Not getting emails with attachments over a few kilobytes

[jbs]

I'm using a sg address as the destination for my jfax number, an electronic fax service, wherein you fax to xxx-xxx-xxxx and they turn it into a tif file, about 40k or so for one page, and send it to your email address.

Today I had a couple of larger faxes not come through to my email account, and set out to understand why.

So using the decrement counter, I can tell that the messages are being sent TO sg, and bypassing sg and faxing straight to my email provider I can see that THEY are able to receive them, but for some reason the larger faxes which go through SG never get to my email provider.

These larger faxes are still only on the order of 100 to 400 kilobytes . . . what is the size limit for SG forwarding? (FAQ doesn't mention the words size kilobytes or attachment).

Earlier in the day I did receive a 340k fax attachment through SG, but not since then -- is there a daily limit? Or is there a current backlog of some sort??

Thanks!

--Jason

[jbs]

The large fax referenced above finally came through, and the headers point to a delay within spamgourmet -- any ideas why?

It seems to occur during the second line of the headers, from gourmet.spamgourmet to localhost. Gourmet.spamgourmet got the message at 14:50:51, but the message spent 3.5 hours there before moving to localhost (still part of SG).

Am I reading this correctly? And is there anything we can do to improve?

Thanks!

--Jason

Received: from [216.218.230.146] (helo=gourmet.spamgourmet.com)
by fetch.runbox.com with esmtp (Exim 4.34)
id 1CSnVl-00063q-7S
for xxxxxxxx+sg@runbox.com; Fri, 12 Nov 2004 18:19:38 -0800
Received: from gourmet.spamgourmet.com (localhost [127.0.0.1])
by localhost (8.12.10/8.12.9) with ESMTP id iAD2E2K5001989
for <xxxxxxxx+sg@runbox.com>; Fri, 12 Nov 2004 18:19:27 -0800
Received: (from jqh1@localhost)
by gourmet.spamgourmet.com (8.12.10/8.12.10/Submit) id iACMop5q005776
for xxxxxxxx+sg@runbox.com; Fri, 12 Nov 2004 14:50:51 -0800
Received: from efax.j2.com (iad12.efax.com [199.232.57.35])
by gourmet.spamgourmet.com (8.12.10/8.12.9) with ESMTP id iACMomU5005734
for <jfax.10.xxxxxxxx@xoxy.net>; Fri, 12 Nov 2004 14:50:49 -0800
Received: from media1.iad1.colo.j2noc.com (media1 [10.11.57.103])
by efax.j2.com (8.11.6/8.11.6) with ESMTP id iACModW23831;
Fri, 12 Nov 2004 22:50:39 GMT
Received: (from j2apps@localhost)
by media1.iad1.colo.j2noc.com (8.11.6/8.11.6) id iACMm1627638;
Fri, 12 Nov 2004 22:48:01 GMT

[SysKoll]

I recently got a message with a 1.2 MB attachment on one of my disposables. Could you try to send 3 messages to your SG address with an attachment of 1, 2 and 4 MB and see if there is a cut-off at some critical size?

[jbs]

This does not seem to be related to any specific cutoff, since I can get large attachments through SG addresses.

But the problem continues with this fax service. I've added another (non-SG) address to my fax account because this problem seems to be continuing.

Attachments of even one or two pages decrement my count on SG almost immediately, as reflected in the message headers, but then spends 3 or more hours queued at SpamGourmet before heading out to my account:

Received: from gourmet.spamgourmet.com (localhost [127.0.0.1])
by localhost (8.12.10/8.12.9) with ESMTP id iAD2E2K5001989
for <xxxxxxx@runbox.com>; Fri, 12 Nov 2004 18:19:27 -0800
Received: (from jqh1@localhost)
by gourmet.spamgourmet.com (8.12.10/8.12.10/Submit) id iACMop5q005776
for xxxxxxx@runbox.com; Fri, 12 Nov 2004 14:50:51 -0800

In this case, 3 and a half hours.

Any thoughts on why this is happening? The attachment was only 314 kilobytes.

Thanks for any help you can offer!

--Jason

[josh]

I'm not sure what would have caused that delay, other than an unusual surge of activity that put the server behind. I'll look into it.

[jbs]

And just to be clear, it's not a temporary thing, as of the time of this posting, it's still happening.

I added my destination (runbox) address to my jfax account, so that jfax fires off the fax simultaneously to both Runbox and Spamgourmet. Tonight I got 2 copies of a fax, separated by several hours, which the header revealed to be a gap at Spamgourmet . . .

Is there any possibility that it's related to the server the messages are coming from? The strange thing is that if I turn around and forward the message immediately to both Spamgourmet and my destination address, both arrive within seconds. That is to say, the same attachment coming FROM me TO me goes through SG in a matter of seconds, but when it comes FROM jfax and TO me, it is held for several hours at SG.

Any ideas?

Thanks!

--Jason

[SysKoll]

JBS,

That's the smoking gun. If the same attachment goes through sg to jfax very quickly, then clearly it's not the fault of sg.

I *think* that what's probably happening is that jfax (ot rather its faxing server) is saturated and takes a very long time to trickle the message to sg.

Does that make sense? Or am I missing something?

[jbs]

No, I probably did not explain it well enough.

So, a fax is sent to my number, which is hosted by jfax. They turn it into a TIF file, and email it to 2 addresses: my SG address and another one at Runbox which is the address to which SG forwards.

Meanwhile, I log into my Runbox account and look for the fax. The one which comes straight to Runbox shows up within a matter of seconds, BUT, the one which went through SG takes several hours.

Moreover, the headers which I've copied earlier in this thred show that jfax is getting the message to SG within a few seconds, but the delay is between servers at SG. If you look at the various entries, you'll see that the handoff from SG to Runbox occurs several hours after the handoff from jfax to SG.

Finally, I've watched the increment count at SG for the disposable address I use for jfax -- the count on the SG website goes up within a few seconds of the message being sent from jfax, but then it does not show up in my Runbox account for several hours.

Does this help?

--Jason

[SysKoll]

OK, I understand better, thanks for the detailed explanation.

So, when this TIFF image is transmitted:

[li]from jfax to runbox, it takes seconds
[li]from jfax to sg, it takes seconds
[li] from sg to runbox, it takes hours

An interesting test would be to logon to the email server and send a 1 MB message to two accounts, your Runbox and a control account.

Please send me your email address at runbox in a private message (don't post it on the forum). I'll try to do that test ASAP.

[jbs]

Well, if I'm reading the email headers correctly, it's also only seconds from SG to Runbox, the issue is happening entirely within the SG servers. There are (again, according to my read of the headers) multiple SG hops within the handoff chain.

The large gap in time occurs entirely BETWEEN SG servers, not between the SG and Runbox servers.

I've tested it as well by sending the same attachment from an outside server (in this case an AOL account) both directly to my Runbox account and to my SG account. While the direct message arrives slightly sooner, the difference is merely 30-120 seconds, which is to be expected.

It seems to be only when the message is sent FROM jfax through SG that it gets caught for so long . . .

Strange, huh?

By the way, can someone confirm that I'm reading the headers correctly, and that there is, in fact, a delay between SG servers handling the message . . .

--Jason

[maratheamit]

If I remember correctly, the perl code reads in the entire message into memory, line by line. Could that be causing a big delay when the message size is in the hundreds of KB?

That does not explain why jfax is singled out for the delay. I can confirm that a large 2MB attachment from gmail to sg to yahoo arrives without delay.

[jbs]

If anyone else is interested in testing this, you can set up a free account at jfax.com, limited to 20 pages a month receiving. Anything more than a couple of pages seems to trigger the delay, though.

And I concur that large attachments (even much larger) from other sources get to my runbox account in seconds.

Can someone with better knowledge of headers than I have glance at the ones I've copied and confirm that this delay is happening between SG servers, and not in transit to or from SG?

--Jason

[SysKoll]

JBS,

I'm ready to perform the test. Please let me know through private message what your runbox.com address is and how large is the average fax attachment. I'll send you an email from the command line and watch the log for problems.

[jbs]

Syskoll-

Email address sent via PM.

Thanks! I'll let you know what and when I receive.

--Jason