making a backup of my gmail account with fetchmail and procmail

Gentle introduction to email terminology

In this post I’ll be talking about procmail, fetchmail and mutt. Three programs to read and backup emails … Wait, three programs just to read emails !? None sens !
Well, actually it makes perfect sens: to read an email on your computer with your favorite email client, you need to follow these steps:

  1. grab the email from your email provider (connect and download)
  2. filter/process is the email (in which mailbox do we store it – possibly the spam mailbox, or your regular inbox)
  3. read it

For each of these steps, the following jargon is used:

  • when you download your email, you are using a mail transfer agent i.e. MTA
  • when filtering/processing it, you are using a mail delivery agent i.e. MDA
  • and finally when reading it, you are using a mail user agent i.e. MUA

Weird ? Not really if you know the UNIX philosophy: each program does one
thing, but it does it well (Yeah, I know: sounds like a real cliche).

When you use thunderbird or outlook, you are using a MTA, MDA and MUA at the
same time. I will not talk about mail submission agent
MSA that is in charge of sending the email you wrote: this is out of scope :-).

Ok, now let’s go back to the purpose of this post: backing up my gmail account.

the big plan

So, given the previous steps described above, I will:

  • use fetchmail as my MTA to connect to gmail and retrieve my emails
  • use procmail as my MDA to store the content in the various mailboxes
  • read it on mutt, yes mutt is my MUA

step 1: retrieving mail with fetchmail

configuring my gmail account

google provides access to my email using POP3 or IMAP.

I will retrieve my email using POP3, google provides interesting features for it. The one I am interested in is located at settings -> forward and pop/imap -> POP download.
Here I select “Enable POP for all mail (even mail that’s already been downloaded)”: it will allow me to download all my emails, including the ones I already did read.
WARNING: be sure to select “When messages are accessed with POP – keep gmail’s copy in inbox” otherwise, gmail will remove your downloaded email
from your account – which is not my intent.

I also select, in the same page, enabling IMAP, so I can read my emails from everywhere using the IMAP protocol which allows me to view my emails from
anywhere while keeping them on the server i.e. I browse them, when I finish reading my mail, it’s still there on the server: be sure to check IMAP wikipedia entry.

I download them only for backup purpose and use POP3 for that.

Do not forget to save the changes you made to your google account …

dealing with gmail server security

Access to my account is encrypted (a good thing) hence I need to setup some security before beeing able to download some emails.
SSL as used by google, requires you to download some credentials from a certificate authority; it is used to authenticate google server i.e. making sure you really are connecting to google and not to some random (and possibly malicious) server.

First, I need to retrieve the certificate of the CA used by google:

# wget -O Equifax_Secure_Certificate_Authority.pem https://www.geotrust.com/resources/root_certificates/certificates/Equifax_Secure_Certificate_Authority.cer

and convert it to something that can actually be used by our beloved ssl tools:

# openssl x509 -in Equifax_Secure_Certificate_Authority.pem -fingerprint -subject -issuer -serial -hash -noout

I now store it where they can latter be referenced and checked. Since I am always setting up workstations when moving from one workplace to another, I store that kind of information in the UNIX account that has been given to me:

# mkdir ~/.certificates
# mv Equifax_Secure_Certificate_Authority.pem ~/.certificates

Now I make them available to the various programs I use:

# c_rehash ~/.certificates

On opensolaris c_rehash is not available (I don’t know why). It’s a perl script and I found a version using google code here.

cut and paste this code into a file named c_rehash and execute it:

# cd ~
# perl c_rehash .certificates
Doing .certificates/
Equifax_Secure_Certificate_Authority.pem => 594f1775.0

This is it for the security part, SSL credentials are now ready to be used by fetchmail.

configuring fetchmail for POP access

my gmail account is something like:

  • user ‘vodoom@gmail.com’
  • pop3 server is ‘pop.gmail.com’

my local settings are:

  • user on my machine is ‘ppereira’
  • my google certificate is located at ‘~/.certificates’ (see previous step)
  • mda to be used is ‘procmail’ (more on that later)

The corresponding fetchmail configuration (to be stored in ~/.fetchmailrc):

set daemon 600
poll pop.gmail.com with proto POP3 and options no dns
user 'vodoom@gmail.com' there is 'ppi' here
options keep ssl sslcertck sslcertpath '~/.certificates'
mda '/usr/bin/procmail -d %T'

For more details about these options, see fetchmail’s manual.

configuring procmail for delivery

spam filtering ?

Note that there is no spam filtering configuration here: since this setup is only for my gmail account, I rely on gmail spam filter to take care of those.
If you need a spam filtering setup, you may want to have a look at http://wiki.apache.org/spamassassin/UsedViaProcmail.

managing my mailing lists

I use the TO_ recipe rule of procmail: I do not use the list-id header, it is more covenient for me as explained in http://www.ii.com/internet/robots/procmail/qs/#alt2TO.
Procmail recipes are matching rules associated with a mailbox (yes, I am over simplifying).
For example, here is a simple recipe:

:0:
* ^TO_zfs-discuss@opensolaris.org
osol-zfs

:0: is the delivery options, in my case I want locking when accessing to the inbox (that’s the second ‘:’)

* ^TO_zfs-discuss@opensolaris.org
means that we want to match messages that were sent to zfs-discuss@opensolaris.org
One can use regular expressions and special matching rules e.g. ^From to create a matching rule.

osol-zfs tells procmail to store this email in the mailbox ‘osol-zfs’

here are my recipes to match the mailing lists I am subscribed to:

:0:
* ^TO_zfs-discuss@opensolaris.org
osol-zfs

:0:
* ^TO_perf-discuss@opensolaris.org
osol-perf

:0:
* ^TO_opensolaris-code@opensolaris.org
osol-code

:0:
* ^TO_opensolaris-announce@opensolaris.org
osol-announce

:0:
* ^TO_indiana-discuss@opensolaris.org
osol-indiana

:0:
* ^TO_ogb-discuss@opensolaris.org
osol-board

:0:
* ^TO_eeepc-discuss@opensolaris.org
osol-eeepc

:0:
* ^TO_crypto-discuss@opensolaris.org
osol-crypto

:0:
* ^TO_xvid-devel@xvid.org
xvid-devel

:0:
* ^TO_users@crater.dragonflybsd.org
dragonfly-bsd

All those recipes are stored in their own file: rc.mailing-lists.
This file will be included from a ‘master’ file, the main .procmailrc file.
The .procmailrc file looks like:

# Directory for storing procmail configuration and log files
# You can name this variable anything you like, for example
# PROCMAILDIR, or don't set it (but then don't refer to it!)
PROCMAIL_DIR=$HOME/procmail
SHELL=/bin/sh

# LOGFILE should be specified ASAP so everything below it is logged
LOGFILE=$PROCMAIL_DIR/procmail.log

# where are the various mailboxes stored ?
# note that this is *not* the location of the system email
# this is where I want my mailboxes to be stored by procmail i.e. the
# destination after filtering
MAILDIR=$HOME/mail

# recipes/filters are included from here
INCLUDERC=$PROCMAIL_DIR/rc.mailing-lists
INCLUDERC=$PROCMAIL_DIR/rc.banking

Now that procmail is setup, time to retrieve some emails:

# fetchmail -vk
File /export/home/ppi/.fetchmailrc must have no more than -rwx--x--- (0710) permissions.

oops I forgot to setup the proper permissions for the .fetchmailrc file,
let’s fix that:

# chmod 710 .fetchmailrc

Running it again, and failing again for a very different reason, from my logfile in ~/fetchmail/fetchmail.log:

fetchmail: starting fetchmail 6.3.8 daemon
fetchmail: 6.3.8 querying pop.gmail.com (protocol POP3) at Thu Jan 28 00:06:43 2010: poll started
fetchmail: getaddrinfo("pop.gmail.com","pop3s") error: service name not available for the specified socket type
fetchmail: Try adding the --service option (see also FAQ item R12).
fetchmail: POP3 connection to pop.gmail.com failed: Bad file number
fetchmail: 6.3.8 querying pop.gmail.com (protocol POP3) at Thu Jan 28 00:06:43 2010: poll completed
fetchmail: Query status=2 (SOCKET)
fetchmail: sleeping at Thu Jan 28 00:06:43 2010 for 600 seconds

hum, fetchmail is looking for the port number corresponding to the ‘pop3s’ service. AFAIK, on my opensolaris box I don’t have such service configured:

# grep -i pop3s /etc/services
#

To connect to gmail I have to manually setup use the port number
corresponding to ‘pop3s’, as suggested by fetchmail in the log:

'fetchmail: Try adding the --service option (see also FAQ item R12).'

so here I go again:

# fetchmail -vk --service 995

now from my fetchmail log I can see:

fetchmail: 6.3.8 querying pop.gmail.com (protocol POP3) at Fri Jan 29 00:15:07 2010: poll started
fetchmail: Trying to connect to 74.125.93.111/995...connected.
fetchmail: Issuer Organization: Google Inc
fetchmail: Issuer CommonName: Google Internet Authority
fetchmail: Server CommonName: pop.gmail.com
fetchmail: pop.gmail.com key fingerprint: 92:73:17:4C:34:4B:68:F7:B2:17:71:42:0D:7F:9F:33
fetchmail: POP3< +OK Gpop ready for requests from 70.31.248.56 8pf4635265qwj.1
fetchmail: POP3> CAPA
fetchmail: POP3< +OK Capability list follows
fetchmail: POP3< USER
fetchmail: POP3< RESP-CODES
fetchmail: POP3< EXPIRE 0
fetchmail: POP3< LOGIN-DELAY 300
fetchmail: POP3< X-GOOGLE-VERHOEVEN
fetchmail: POP3< UIDL
fetchmail: POP3< .
fetchmail: POP3> USER vodoom@gmail.com
fetchmail: POP3< +OK send PASS
fetchmail: POP3> PASS *
fetchmail: POP3< +OK Welcome.
fetchmail: POP3> STAT
fetchmail: POP3< +OK 354 21544383
fetchmail: POP3> LAST
fetchmail: POP3< -ERR Not supported
fetchmail: Not supported
fetchmail: POP3> UIDL
fetchmail: POP3< +OK
fetchmail: POP3< 1 GmailIdfbe952ed20bfe97
fetchmail: POP3< 2 GmailIdfc26b09f74e3775
fetchmail: POP3< 3 GmailIdfc6f980243d2a13
fetchmail: POP3< 4 GmailId1043d10888f8ed70
...
fetchmail: POP3< 354 GmailId10e7b3433c301f94
fetchmail: POP3< .
fetchmail: 354 messages for vodoom@gmail.com at pop.gmail.com (21544383 octets).
fetchmail: POP3> LIST 1
fetchmail: POP3< +OK 1 1694617
fetchmail: POP3> RETR 1
fetchmail: POP3< +OK message follows
fetchmail: reading message vodoom@gmail.com@pop.gmail.com:1 of 354 (1694617 octets)
...

Yes ! This is working :-)

So now my gmail account is backed up. From time to time I run my fetchmail script, and it downloads the new emails that are not yet backed up.
If like me you have thousands of emails waiting to be download, be patient, as emails are not downloaded in a single connection: each connection will download hundreds of emails, fetchmail will have to reconnect to download the rest. This is fine, that’s why I setup the daemon mode in fetchmail :-).

Ok now I have to access my emails with IMAP, but thats another post :-)

enjoy and share.

Advertisements

About this entry