GMail and Sourceforge E-mail Bouncing Saga

An anonymous reader writes "All e-mail going back and forth from Sourceforge and Gmail is being bounced. This leaves many Open Source projects with helpless mailing lists. Fortunately, Sourceforge blames Google and Google is blaming SourceForge for this. The Sourceforge support site is clogged with support requests for a resolution to this problem. Google's response to this bouncing has been automated e-mails saying it is probably at the other end of mail delivery. This is something that the community needs to know about since it has been going on for a week already with no end in sight." Worth noting that Sourceforge and Slashdot are both part of OSTG. Update 20:07 GMT by SM: According to SourceForge support staff this issue is now resolved. Apparently a few days ago the sender-verify to gmail started resulting in 450 errors. Google has since either corrected this issue or whitelisted SourceForge and several tests of the system have resulted in correct delivery.

Comments

Loss of communication can only mean one thing...

[photozz]

Invasion.

Why is the email bouncing?

[Anonymous Coward]

The summary was useless, there's only a few things I want to know about this spat. Who sends the first DSN, why and why was it rejected by the other party?

Re:Why is the email bouncing?

[doti]

Here you go, a complete bounced message from sf.net:

X-Gmail-Received: ecfafb0784517c3cc7f903105542834cd33fde22
Delivered-To: rodolfo.borges@gmail.com
Received: by 10.35.42.5 with SMTP id u5cs205830pyj;
                Sat, 30 Sep 2006 21:26:16 -0700 (PDT)
Received: by 10.35.61.2 with SMTP id o2mr4364526pyk;
                Sat, 30 Sep 2006 21:26:16 -0700 (PDT)
Return-Path:
Received: by 10.35.61.2 with SMTP id o2mr5005562pyk;
                Sat, 30 Sep 2006 21:26:16 -0700 (PDT)
From: Mail Delivery Subsystem
To: rodolfo.borges@gmail.com
Subject: Delivery Status Notification (Delay)
Date: Sat, 30 Sep 2006 21:26:16 -0700 (PDT)

This is an automatically generated Delivery Status Notification

THIS IS A WARNING MESSAGE ONLY.

YOU DO NOT NEED TO RESEND YOUR MESSAGE.

Delivery to the following recipient has been delayed:

          albert@users.sf.net

Message will be retried for 2 more day(s)

Technical details of temporary failure:
TEMP_FAILURE: SMTP Error (state 9): 451-Could not complete sender verify callout
451-Could not complete sender verify callout for .
451-The mail server(s) for the domain may be temporarily unreachable, or
451-they may be permanently unreachable from this server. In the latter case,
451-you need to change the address or create an MX record for its domain
451-if it is supposed to be generally accessible from the Internet.
451 Talk to your mail administrator for details.

      ----- Message header follows -----

Received: by 10.35.61.2 with SMTP id o2mr1893905pyk;
                Fri, 29 Sep 2006 20:41:07 -0700 (PDT)
Received: by 10.35.42.5 with HTTP; Fri, 29 Sep 2006 20:41:07 -0700 (PDT)
Message-ID:
Date: Sat, 30 Sep 2006 00:41:07 -0300
From: "Rodolfo Borges"
To: procps-feedback@lists.sf.net
Subject: pkill -l
Cc: "Kjetil Torgrim Homme" ,
        "Albert Cahalan"
MIME-Version: 1.0
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit
Content-Disposition: inline

      ----- Message body suppressed -----

Re:Loss of communication can only mean one thing..

[Anonymous Coward]

I for one welcome our new e-mail bouncing, blame placing, invading overlords!

Probably Sourceforge?

[Chris_Jefferson]

The message linked to in the post says the person is having trouble with both gmail and sending mail from his own domain. I have also had trouble with sourceforge, where mails from my ISP seemed to be "eaten" about half the time. I've just moved mailing lists off sourceforge, although I'm still using them as their svn support is good. Unless anyone else is having trouble with gmail, I'm tempted to just lay all of the blame at sourceforge.

Re:Probably Sourceforge?

[srussell]

Unless anyone else is having trouble with gmail, I'm tempted to just lay all of the blame at sourceforge.

Hear, hear. Considering all of the problems I've (personally) encountered with SourceForge (broken databases, unresponsive, utterly down) and how few problems I've seen with Google (as in, none), I'd be inclined to think the problem is on SourceForge's end. Google has a reputation for reliability and quality. SourceForge, on the other hand...

--- SER

Umm

[nizo]

Fortunately, Sourceforge blames Google and Google is blaming SourceForge for this.

I don't think that word means what you think it means. Unless you are glad that no one is willing to take responsibility for the problem and fix it???

Re:Umm

[Rob T Firefly]

Fortunately, the word Schadenfreude does mean what I think it means.

Re:Umm

[Anonymous Coward]

Hello, I'm sarcasm, I don't believe we've met before.

Re:Umm

[onion2k]

A sarcasm detector? Like that'd be useful.

Re:Umm

[narzy]

I would find such a device highly useful. I am one of many people who find it hard to differentiate sarcasm from strait forward conver...wait a minute...

Re:Umm

[houghi]

A sarcasm detector? Like that'd be useful.

I think it would, because many people don't recognize it when they see it.

SourceForge is good for spewing into the ether...

[jazzkat]

I've had cases where mail coming from SourceForge never reached me; their servers never even attempted to connect to my e-mail server (i.e. nothing in the logs to indicate this). I was running my own DNS at the time, at a colocation center, and never had problems sending or receiving e-mail before with any other domains.

I don't see the problem...

[Anonymous Coward]

...you all have the source code, and the developers do not consider this a priority, so feel free to solve your problem and post a patch

well this looks clear as mud

[LiquidCoooled]

Greetings,

This is something recent that has changed in how Google handles
email (other sites have started to get the same errors). We
are investigating how to deal with this.

SourceForge.net Support

Is it because sourceforge is not following the RFCs and google has just tightened up?

We had a similar issue in one of our programs where mailing worked wonderfully for months and months for all customers, then one morning complaints started.
It appears as though we weren't following the RFCs to the letter and the main isp in our country (bt) had updated to a more stringent mail server (we shockingly used an additional CR where one was not expected...).

This all sounds similar.

Open Source vs. Google

[patio11]

Who is the Slashdotter to root for? Hmm... I know, third option! It is Microsoft's fault!

Re:Loss of communication can only mean one thing..

[garcia]

They don't have a "+1 Nerd Movie Quotes" mod, sorry. :)

Google's Answer if they find it is them...

[tecker]

Original Response:

Google's response to this bouncing has been automated e-mails saying it is probably at the other end of mail delivery.

New Response:
"Well Gmail is still in beta so don't blame us."

I beg to differ

[geoffspear]

"All e-mail going back and forth from Sourceforge and Gmail" is certainly not being bounced. My Gmail account has been getting plenty of email from Sourceforge during the period when "all e-mail" has supposedly been bouncing.

Of course, this is the sort of accuracy I expect from Slashdot.

SourceForge uses Mailman

[ben there...]

In my opinion, the problem with SourceForge lies in that Mailman doesn't work well with Gmail addresses. I use Mailman discussion lists on my DreamHost account, and while testing I couldn't get the emails to work until I added a non-gmail account. I contacted support, blaming them for a while and getting frustrated, until I tried a different email account.

This was DreamHost's response:

I've closed out this ticket for you. I thought I should mention however
that quite a few people that have forwards to gmail have ran into similar
problem, the only thing that is consistent is that the messages make it
to the gmail relays and then disappear.

I don't know if that means that GMail rejects Mailman messages, or Mailman has problems sending to Gmail addresses, but one way or another, it doesn't work right.
--
Use coupon DH75OFF to get $75 off hosting at DreamHost.com

It is definitely Sourceforge's problem

[kindbud]

Sourceforge is posting the following message to bug reports about this problem.

Greetings,
    We're aware of the difficulties in the interaction
between
our mailing list services and Gmail. Our network operations
team
is currently aware of the issue and is working with Gmail
administration on a resolution.

-Jay Bonci
Systems Programmer Analyst,
Sourceforge.net

Somebody posted a SMTP dialog to one of the bug reports:

Example:
telnet mail.sourceforge.net 25
Trying 66.35.250.206...
Connected to mail.sourceforge.net.
Escape character is '^]'.
220 mail.sourceforge.net ESMTP Exim 4.44 Sat, 30 Sep
2006 01:12:02 -0700 sc8-sf-mx1.sourceforge.net
HELO aisa.fi.muni.cz
250 mail.sourceforge.net Hello 14397 at aisa.fi.muni.cz [147.251.48.1]
mail from:
250 OK
rcpt to:
451-Could not complete sender verify callout
451-Could not complete sender verify callout for <anyone@gmail.com>
451-The mail server(s) for the domain may be temporarily unreachable, or
451-they may be permanently unreachable from this server. In the latter case,
451-you need to change the address or create an MX record for its domain
451-if it is supposed to be generally accessible from the Internet.
451 Talk to your mail administrator for details.
QUIT
221 mail.sourceforge.net closing connection
Connection closed by foreign host.

Sourceforge's mail server is doing a callback to gmail.com, to verify the sender address is accepted by gmail.com. This check is screwing up. It's Sourceforge's problem. Callback verify is not covered by any RFC, so SF has gone above and beyond the standards, it is their responsibility to make sure their SMTP service is interoperable with standard servers, not the other way around. Google can provide logs of the failed callbacks, but that's all the burden they should assume. It's SF's problem to fix.

Re:SPF records....

[nuzak]

> Google has SPF records. Sourceforge seems to reject mail that seems spoofed (eg people 'pretending' to be allowed to send user@gmail.com mail without going through google.

SPF has nothing to do with it. Sourceforge is employing callback verification, which is not only abuse itself (it's basically a dictionary attack that we're just supposed to trust is for good and not evil), it's also incredibly broken.

See http://atm.tut.fi/list-archive/nanog/msg37172.html for an explanation.

Just one more reason to jump ship from sourceforget.

Re:Callbacks Are Evil

[140Mandak262Jamuna]

It is true that there are vast botnets. And the spammers routinely change the bots. And most bots are on dynamic ip address that keeps changing. You are right in saying that I or you or most other companies would not have the resources to combat spam by tagging the ip addresses. But if there is a company that has the resources, both in terms of money and in terms of searching, organizing and finding patterns it would be Google.

Most legitimate mailservers are running on static ip addresses. Google will be able to compile a list of legitimate good mailservers rather quickly. Google is also an IP address registrar. It has the routing tables and other registration information and netblock ownership information. It will know the dynamic ip addresses by the block. Mailservers running on dynamic addresses, or relays running dynamic addresses are suspect immediately. It is not proof. But more like preponderance of evidence (IANAL).

Can they determine spam without callbacks in three months. No way. Can they reduce the number of callbacks to confirm legitimacy of email by atleast an order of magnitude? Yes, they can by collecting relay ip addresses, mail server ip addresses, netblock ownership data and putting them all together like "page-rank", "mailserver-rank". They might even find the bots and inform the ISP that they probably have a bot and the ISPs might even contact the boob with the infected machine. Good things can come out of this.

Will they? There you got me. Dont know if they will. But I hope they do.