[Update — fixed the link!]

Automated Backups are a good thing. Automated Backups make the little birds sing, the rainbows shine, and little fauns gambol about in beautiful green forests. When computers are backed up, the butterflies flutter, the flowers bloom, and the fruit from the trees taste just a little sweeter. But when computers are not backed up, the universe becomes angry.

An angry universe is not a good thing. An angry universe makes little birds cry. An angry universe makes Cthulhu come and visit.

So. Automated backups. I’m partial to rdiff-backup because it allows me to not only back up data, but keep previous versions available. Backing up nightly doesn’t help if you accidentally overwrite the contents of a file with something, and don’t notice for a day or two. But with rdiff-backup, you can restore the version before the error.

Unfortunately, rdiff-backup really is designed for server-to-server backups, where each end of the transaction has shell access. Enter duplicity, a related project. It’s more designed for storing backups on servers that you don’t control and/or don’t trust. It allows encryption of your backup sets, as well as supporting a wider variety of protocols (ftp, scp, s3, etc.)

So with a combination of these two scripts, you can backup pretty much any POSIX-ish server to pretty much anything that you can ftp or ssh into. Still, it’d be nice if you could:

• Check that the backups completed successfully, and get email confirming that success or warning on a failure.
• Configure up all of your various backups by a simple text file, rather than remembering the different command-line formats.
• Create groups of options that can be applied to backup tasks.
• Issue commands on the backup source and destinations before and/or after the backup (good for dumping databases into a flat file, for example, and then deleting it after it’s backed up).
• Get email confirmation on completion of backups.
• Have some tools to simplify the securing of the backup process.

For these reasons, I put together this backup script, which is basically a Ruby wrapper for rdiff-backup and duplicity. It’s almost entirely configured via two human-readable yaml files.

It’s flexible, reasonably simple to use, and comes without any guarantees whatsoever. Feel free to use it yourself!

DISCLAIMER: it’s as-is. Not to be used in place of a certified Cthulhu-deterrent. Use at your own risk. To quote the duplicity page: “[it] is not stable yet. It is thought to have a few bugs, but will work for normal usage, and should continue to work fine until you depend on it for your business or to protect important personal data.” — that goes for me too, only double.

OK. Entrepreneurs, read up. I’m gonna give you some ideas that’ll make you rich.

Start my ranting:

1. Can I really be the only person who wants to share Thunderbird/Seamonkey address books with a spouse? I mean, how hard can it be?

What I’d like:

• Each of our “Personal Address Book” collections show up as a list on one another’s address books as a list (e.g., mine shows up on my wife’s machine as “Samuel’s Address Book”. It could use the machine name instead, if it’s easier).
• We can see one another’s mailing lists in our address books
• Manual sync is fine — automatic would be even better
• Simplistic merging is OK, so long as there’s a way to resolve conflicts
• Ability to mark lists as private or shared

2. Can I really be the only person who wants to share a checkbook program with a spouse? I mean, how hard can it be?

What I’d like:

• Ability to enter checks / charges / deposits into a common account register
• Ability for either person to perform reconciliation
• Ability to have accounts that are not shared

3. Can I really be the only person who wants to have an intelligent, revision-capable backup script that doesn’t require shell on the destination end? I mean, how hard can it be?

What I’d like:

• rdiff-backup, only permitting an ftp-based push of the backup file.

More to come, as I experience more outrage.

I’ve been busy setting up a new hosting environment for a bunch of static HTML and PHP-based web sites on a Go Daddy Virtual Server. It was going swimmingly, until I came to an old CMS Made Simple site (running 0.10.x), which merely returned blank pages. Newer versions of CMS Made Simple ran fine. I could find nothing in the virtual host’s web error logs, the php log, the mysql error logs, the eaccelerator error logs, or any other system logs — except in the main Apache error log, there was:

child pid XXXXX exit signal Segmentation fault (11)

Searching around, this looks like it could be some kind of threading issue; however, I’m following the recommendations and using the Apache 2 prefork MPM.

Eventually, the (weak) solution I came up with is to turn off eaccelerator for that virtual host. This remedies the situation, although I can’t say it makes me very happy, since I don’t understand exactly what’s going on (or what the problem is).

I’d welcome insight into this.

Details: Fedora Core 4, Apache/2.0.54 (Fedora), PHP 5.0.4, eAccelerator 0.9.5.

I’m one of the many people who will be using VNC to do remote assistance for a relative using Windows.

There are a number of tutorials out there. Most of them fail because they require the ability to VNC in to the remote system, which won’t work in my case because the remote Windows box is behind a firewall/router that I can’t configure. There are also several reverse approaches out there, where the user needing assistance initiates the connection. The first of these I say was Gina Trapani’s approach at Geek to Live, which uses UltraVNC on both ends. This is almost the solution I want, except that it requires Windows on my end as well. It also assumes that I’m at a fixed location.

In the comments, I came across Fazal Majid’s response. He had the same requirements as I do, and links to his source where he built a customized VNC server that targets a fixed IP address. Fazal’s approach matches my needs exactly.
But then I ran into the problem of the last step: the reverse SSH tunnel from my known server (which gets hard-coded into the executable) to my notebook running Chicken of the VNC.
Building reverse SSH tunnels is really not that difficult. But when I created the setup, I was able to make it work from a Linux machine and from a Cygwin terminal under Windows, but it mysteriously failed under Mac OS. Using lots of -v flags, I kept seeing the service for the port on the Mac side refusing the connection from the tunnel. The ssh debug looked like:

debug1: remote forward success for: listen 5900, connect localhost:5500
debug1: client_input_channel_open: ctype forwarded-tcpip rchan 2 win 131072 max 32768
debug1: client_request_forwarded_tcpip: listen localhost port 5900, originator ::1 port 60475
debug1: channel 0: new [::1]
debug1: confirm forwarded-tcpip
debug3: channel 0: waiting for connection
debug1: channel 0: not connected: Connection refused
debug2: channel 0: zombie
debug2: channel 0: garbage collecting
It turns out that this means the tunnel doesn’t even see the service. After wasting time with firewall tests and a lot of other false leads, I finally noticed the [::1] notation in there. Yup, that’s an IPv6 address. The solution is to make sure the ssh tunnel is using IPv4. For reference, the command that works is:

ssh -nNT4 -R 5500:localhost:5500 -l my_username myhost.com

Why do they have to be so damn difficult?

I’ve been working on upgrading my grandmother’s machine from a five or six year-old eMachine PII/300 running Windows 98 to a brand new Compaq deal I got at CompUSA along with a monitor and printer. The machine is reasonably fast, and, after I uninstalled all the crap that it came with and threw on some more reasonable software, I’m nearly at the point where it works the way she can use it.

Frustration 1. Moving files from one machine to another. I knew I wasn’t going to be able to use the network, as the old machine doesn’t have ethernet. New machine has no floppy. I could have uploaded all of the files via modem (probably get about 2 kB/sec), but that would have been painful. Easy, though, I’d use my USB memory key. Except that Win98 doesn’t natively support USB keys. Download the drivers… discover that they all want Win98SE, and won’t install. Grind teeth. Outcome: Success — finally had to use a USB CDR I had at home, since it at least had drivers on CD for Win98.

Frustration 2. Printer failure. The HP Deskjet 3915 just sits there flashing its light. Figure it’s a driver problem, so download 8MB of updates (over that mighty 2kB/sec connection). Still, nothing. Paper manual says that I should read documentation on CD. Documentation on CD says that I should consult the error code provided by the HP software. HP software says everything is fine. CD documentation’s best suggestion is to reboot. Windows thinks the driver is fine; HP thinks the printer is fine. The printer queue says it’s printing. Nothing ever happens. Reinstall drivers. Repeat. Outcome: still unsolved. Next step — call HP tech support. Oh, joy.

Frustration 3. Importing old mail from Netscape 4.8 to Thunderbird. Importing address book was simple — worked beautifully. But for old email, no such luck. There’s a tool “Wizard” for importing mail. But it doesn’t allow you to point it at files, it wants you to pick the profile. But it doesn’t see any profiles. I try putting the Netscape 4.8 user directories in all of the reasonable places (no, really. I tried all of them), but it never sees them. Documentation doesn’t yield any help. Try other directories. Try copying mail files directly into the appropriate directory in Thunderbird’s profile to no avail. Outcome: gave up.