Re: [Mailman-Users] Mailman redundancy
> Ok, so you then add the "domain" that is used for the mailing > list to the sendmail.cw file or it's equivalent (domains I > accept mail for). > > Would the secondary/third servers then then try to deliver > it locally? That is where I get confused :-) Yes. The sendmail.cw file specifies which hosts should be treated the same as localhost, so it would try to deliver. The backup servers should not specify each domain in sendmail.cw; rather, RELAY_DOMAIN_FILE should be used in your .mc file and populated with domain names to queue. If the MX records are right, it will only queue until the primary becomes available. Chris -- Mailman-Users maillist - [EMAIL PROTECTED] http://mail.python.org/mailman/listinfo/mailman-users
Re: [Mailman-Users] RFC or algorithm for creation of Message-ID
> Does anyone on the list know of an RFC or, better yet, an "official" > algorithm to use for the creation of a message's Message-ID: header value? > I'm gating web forums to Mailman lists and need to generate Message-IDs for > messages created via the web interface. I need to insert the ID in a forum > DB field and deposit it into the outgoing message. Hi Kevin, Take a look at RFC 822 -- it describes the Message-ID: header. If I remember correctly, it's in the form of @ and should be unique. Most sites choose to do @example.com (replacing example.com with their domain name). The unique identifier is usually generated based on the current time and process id, i.e. time() . $$ in perl. It can be left as base 10, converted to base 16, run through md5sum to create a 32-bit number, etc... Chris -- Mailman-Users maillist - [EMAIL PROTECTED] http://mail.python.org/mailman/listinfo/mailman-users
Re: [Mailman-Users] demime.pl with smrsh
On Feb 28, Clark E. Morgan wrote: > foo: "|/home/mailman/mail/demime.pl - |/home/mailman/mail/wrapper post foo" > >- Transcript of session follows - > sh: cannot use | in command > 554 "|/home/mailman/mail/demime.pl - |/home/mailman/mail/wrapper post > shaka"... Service unavailable > > So I'm thinking it's the pipe into a pipe that is creating the problem. > I'm running smrsh and am not sure if the limitation is being imposed there > or not. I do have the approp sym links for both the mailman wrapper and > demime.pl in sm.bin. Has anyone seen this and can recommend a viable > solution? Yes, that's a smrsh limitation. The only real solution is to create a script that takes input and pipes it to wrapper, then make that script runable by smrsh. Chris -- Mailman-Users maillist - [EMAIL PROTECTED] http://mail.python.org/mailman/listinfo/mailman-users
Re: [Mailman-Users] Next: newlist admin passwords not working?
> I ran into some problems a while back where the Web server didn't have > the Python crypt module installed, but the mail server did. The upshot > was that Mailman used crypt() to create passwords on the mail server, > but then couldn't authenticate against them on the Web server. :) Actually, that's not quite right. Sorry. :) But it might still be applicable. The problem that I had was that both servers were running RedHat 4.2, which had python sans the crypt module. Even though USE_CRYPT was set to 1 in mm_cfg.py, it wouldn't work since the module wasn't there. Once I upgraded one of the servers to RedHat 6.2 (which did have the crypt module) Mailman started trying to use crypt() against passwords that had previously been generated without (because USE_CRYPT was set). Changing USE_CRYPT to 0 solved the problem (and I personally feel that this should be the default given the problems that this can cause). Chris -- Mailman-Users maillist - [EMAIL PROTECTED] http://mail.python.org/mailman/listinfo/mailman-users
Re: [Mailman-Users] Next: newlist admin passwords not working?
> When I get my email that the list has been created, this is the password > it shows me. > > But, when I go to the admin page and try to login using that password (I > cut/paste from Emacs to Netscape it so I know it's right), I always get > "Authentication failed". > > I have to log in with the site password and change the list password > that way. Are you running the Web server on the same system as the email services? I ran into some problems a while back where the Web server didn't have the Python crypt module installed, but the mail server did. The upshot was that Mailman used crypt() to create passwords on the mail server, but then couldn't authenticate against them on the Web server. :) What is USE_CRYPT set to in mm_cfg.py (or is it just the default)? Chris -- Mailman-Users maillist - [EMAIL PROTECTED] http://mail.python.org/mailman/listinfo/mailman-users
Re: [Mailman-Users] Message-id?
> When people subscribe, using the web, and mailman sendt the mail with > the secret to prove your identity, it sends it without any Message-Id > line. I process mails using SMTP to my local qmail installation (so > SMTP to localhost) and qmail does not add any Message-Id either. If qmail doesn't add a Message-Id: header it's not reasonable to expect Mailman to insert one either. However, RFC 822 specifically states that a Message-Id: header is *optional*, so Mailman isn't violating anything by not adding one. > The result is, that the mail is delivered to the target users mailhost > without any Message-Id line at all, and some mailsystem is configured > to performs header check and will reject such mails. The systems that reject emails without Message-Id: headers are non-RFC compliant and should be flogged. Sending 10MB attachments with the envelope sender set to [EMAIL PROTECTED] without a Message-Id: is the only appropriate course of action... (no, I'm only kidding about this last sentence). Chris -- Mailman-Users maillist - [EMAIL PROTECTED] http://mail.python.org/mailman/listinfo/mailman-users
Re: [Mailman-Users] how to remove badly formed email address from list
> I've got a user on one of my lists who's subscribed himself with the email > address with a form of "myname@home"@isp.co.uk. > > Of course, this address is completely undeliverable, and every mail out to > the list bounces. So I'd like to remove this guy. But I can't figure out > how! It *is* a valid RFC 822 address. Unfortunately, Mailman doesn't work well with quoted addresses, especially those containing whitespace (it often breaks things up based on white space, so something like "Bobo the emu of luv"@example.com will get split into 5 parts). > If I deselect him from the web interface, mailman complains that the user > doesn't exist. Same if I try it via the mail interface. > > I've tried every combination of quotes and escapes I can think of. I had no problems removing entries like this with $PREFIX/bin/remove_members. Does it give you an error? Chris -- Mailman-Users maillist - [EMAIL PROTECTED] http://mail.python.org/mailman/listinfo/mailman-users
Re: [Mailman-Users] Out-of-office replies
> Unless the vacation program quotes back the message it is reponding
> to, it is difficult for Mailman to distinguish between a vacation
> loop and am active poster. That said, implementing something like
> Formail's MsgID cache for Mailman, except instead of cacheing
> MessageIDs cacheing MD5's of message bodies might be a good idea...
All of my messages have an X-MD5SUM: header added, so doing this with
formail is easy... Just use this instead of the traditional formail -D
line:
formail -RX-MD5SUM: From: -R "" A -rD 8192 idcache
If you don't have an X-MD5SUM: header, you could add one. Here's the
whole blob, including the formail test to see if a message is a duplicate:
:0
* B ?? ()^^\/((.*$)*|.*)
{
MD5SUM=`echo $MATCH | /usr/bin/md5sum`
:0
* MD5SUM ?? ()^^\/
{ MD5SUM = "$MATCH" }
:0 fhw
| formail -I"X-MD5SUM: $MD5SUM"
}
:0 Whc: msgid.lock
| formail -RX-MD5SUM: From: -R "" A -rD 8192 md5sum.cache
# Message-Id: was in the cache, so bounce the message to the admin
:0 a
! [EMAIL PROTECTED]
Chris
--
Mailman-Users maillist - [EMAIL PROTECTED]
http://mail.python.org/mailman/listinfo/mailman-users
Re: [Mailman-Users] PGP for Mailman
> Always wanted to have a mailing-list handler that could optionally use > PGP :) > > * So that users subscribed could (optionally) send in their public PGP > key, and have all emails delivered to them from then on encoded. > > * Ask the server for its public key so mails could be sent to the list > encrypted > > * Optional Header keywords in emails to specify delivery of mail to > users only with PGP keys specified. > > But perhaps I'll just be flamed for such an idea... :) I think it's a good idea... Sympa (www.sympa.org) does this via S/MIME encryption with X509 certificates... Chris -- Mailman-Users maillist - [EMAIL PROTECTED] http://mail.python.org/mailman/listinfo/mailman-users
Re: [Mailman-Users] Unsubscription Script, Try #3
> We are using your script on our live lists with great success. > I'm sure many lists will find this useful and I'm hoping to see it > in one of the future Mailman releases as a built-in feature. A > 'confirmed unsubscription' similar to this but with a 'pending > unsubscriptions' db and random numbers would be the final > step, but we're more than satisfied now. I just looked at the script for the first time, and it looks pretty good. Kudos, Bob! I did have one comment about the use of /tmp for the temporary file... Since /tmp isn't a secured directory, it's possible for someone to exploit the known filename... For example, if I create a symlink from [EMAIL PROTECTED] to a file owned by the user that sendmail runs as, then send mail to your script with a From address of [EMAIL PROTECTED] it'll overwrite the original file. You're probably better off creating a directory like ~mailman/tmp, giving your script permissions to use it, then using that... Of course, this only applies if people have access to your systems, but it's still good practice (heck, you could exploit it via a .procmailrc or a .forward, too). I don't know if anyone mentioned this or not, but the script should probably only be readable by the user that can execute it, otherwise your mmsitepass password could be readable by others too (again, via .procmailrc, .forward, etc)... Chris -- Mailman-Users maillist - [EMAIL PROTECTED] http://mail.python.org/mailman/listinfo/mailman-users
