Re: [spamdyke-users] recipient-blacklist-file=FILE with RegExes?

2015-06-20 Thread Sam Clippinger via spamdyke-users
Regex support is on the (rather lengthy) to-do list, but frankly it's not a 
very high priority -- there's a lot of low-hanging fruit that would be of much 
more benefit right now.  Plus, since I'm not one of the 10 people in the world 
who completely understands regexes, I doubt I would actually use them myself; 
I'd rather add globbing support, which I do understand. :)

spamdyke's header filter runs at connection time, as all of its filters do.  If 
a header line matches a blacklisted pattern, the entire message is rejected 
(the sending server receives an error code, qmail never sees the message).

-- Sam Clippinger




On Jun 19, 2015, at 9:09 PM, Philip Rhoades via spamdyke-users 
spamdyke-users@spamdyke.org wrote:

 Sam,
 
 See inline comments:
 
 
 On 2015-06-20 11:53, Sam Clippinger via spamdyke-users wrote:
 You're correct spamdyke does not support regexes for any of its
 options, but you can use a wildcard in a sender or recipient
 white/blacklist file to match entire domains by prefixing the line
 with an @ symbol. For example:
 @example.com [1]
 
 
 Yep, saw that - is it possible to support regexes in the future?
 
 
 Full documentation here:
 http://www.spamdyke.org/documentation/README.html#REJECTING_RECIPIENTS
 [2]
 BUT! Be careful -- the To and From lines in the message header are
 not the same as the sender and recipient. The sender and recipient
 are part of SMTP, the To and From lines are part of the message data
 and are completely unrelated. Think of it this way: when a letter is
 sent through the post office, the name on the outside of the envelope
 tells the postman which mailbox gets the envelope (or where to send it
 back to) but top of the letter inside may have a completely unrelated
 letterhead and salutation. Whenever spamdyke's options/documentation
 refer to a sender or a recipient, it means the name on the outside
 of the envelope. The user never sees those values in their mail client
 unless the sender chooses to use those values in the To and From
 fields. Spammers typically fake all sender/recipient/To/From fields,
 but other software does too for perfectly legitimate reasons (e.g.
 mailing lists, autoresponders).
 
 
 Right.
 
 
 If you want to block based on the To
 and From lines the user sees in their mail client, you should look at
 spamdyke's header blacklist filter:
 http://www.spamdyke.org/documentation/README.html#HEADERS [3]
 
 
 In that case the mail has already been accepted?  When I was using the 
 qmail-qfilter+Ruby script method - my understanding of it at least - was that 
 my Ruby script could process the header and body of the email and exit with a 
 particular error code if the mail was bad and this would terminate the SMTP 
 negotiation with that error message (eg drop the mail silently).  So in this 
 case I was able to look at all the header fields as well as the mail body and 
 do whatever I wanted before accepting the mail.
 
 
 Header filtering doesn't support regexes either, but it does use
 globbing to allow more wildcard options.
 
 
 Right.
 
 Thanks,
 
 Phil.
 
 
 On Jun 19, 2015, at 7:47 PM, Philip Rhoades via spamdyke-users
 spamdyke-users@spamdyke.org wrote:
 People,
 As well as using GreyLite I have done my own thing for many years
 with qmail-qfilter and a Ruby script (it started off as a Ruby
 learning exercise . . ) - anyway for my white and black lists I was
 able to have in the plain text files things like:
 ad...@phillipsfinancial.com.au
 administrator@(booksjournals.com(|.au)|(prix.|)pricom.com.au|qps.com.au)
 adwords-noreply
 america.com
 ecolife
 where if any of those particular regexes appeared in the To: or
 From: or whatever, they could be allowed or blocked or whatever - I
 am guessing that eg the recipient-blacklist-file=FILE only allows
 for full email addresses?
 Thanks,
 Phil.
 --
 Philip Rhoades
 PO Box 896
 Cowra NSW 2794
 Australia
 E-mail: p...@pricom.com.au
 ___
 spamdyke-users mailing list
 spamdyke-users@spamdyke.org
 http://www.spamdyke.org/mailman/listinfo/spamdyke-users
 Links:
 --
 [1] http://example.com
 [2] http://www.spamdyke.org/documentation/README.html#REJECTING_RECIPIENTS
 [3] http://www.spamdyke.org/documentation/README.html#HEADERS
 ___
 spamdyke-users mailing list
 spamdyke-users@spamdyke.org
 http://www.spamdyke.org/mailman/listinfo/spamdyke-users
 
 -- 
 Philip Rhoades
 
 PO Box 896
 Cowra  NSW  2794
 Australia
 E-mail:  p...@pricom.com.au
 ___
 spamdyke-users mailing list
 spamdyke-users@spamdyke.org
 http://www.spamdyke.org/mailman/listinfo/spamdyke-users

___
spamdyke-users mailing list
spamdyke-users@spamdyke.org
http://www.spamdyke.org/mailman/listinfo/spamdyke-users


[spamdyke-users] Softlimit messages

2015-06-20 Thread Philip Rhoades via spamdyke-users

People,

I played around with the logging verbosity and found if I used debug 
mode I saw suggestions (commands!?) in the log about remove the 
softlimit function from the start script for qmail-smtpd - while I was 
trying to sort out the last bug that was preventing eQmail from working, 
I did actually do that - is the softlimit function even necessary these 
days on a lightly loaded server with 8GB RAM?


Thanks,

Phil.
--
Philip Rhoades

PO Box 896
Cowra  NSW  2794
Australia
E-mail:  p...@pricom.com.au
___
spamdyke-users mailing list
spamdyke-users@spamdyke.org
http://www.spamdyke.org/mailman/listinfo/spamdyke-users


Re: [spamdyke-users] recipient-blacklist-file=FILE with RegExes?

2015-06-20 Thread Philip Rhoades via spamdyke-users

Sam,


On 2015-06-21 03:12, Sam Clippinger via spamdyke-users wrote:

Regex support is on the (rather lengthy) to-do list, but frankly it's
not a very high priority -- there's a lot of low-hanging fruit that
would be of much more benefit right now. Plus, since I'm not one of
the 10 people in the world who completely understands regexes, I doubt
I would actually use them myself; I'd rather add globbing support,
which I do understand. :)



OK, no worries - SD is going well so far so I may not need some of the 
mechanisms that I used in my own setup - we'll see how things go.




spamdyke's header filter runs at connection time, as all of its
filters do. If a header line matches a blacklisted pattern, the entire
message is rejected (the sending server receives an error code, qmail
never sees the message).



Right - thanks for the clarification.

Regards,

Phil.



-- Sam Clippinger

On Jun 19, 2015, at 9:09 PM, Philip Rhoades via spamdyke-users
spamdyke-users@spamdyke.org wrote:


Sam,

See inline comments:

On 2015-06-20 11:53, Sam Clippinger via spamdyke-users wrote:


You're correct spamdyke does not support regexes for any of its
options, but you can use a wildcard in a sender or recipient
white/blacklist file to match entire domains by prefixing the line
with an @ symbol. For example:
@example.com [1] [1]


Yep, saw that - is it possible to support regexes in the future?


Full documentation here:




http://www.spamdyke.org/documentation/README.html#REJECTING_RECIPIENTS

[2]
[2]
BUT! Be careful -- the To and From lines in the message header
are
not the same as the sender and recipient. The sender and
recipient
are part of SMTP, the To and From lines are part of the message
data
and are completely unrelated. Think of it this way: when a letter
is
sent through the post office, the name on the outside of the
envelope
tells the postman which mailbox gets the envelope (or where to
send it
back to) but top of the letter inside may have a completely
unrelated
letterhead and salutation. Whenever spamdyke's
options/documentation
refer to a sender or a recipient, it means the name on the
outside
of the envelope. The user never sees those values in their mail
client
unless the sender chooses to use those values in the To and From
fields. Spammers typically fake all sender/recipient/To/From
fields,
but other software does too for perfectly legitimate reasons (e.g.
mailing lists, autoresponders).


Right.


If you want to block based on the To
and From lines the user sees in their mail client, you should look
at
spamdyke's header blacklist filter:
http://www.spamdyke.org/documentation/README.html#HEADERS [3] [3]


In that case the mail has already been accepted? When I was using
the qmail-qfilter+Ruby script method - my understanding of it at
least - was that my Ruby script could process the header and body of
the email and exit with a particular error code if the mail was bad
and this would terminate the SMTP negotiation with that error
message (eg drop the mail silently). So in this case I was able to
look at all the header fields as well as the mail body and do
whatever I wanted before accepting the mail.


Header filtering doesn't support regexes either, but it does use
globbing to allow more wildcard options.


Right.

Thanks,

Phil.

On Jun 19, 2015, at 7:47 PM, Philip Rhoades via spamdyke-users
spamdyke-users@spamdyke.org wrote:
People,
As well as using GreyLite I have done my own thing for many years
with qmail-qfilter and a Ruby script (it started off as a Ruby
learning exercise . . ) - anyway for my white and black lists I was
able to have in the plain text files things like:
ad...@phillipsfinancial.com.au
administrator@(booksjournals.com [4](|.au)|(prix.|)pricom.com.au
[5]|qps.com.au [6])
adwords-noreply
america.com [7]
ecolife
where if any of those particular regexes appeared in the To: or
From: or whatever, they could be allowed or blocked or whatever - I
am guessing that eg the recipient-blacklist-file=FILE only allows
for full email addresses?
Thanks,
Phil.
--
Philip Rhoades
PO Box 896
Cowra NSW 2794
Australia
E-mail: p...@pricom.com.au
___
spamdyke-users mailing list
spamdyke-users@spamdyke.org
http://www.spamdyke.org/mailman/listinfo/spamdyke-users
Links:
--
[1] http://example.com [1]
[2]


http://www.spamdyke.org/documentation/README.html#REJECTING_RECIPIENTS

[2]
[3] http://www.spamdyke.org/documentation/README.html#HEADERS [3]
___
spamdyke-users mailing list
spamdyke-users@spamdyke.org
http://www.spamdyke.org/mailman/listinfo/spamdyke-users


--
Philip Rhoades

PO Box 896
Cowra NSW 2794
Australia
E-mail: p...@pricom.com.au
___
spamdyke-users mailing list
spamdyke-users@spamdyke.org
http://www.spamdyke.org/mailman/listinfo/spamdyke-users



Links:
--
[1] http://example.com
[2] 
http://www.spamdyke.org/documentation/README.html#REJECTING_RECIPIENTS

[3] 

Re: [spamdyke-users] Softlimit messages

2015-06-20 Thread Sam Clippinger via spamdyke-users
IMHO, everyone should delete the softlimit program from their servers 
immediately.  Not that I have a strong opinion on the matter or anything. :)

The softlimit program seems like a good idea -- set an upper limit on the 
amount of RAM a program can use, to guard against memory leaks (but not buffer 
overflows).  In practice however, it causes far far more problems than it 
causes.  When a program hits the limit, it always happens inside a library 
function and not the application itself.  So the user sees strange errors from 
glibc or OpenSSL functions that are never related to memory allocation.  Those 
errors always look like real bugs, because there's never any indication the 
limit was hit.

There's also no way to even estimate how much memory is correct.  Does anyone 
really understand how many libraries a program loads and how much memory they 
need?  spamdyke uses OpenSSL and on some systems, separate libraries for math 
and DNS functions.  Unpatched qmail doesn't use many libraries, but if patches 
have been applied to allow TLS or authentication, it may use many (who uses 
unpatched qmail anyway?).  If vpopmail is in use, it may need MySQL, depending 
on how it was compiled.  If the server is configured to use stack guarding or 
memory profiling, the virtual memory use could be astronomical.  Every guide 
I've ever read says to use trial-and-error to find the lowest value that 
appears to work, then double (or triple) it.  Crazy!

I've spent way way too much time trying to track down bugs that were caused 
by softlimit and I finally reached my own limit this year.  That's why spamdyke 
5.0.1 examines the limits it starts with and, if it can, resets them.  It can't 
undo hard limits set by the ulimit program, but it can (and does) undo 
softlimit.

-- Sam Clippinger




On Jun 20, 2015, at 2:05 PM, Philip Rhoades via spamdyke-users 
spamdyke-users@spamdyke.org wrote:

 People,
 
 I played around with the logging verbosity and found if I used debug mode I 
 saw suggestions (commands!?) in the log about remove the softlimit function 
 from the start script for qmail-smtpd - while I was trying to sort out the 
 last bug that was preventing eQmail from working, I did actually do that - is 
 the softlimit function even necessary these days on a lightly loaded server 
 with 8GB RAM?
 
 Thanks,
 
 Phil.
 -- 
 Philip Rhoades
 
 PO Box 896
 Cowra  NSW  2794
 Australia
 E-mail:  p...@pricom.com.au
 ___
 spamdyke-users mailing list
 spamdyke-users@spamdyke.org
 http://www.spamdyke.org/mailman/listinfo/spamdyke-users

___
spamdyke-users mailing list
spamdyke-users@spamdyke.org
http://www.spamdyke.org/mailman/listinfo/spamdyke-users


Re: [spamdyke-users] Softlimit messages

2015-06-20 Thread BC via spamdyke-users



Wow. So for example, the starting linefor my smtpd-run file looks like 
this:


exec /usr/local/bin/softlimit -m 2 /usr/local/bin/tcpserver 
-4v -R -l $LOCAL \


and I can simply change it to this:

exec /usr/local/bin/tcpserver -4v -R -l $LOCAL \

with impunity?



On 6/20/2015 5:12 PM, Sam Clippinger via spamdyke-users wrote:
IMHO, everyone should delete the softlimit program from their 
servers immediately.  Not that I have a strong opinion on the matter 
or anything. :)



___
spamdyke-users mailing list
spamdyke-users@spamdyke.org
http://www.spamdyke.org/mailman/listinfo/spamdyke-users