[Raw Msg Headers][Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: mailbox processes growing
On Thu, Dec 07, 2000 at 08:22:34PM +0300, Eugene Crosser wrote:
> It seems that unreasonable growth of mailbox processes is not connected
> with "-S" option: they grow without it as well. That is, before
> mimeheaders reorganization. Now I put in production recent CVS version,
> will see if things change...
I saw such hyper-growth two days ago myself too.
It did lead to out-of-memory condition, and all merriment
what *that* one causes..
Out of Memory: Killed process 26690 (mailbox).
Out of Memory: Killed process 26690 (mailbox).
Out of Memory: Killed process 26690 (mailbox).
Out of Memory: Killed process 26690 (mailbox).
Out of Memory: Killed process 26690 (mailbox).
Out of Memory: Killed process 26690 (mailbox).
Out of Memory: Killed process 26690 (mailbox).
Out of Memory: Killed process 26690 (mailbox).
__alloc_pages: 0-order allocation failed.
VM: killing process mailbox
(yeah, linux kernel really repeated itself like that..)
I have that mailbox running with -S, and I *think* the case
is related to having malformed incoming headers, but damned
if I could determine what -- naturally it cleared itself auto-
magically without enabling me to repeat the test.
I had a hunch that *lack of* Content-Transfer-Encoding:
header with otherwise normal MIME message is related to this,
but the blasted thing didn't repeat :-/ --> my hunch could
be wrong.
I think the message in question was some sort of spam coming
into my system. I think it had these headers ( + some Receiveds):
From: <lauren554@achill.net>
To: matti.aarnio@sonera.com
Subject: HI ...
Reply-To: lauren32@jahoopa.com
Mime-Version: 1.0
Content-Type: text/plain, charset="iso-8859-1"
Date: Thu, 7 Dec 2000 01:09:13
Notably it is missing Message-ID and Content-Transfer-Encoding.
In couple hours time I will be leaving for flights to San Diego
IETF (www.ietf.org), and while I have quite excellent network
connectivity from there, I am also quite busy.
... but in case I have free cycles, and nothing better to do,
I might look into getpwnam() return code checks, at least...
> Eugene
--
/Matti Aarnio <mea@nic.funet.fi>