Month: December 2010

Disaster recovery with MondoRescue

 

Important
This is an edited version of a post that originally appeared on a blog called The Michigan Telephone Blog, which was written by a friend before he decided to stop blogging. It is reposted with his permission. Comments dated before the year 2013 were originally posted to his blog.
The Great Desktop Fire
Image by mattbraga via Flickr

Many of us face the problem of having a server that we know we should backup frequently, but we don’t do it because it’s either too difficult to figure out how, or the backup solutions offered don’t actually restore the entire system if it crashes, so we figure, “why bother?”  If your system crashes, the thing you really need is a way to restore the entire system from some recent point in time.

Well, here’s one possible solution for you, assuming your server runs some form of Linux, and it’s from the fine folks at Sunshine Networks in Brisbane, Australia. I refer you to their article:

Disaster Recovery with Elastix 2.0

Now, don’t let the title throw you – there’s nothing Elastix-specific in this article.  The instructions should work with just about anything running under the CentOS operating system, and with minor tweaks to the installation process, under other versions of Linux.  What this software is supposed to do is give you an ISO file that can be burned to CD’s or DVD’s, or stored on a network share on another machine.  If the worst happens, you fix the hardware problems and then reinstall from the ISO file, and the way it’s supposed to work is that you get back to exactly where you were at the time of the last backup.  Now, I haven’t personally ever had to attempt a restore, but apparently others have and consider this a great piece of software. Obviously, I’m not making any guarantees, but it’s got to be better than no backup at all, right?

EDIT: Since I originally wrote this article, I’ve actually had the opportunity to use MondoRescue to restore a failed system (in this particular case, one that runs on a virtual machine). To say it worked great is an understatement. You just boot from the .iso file and it installs EVERYTHING back as it was. The only issue I had was that it couldn’t communicate with the network because the name of the network adapter was apparently different on the original and new systems — once I reconfigured the network settings to select a valid adapter (eth0, for example) it appeared to work just as it had on the day of the backup. And the restore process was surprisingly fast (much faster than the original installation, in fact)! Of course I cannot guarantee it will work that well for you, but I was blown away by the speed of the restoration, and I’m not that easily impressed!

I must also note that the article on the original Sunshine Networks site seems to have disappeared, so I changed the link to point to an archived copy on the Wayback Machine. However, in case that fails at some point, here is how I installed MondoRescue. Their instructions gave three different ways to do it, and I used this one, which (with perhaps a change in the file used) should work on any Red Hat or Centos based system (this was noted as “Tested on Elastix 2.0 32-bit” — if you are running something else, don’t just follow these instructions because you may need a different file):

cd /root/
wget http://packages.sw.be/rpmforge-release/rpmforge-release-0.5.1-1.el5.rf.i386.rpm
rpm -Uhv rpmforge-release-0.5.1-1.el5.rf.i386.rpm
yum install mondo

after mondo installed correctly, you should disable the RPMForge repository, just to be on the safe side :
nano /etc/yum.repos.d/rpmforge.repo
change “enabled = 1” to “enabled = 0”

(They used vi to edit the repository; I changed it to nano. Use whichever text editor you like).

However, the file shown here is probably NOT the right one for your system. So, first go to http://packages.sw.be/rpmforge-release/ and read the descriptions for each file, and be careful to select the right one for your system, and substitute that filename in the two lines where it is used above.

After installation, you can start the program by running /usr/sbin/mondoarchive, which will bring up a GUI (of sorts). The original article notes that:

your full iso will ( under default settings ) be created in the following directory :
/var/cache/mondo/mondorescue-1.iso
there is a small recovery CD here :
/var/cache/mindi/mondorescue.iso

END OF EDIT.

The article has you use the mondoarchive GUI to make the backups (well, they actually say mondorescue, but when I downloaded the software the program was called mondoarchive), and that’s fine to start with.  But eventually, you’re going to want to automate the process so you can use it in a cron job to do unattended scheduled backups on a regular basis.  I have this running on one machine and send copies of the backups to another, like this (cut and paste from this article to get the full lines without wrapping) :

#!/bin/bash
mondoarchive -OVi -d "/var/cache/mondo" -E "/asterisk_backup" -N -9 -G -s 4G
ssh myaccount@server2.net rm /home/myaccount/server1backup/mondo/mondorescue-1-old.iso
ssh myaccount@server2.net mv /home/myaccount/server1backup/mondo/mondorescue-1.iso /home/myaccount/server1backup/mondo/mondorescue-1-old.iso
scp /var/cache/mondo/mondorescue-1.iso myaccount@server2.net:~/server1backup/mondo
ssh myaccount@server2.net rm /home/myaccount/server1backup/mindi/mondorescue-old.iso
ssh myaccount@server2.net mv /home/myaccount/server1backup/mindi/mondorescue.iso /home/myaccount/server1backup/mindi/mondorescue-old.iso
scp /var/cache/mindi/mondorescue.iso myaccount@server2.net:~/server1backup/mindi

The first line calls the mondoarchive program to create the backup – the -E argument excludes any directories you don’t wish to back up (I have a directory of backups made using another method that I didn’t want backed up) and you can read about the other arguments in the documentation (also see the full HOWTO).  The remaining lines connect to the external server and delete the oldest backups, rename the previous backup, and then copy the new backups over.  To do it the way I’ve done it here, you must have ssh access to the other server and you must be able to connect without using a password, using public/private key authentication.  You may also have to log into the remote server and create the directories (/home/myaccount/server1backup/mindi/ and /home/myaccount/server1backup/mindi/ in this example – obviously you can call the directories whatever you wish, it’s entirely up to you).

There is, of course, more than one way to remove the pelt from a deceased feline, and you’ll probably have your own method for moving the files to another server.  In some situations it appears that MondoRescue could do it for you (look at the n option), but it doesn’t include a provision to remove the oldest file and rename the previous one (not that I could see, anyway), so that’s why I did it in a shell script.

The folks at Sunshine Networks have several other great how-tos – you might want to give them a look! And for more useful information on MondoRescue, particularly how to perform a restore, see Configure IT Quick: Use Mondo Rescue to back up Linux servers (but please realize that article was written in 2003, and the install has apparently been made less complicated since then, so don’t use their installation instructions).

Related Articles:
How to Clone/Backup Linux Systems Using – Mondo Rescue Disaster Recovery Tool (TecMint.com)
Redo Backup and Recovery Tool to Backup and Restore Linux Systems (TecMint.com)

Asterisk 1.8.x and FreePBX users: How to NOT answer Google Voice calls UNTIL the called extension answers

 

Important
This is an edited version of a post that originally appeared on a blog called The Michigan Telephone Blog, which was written by a friend before he decided to stop blogging. It is reposted with his permission. Comments dated before the year 2013 were originally posted to his blog.

EDIT (May, 2018): FreePBX and Asterisk users that wish to continue using Google Voice after Google drops XMPP support should go here: How to use Google Voice with FreePBX and Asterisk without using XMPP or buying new hardware.

This article was originally published in December, 2010 and may contain out-of-date information.

Many folks are experimenting with Asterisk 1.8.x and Google Voice.  In most cases the way it’s set up is that when a Google Voice call arrives, Asterisk answers the call, then sends a touch-tone digit “1” to Google Voice to answer the call, then proceeds to ring the destination extension.  This is necessary because when you configure Google Voice to use a Gtalk destination, they require you to press “1” to accept the call, even if you’ve configured Google Voice not to require that.  I don’t know if this is a bug in Google Voice or if they did it that way deliberately for some reason, but answering the call and accepting it upon arrival at the PBX has a few unintended side effects:

  • If your callers pay for long distance by the minute, they get charged from the moment the called extension begins ringing – even if you never answer the call.
  • You can’t use Google Voice’s Voicemail, nor their transcription service, because you’ve already answered the call.
  • Callers may hear a confusing double ringing tone at the start of ringing — one ring from Google Voice and the rest from Asterisk.

On the other hand, there are some advantages to doing it that way:

  • Because you’ve answered the call, you can let the extension ring as long as you like before sending it to voicemail, and Google Voice won’t snatch it away in 25 seconds and send it to their voicemail.
  • You can use Asterisk’s voicemail, if that’s what you prefer.

For those who’d prefer to let Google Voice handle their voicemail, or who object to making callers pay to listen to up to 25 seconds of ringing, there is a way to not answer the call and send the touch tone “1” until  after the destination extension has actually picked up the call.  If you are using plain vanilla Asterisk, all you have to do is make sure your Dial() command contains two additional options.  Consider this example line of Asterisk dialplan:

exten => gvoicein,n,Dial(SIP/1004,35,rTWtwaD(:1))

The important part here is the aD(:1) — the other options can be whatever you’d normally use, if any, but it’s the aD(:1) that does the magic. Now at this point, if you’re a FreePBX user you may be wondering how on earth you can modify the Dial() string, since the code that generates it is buried deep within the bowels of FreePBX. Fortunately, there is a way. Consider the following piece of code that might be used in extensions_custom.conf to bring in Google Voice calls:

[googlein]
exten => _[0-9a-z].,1,Noop(Incoming Google Voice call for ${EXTEN})
exten => _[0-9a-z].,n,Set(CALLERID(name)=${CUT(CALLERID(name),@,1)})
exten => _[0-9a-z].,n,GotoIf($["${CALLERID(name):0:2}" != "+1"]?notrim)
exten => _[0-9a-z].,n,Set(CALLERID(name)=${CALLERID(name):2})
exten => _[0-9a-z].,n(notrim),Set(CALLERID(number)=${CALLERID(name)})
exten => _[0-9a-z].,n,Wait(1)
exten => _[0-9a-z].,n,Answer
exten => _[0-9a-z].,n,Wait(1)
exten => _[0-9a-z].,n,SendDTMF(1)
exten => _[0-9a-z].,n,Goto(from-trunk,gv-incoming-${CUT(EXTEN,@,1)},1)
exten => h,1,Macro(hangupcall,)

With this context you’d use gv-incoming-username (where username is the part of the associated gmail address before the @) as the DID in your inbound route — a DID doesn’t have to be numeric even if FreePBX whines about it, and the advantage is you only need one context to handle incoming calls for all your Google Voice accounts.  This particular context is slightly modified from one found in the PBX in a Flash forum, but note that it contains these four lines that wait ONE second, answer the call, wait ONE second (you do NOT have to wait two seconds, despite what any other article may say, and in fact the one second wait might be unnecessary), and then send the touch tone digit 1:

exten => _[0-9a-z].,n,Wait(1)
exten => _[0-9a-z].,n,Answer
exten => _[0-9a-z].,n,Wait(1)
exten => _[0-9a-z].,n,SendDTMF(1)

You will find those four lines, or some variation on them (sometimes just the last three), in just about every published method for using Google Voice with Asterisk and FreePBX.  But, in FreePBX at least, you can replace them with this:

exten => _[0-9a-z].,n,Set(DIAL_OPTIONS=${DIAL_OPTIONS}aD(:1))

This slides the aD(:1) into the options that will be used with the Dial command, so when the extension answers, the call will be answered and then the touch tone “1” will be immediately sent to Google Voice, and then the audio between Google Voice and the called extension will be bridged as usual.

Unfortunately, or maybe fortunately depending on your point of view, it appears that if the call should go to Asterisk’s voicemail, the call will not be answered and the DTMF 1 will never be sent.  This means that if, for whatever reason, you don’t answer the incoming call, after 25 seconds it will go to Google’s voicemail.  There are doubtless ways around that (and if anyone’s truly interested, leave a comment and I’ll suggest a way that may work, that involves routing the incoming call to a ring group first) but I suspect that the majority of people who want to do this will be doing it because they want to use Gmail’s voicemail.

I’ve tested this and it works for me, though I would not use it on a regular basis because I prefer Asterisk’s voicemail.  If it doesn’t work for you for some reason, the only suggestion I can offer is adding a w before the :1, so the added options look like aD(w:1) – that will add a one-half second delay before the “1” is sent, and more than likely it won’t help one bit, but may cause callers to not hear your “hello” or other greeting.  But, you can try it and see — at least one user has reported it to be necessary.  If that doesn’t work, I probably won’t be able to help you but if you leave a comment, maybe someone else can.

And, should anyone from Google Voice read this, it would be really helpful if you’d do two things:

  1. Give us a way to disable Google Voice’s voicemail so we don’t have to resort to hacks like this to discourage callers from leaving a message there.
  2. Fix the bug (or “feature”) so that when we turn off call screening, it’s off for ALL destinations, including Gtalk!