Mailing List Archive

On Sun, 2008-12-14 at 21:45 +0000, G.W. Haywood wrote:
> Hi there,
>
> Memory usage graph for spfmilter using libspf2-1.2.9:
>
> http://www.jubileegroup.co.uk/JOS/misc/spfmilter_memory_consumption_one_month.gif
>
> This is a fairly quiet mailserver handling about 5,000 to 10,000 message
> attempts per day, most of which are of course rejected as spam either by
> the numerous firewall rules (50,000+) or by the mailserver itself.
>
> Does anyone have any comments to make on the memory consumption?

Hrm. At a guess, something's losing a chunk, but neither I nor our
earlier coverity scan could immediately see it. There is always the
possibility that it's just filling in the DNS cache, and that the jumps
in usage are just finding a free bucket in the open hash table. It's not
climbing drastically. Have you run the system under valgrind? Do you
know how many messages actually hit the milter?

> The milter was originally version 0.97 from
>
> http://www.acme.com/software/spfmilter/
>
> but I have modified it to work with libspf2-1.2.9 as its development
> seems to have been abandoned by the author. The source is available
> if anyone wants it. I'm still working on the comments... :)

I would be tempted, at this stage, to fold spfmilter into the libspf2
tree, for ease of distribution and installation, but I am not keen to
maintain it. Would you be interested in being a co-maintainer in
subversion if I were to do so?

S.



-------------------------------------------
Sender Policy Framework: http://www.openspf.org
Modify Your Subscription: http://www.listbox.com/member/
Archives: https://www.listbox.com/member/archive/1007/=now
RSS Feed: https://www.listbox.com/member/archive/rss/1007/
Powered by Listbox: http://www.listbox.com

Hi Shevek,

On Mon, 15 Dec 2008, Shevek wrote:

> On Sun, 2008-12-14 at 21:45 +0000, G.W. Haywood wrote:
> >
> > Memory usage graph for spfmilter using libspf2-1.2.9:
> > ...
> > Does anyone have any comments to make on the memory consumption?
>
> Hrm. At a guess, something's losing a chunk, but neither I nor our
> earlier coverity scan could immediately see it. There is always the
> possibility that it's just filling in the DNS cache, and that the jumps
> in usage are just finding a free bucket in the open hash table.

I was hoping you might have suggestions of that sort, that's why I
posted before investigating further.

> It's not climbing drastically. Have you run the system under
> valgrind? Do you know how many messages actually hit the milter?

That's right, it's perfectly manageable on this fairly quiet system.
No, I haven't run it under valgrind, nor have I done any investigation
other than looking at the graph and wondering. :) Approximately 2,000
messages actually hit the milter in the period covered by the graph.

> I would be tempted, at this stage, to fold spfmilter into the libspf2
> tree, for ease of distribution and installation, but I am not keen to
> maintain it. Would you be interested in being a co-maintainer in
> subversion if I were to do so?

That sounds fine to me, I'll be happy to do that, and I'm happy to look
deeper into this memory issue too. Watch this space (but don't hold
your breath... :)

--

73,
Ged.


-------------------------------------------
Sender Policy Framework: http://www.openspf.org
Modify Your Subscription: http://www.listbox.com/member/
Archives: https://www.listbox.com/member/archive/1007/=now
RSS Feed: https://www.listbox.com/member/archive/rss/1007/
Powered by Listbox: http://www.listbox.com