SMTP id TAA19254;
>> Fri, 12 Feb 1999 19:02:26 -0500 (EST)
>>Message-Id: <3.0.6.32.19990212190141.00891c70@pop.netaddress.com>
>>X-Sender: pderbysh@pop.netaddress.com
>>X-Mailer: QUALCOMM Windows Eudora Light Version 3.0.6 (32)
>>Date: Fri, 12 Feb 1999 19:01:41 -0500
>>To: craig@jcb-sc.com
>>From: Paul Derbyshire <pderbysh@usa.net>
>>Subject: Enough is enough... truce time. Re: kernel blah blah blah
>>Cc: egcs@egcs.cygnus.com
>>In-Reply-To: <19990212221225.22651.qmail@deer>
>>References: <7a203j$9cf$1@palladium.transmeta.com>
>> <Pine.LNX.3.95.990209135218.2116D-100000@penguin.transmeta.com>
>> <19990211201836.2781.qmail@deer>
>> <19990211231523.14202.qmail@piglet.chem-eng.nwu.edu>
>> <19990212031118.4271.qmail@deer>
>> <7a203j$9cf$1@palladium.transmeta.com>
>>Mime-Version: 1.0
>>Content-Type: text/plain; charset="us-ascii"


[Getting back to your most recent email....]


>POSTED BECAUSE CRAIG'S ADDRESS STILL BOUNCES.
>PLEASE DISREGARD. I APOLOGIZE TO OTHERS ON EGCS, ON BEHALF OF THE MORONS AT
>WORLD.STD.COM WHOSE FAULT THIS UNDOUBTEDLY IS.

Maybe.  I don't know enough about the relevant standards and protocols
to know.  I do know that I get tons of email from lots of different
places and people.  (Probably more than 100 pieces per day, and,
thankfully, hardly any spam at all, most of which comes via the
old egcs aliases and my [forwarded] account at Cygnus...whereas, at
gnu.org, it's down to 10 or so emails per day, almost all spam.)

Although, strangely, I don't have a copy of my own email, the first
one I sent to this list with the *current* subject line, to which
you responded.  Maybe I forgot to dump at least one of the two
copies I should have gotten into my archive (I do it "by hand"),
but it's a slightly worrisome glitch.  I'll watch carefully what
happens to *this* email, which is going to three of my four
email addresses, plus the egcs list, so I should get four copies,
picking up three at world.std.com, the fourth at gnu.org.  Ain't
technology wonderful?  Sigh.

>At 08:20 PM 2/14/99 -0000, you wrote:
>>Thanks.  That's so weird: I saw the email you enclosed sometime late
>>Friday night, when I fetched my email (I could narrow that down by
>>examining my archives, if necessary, I suppose).
>
>Via the list, or a personal copy (with no X-Mailing-List, etc.)?

Via the list.  Had "owner-egcs" on it as a resender or whatever.

>[...]
>
>>So, I *suppose* I could fire myself, as both postmaster and `noc',
>>whatever that is ("network oxymoron consultant"?  :).  But there's
>>got to be a better way.
>
>Hm. Well if anyone needs firing or something it's at world.std.com then.
>
>>So far, I'm not aware that I've actually missed anything.  Are *all*
>>my messages bouncing (this wouldn't count the ones from egcs-cvs,
>>since those are sent by the egcs repository server, I assume)?  Or
>>just some of them?
>
>It must be intermittent. You clearly are getting mail from the list fairly
>conistently with only a small amount of trouble, or else it'd have kicked
>you off and you'd have failed to reply to some things or reported seeing
>messages following up to messages you didn't receive even days later.
>
>>As far as what little I can make of the header below: there seems to
>>be little semantic information in the message "550 Wrong." from which
>>I can gather a clue.
>
>Well, it didn't make sense to me. It seemed to suggest my address was
>ill-formed, which is about as sensible as a compiler responding to
>
>int main (void) {
>  int j;
>  j=5;
>  ++j;
>  return j;
>}
>
>with
>
>foo.c: 5: Parse error before 'j'
>
>
>>But, from "554 <craig@jcb-sc.com>... Service
>>unavailable", it looks like maybe what happened is that some MTA
>>tried to deliver the mail *directly* to `jcb-sc.com', instead of
>>to `world.std.com'.
>
>jcb-sc.com (your box) doesn't run a working SMTP server? But to generate
>this message it'd have to be running an SMTP server programmed to bounce
>all mail, when it seems easier to just not run any at all.

My box *does* run an SMTP server, but only for emailing within my
LAN (behind a firewall).  (Which, by the way, constitutes only one
machine right now, firewall and all, but soon after I get the network
card fixed, I'll add the other two.  So, right now, the only user
of my LAN's SMTP server is, I believe, fetchmail; later users will
be whatever daemons run on the other machines, e.g. building and
testing egcs and emailing results to myself and/or to egcs-results.)

Anyway, the bounce message doesn't indicate any attempt to open a port
to `jcb-sc.com', which is just as well, since, for the purposes of
SMTP from the Internet, it doesn't exist.  It's supposed to contact
`world.std.com', but I can't tell whether it's doing that correctly
*and* giving it the proper MAIL FROM: command, etc.  (I suppose
world.std.com's SMTP server could be choosing to reject email
from your direction, thinking that spam issues therefrom, but I
don't know.)

>And "Wrong"
>would mean I (or rather an MTA involved in my mail) mailed the wrong
>machine instead of world.std.com. Except that doesn't explain world.std.com
>itself bouncing the same way, and it seems odd your box might run an SMTP
>server configured this way wihout you, "root", knowing... although you
>could then try to find it and change it to forward to burley@world.std.com
>any incoming mail to patch over the problem.

That's what happens anyway.  All email sent to `anyuser@jcb-sc.com'
is placed, by my ISP's SMTP server (on the whole) directly into my
mailbox at the ISP, from what I understand.  Doesn't matter whether
my LAN is connected -- even when it is, the Internet (DNS) has no
clue that `jcb-sc.com' is connected, it remains just an MX record
pointing to `world.std.com'.

>>How could this happen, with me still getting the email, I ask myself,
>>since my LAN shouldn't accept any outsiders trying to connect to its
>>SMTP server?
>
>Oh you've got a LAN? And apparently I'm the outsider...

Right, my little LAN is behind one of it's machines own firewalls.
I just PPP into my ISP, get a dynamic IP, then work from there.

So everyone is an "outsider" to my LAN, including my wife (when
she's using her laptop connected to her computers at work -- which
is just fine, they run all that MS software).

>>And perhaps *that* copy never made it here, because some MTA didn't
>>take the advice of the MX record to deliver it to `world.std.com',
>>and I happened to be off-line and/or my SMTP server (qmail) rejected
>>the connection (I have a non-null rcpthosts, which will mean something
>>to the qmail-aware).
>
>hmm.
>
>>[craig@deer gnu]$ nslookup
>>*** Can't find server name for address 192.168.0.1: Non-existent
host/domain
>>Default Server:  world.std.com
>>Address:  192.74.137.5
>>
>>> jcb-sc.com
>>Server:  world.std.com
>>Address:  192.74.137.5
>>
>>*** world.std.com can't find jcb-sc.com: Non-existent host/domain
>>> ls -d jcb-sc.com
>>[world.std.com]
>>$ORIGIN jcb-sc.com.
>>@ 1d2s IN SOA world.std.com. netadmin.world.std.com. (
>> 98101501 ; serial
>> 5D ; refresh
>> 1H ; retry
>> 2w6d ; expiry
>> 1d2s ) ; minimum
>>
>> 1d2s IN NS world.std.com.
>> 1d2s IN NS AUTH01.NS.UU.NET.
>> 1d2s IN MX 10 world.std.com.
>> 1d2s IN SOA world.std.com. netadmin.world.std.com. (
>> 98101501 ; serial
>> 5D ; refresh
>> 1H ; retry
>> 2w6d ; expiry
>> 1d2s ) ; minimum
>>
>>So it might not be a problem on my end, or even my ISP's end at all,
>>though I'm *really* speculating here, as I'm a newbie with all this
>>email stuff (whatever happened to smoke signals?).
>
>I dunno. I'm sure there's no problem with globalserve.net's SMTP since I
>have no problem with any other mailing, but I'll copy my postmaster on it
>anyways to see if they have any idea.

There's someone else who emailed me about how he had no troubles
emailing me, no bounces, but how he *also* has an email address
with a host name that the Internet sees as just an MX record.  I
suggested he considering asking you to email him some test messages,
so maybe he'll contact you.

Because, it could be the case that your ISP's SMTP has a bug in that
it isn't doing the kosher thing emailing to a host with only an MX
record pointing to another host (whatever that kosher thing should
be; I don't know).

>>If you want to test my theory, and you've had bounces from me
>>at <craig@jcb-sc.com>, try sending the same test message to that
>>*and* to <burley@world.std.com>...
>
>Done. It bounces, kiosk bounces, and craig@jcb-sc.com still bounces.

Now, *that* makes me think maybe there's a (spurious?) spam-guard
at world.std.com blocking emails from your direction after all.

(Just this past week, world.std.com got their outgoing email server
off of the ORBS data base, after a bit of prodding, which has
allowed my emails to the egcs.cygnus.com lists to make it through,
which is why I've only finally started getting back to serious
work since then.  So they do seem to know what they're doing,
even though, at first, they claimed ORBS listed sites for spurious
reasons -- apparently, that *was* true, at some point in the past.)

I mean, presumably you can understand that if world.std.com can't
receive email sent to its users, its special-services staff, and
an email address provided via one of its special services, it
wouldn't take *our* complaints to get their attention!!

(To give you some idea, I've enclosed the the outputs of `uptime' and
`who' commands I just performed on their main system, this being
late Sunday evening, certainly not the busiest of time or load for
them.)

>Retards. The idiots that run world.std.com's mailserver deserve to come
>down with a really bad std. I've posted all of this to
>news.admin.net-abuse.email. Usually email net-abuse means spam, but this
>SMTP server problem at world.std.com strikes me as a sort of net-abuse
>related to email, since arbitrary bouncing of well-formed and non-abusive
>messages is abusive, particularly when it even makes your zone and
>technical contacts unreachable.

That's why, in a similar situation, I would lean more towards thinking
something was wrong on the sending end, since an ISP screwed up to the
extent apparently visible from your end would hardly need any public
attention called to their problem -- their customers would be complaining
like crazy.

But I'll allow you're probably doing the right thing -- I didn't even
*know* about that newsgroup (and I'm not reading USENET these days
anyway).  It's nice of you to go after this problem on my behalf (and
on the behalf of world.std.com users) -- I've had enough of yelling
at people this week to last me for the rest of the millenium, and,
as I say, I just don't know enough to have a clue *who* to yell at
here, since it isn't clear to me who's at fault.

        tq vm, (burley)


[$ uptime]
 10:24pm  up 15 days,  5:17, 275 customers,  load average: 10.61, 9.94, 9.41

(That's 275 people logged in.  The system did not feel sluggish at
all: telnet'ing and ftp'ing in was a breeze.  They're running
some pretty *serious* iron.  They're no mom & pop outfit running
an ISP out the back of an old 486.  I decided to not send the
"who" output due to privacy concerns, etc., but at the time I did
it, 273 people were logged in.  I personally know a few people
who use, or have used, this ISP.  They're not mythical, and I
don't recall hearing any complaints, like I have heard from friends
about *other* ISPs.  I believe two or three egcs luminaries
recommended this ISP when they learned I was looking for one to
pay, in lieu of relying on free access to gnu.org.)




Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]