How I Cut Myself on Edgy Eft’s Bleeding Edge (Twice)

As a loyal Ubuntu user, I was excited when Ubuntu 6.10 Edgy Eft was released last week. I downloaded my ISO via BitTorrent to have available for any new installs, but planned on upgrading my laptop and desktop machines via these four commands at the terminal:

sed -i 's/dapper/edgy/g'/etc/apt/sources.list
apt-get update
apt-get upgrade
apt-get dist-upgrade

After getting the list of packages to upgrade, I found that a few packages, most notably Beagle, Drivel, Tomboy Notes, and a few Python packages, were held back, blocking my upgrade. I knew that the apps were included in the new release, so I figured maybe they needed to be removed because they weren't in the main repository or some such. Given that, I uninstalled them and re-ran the upgrade commands. The upgrade ran fairly smoothly for several hours as it downloaded hundreds of megabytes of files. While the upgrade was running in the background, I continued to work on my system. The first thing I noticed was some cosmetic changes. The Kate editor had a new splash screen, which I noticed when I used it to edit some files. Thunderbird had a new default font, as did some system menues. After switching to a blank desktop, I realized that the background and default theme had been updated. I thought all of these were pretty cool, as they happened transparently while I was working without so much as a hiccup.

When the upgrade was complete, I was prompted to reboot. I happily rebooted the system.

That's when it got ugly.

Somehow, despite Dapper and Edgy being able to recognize my display and configure it properly (Dapper when booting from CD or the drive, Edgy when booting from the CD), it choked when trying to launch X11. I looked over the diagnostic info in the log, but after a few failed attempts to edit the xorg.conf file to a state of happiness, I decided to back up my home directory and do a clean install of Edgy from CD.

In addition to my ext3 and swap partitions, I have an NTFS and FAT32 partition on this machine. I backed up my home directory to the FAT32 partition, then booted from the Edgy CD and selected the options to install over the now-corrupted Dapper to Edgy upgrade. The clean installation went fairly quickly, but I was not able to find the copy of my home directory that I had backed up!

I booted into Windows and couldn't find the files there either. I ran CHKDSK on the FAT32 partition, which found and corrected errors, but didn't find my files. Deciding that there wasn't anything critical there (well, hoping so, anyway) I booted back into Edgy to see what was new with this release.

That's when I got my next surprise. Suddenly it would not mount my FAT32 or NTFS partitions. The partitions simply did not show up. At this point I was getting concerned that I was having hardware problems, so I ran SpinRite on the drive. It found no problems, but my data didn't come back either. Satisfied that the drive was okay, I booted into Edgy again.

This time, after seeing the Ubuntu splash screen momentarily, the whole display went black, except for a blinking cursor in the upper left corner of the screen. I waited for a couple of minutes, and after no disk activity and no login prompt, cold-booted the system. The same thing happened again. Not sure exactly where the machine was in the boot process when it hung, I pressed Ctrl-Alt-Del to see what happened. To my surprise, the Gnome login prompt popped up!

Deciding that this was simply too much weirdness for now, I pulled my 6.06 LTS CD off the shelf, booted from it, and kissed Edgy goodbye for now, as I put Dapper back on my laptop. I'm a little disappointed, because some of the enhancements I saw during the brief time I was in Edgy looked nice, but stability is the main reason I wanted to run Linux in the first place, and I decided Dapper's pretty stable, so I'm going back to Ubuntu 6.06 LTS Dapper Drake for now, at least on my laptop. I'll write again shortly with an update of how the upgrade went on my two desktops at the office.

UPDATE

My X11 configuration was also trashed on my Dell Dimension 2400 after attempting an in-place upgrade over my previous Dapper installation. Beware!

Why the Microsoft POP3 Connector for Exchange is a Piece of Junk

I’ve used the POP3 Connector for Exchange which comes with Small Business Server since SBS version 4.0 (NT 4, Exchange 5.0). The main limitation I’ve always bumped into was the fact that the POP Connector will check mail only as frequently as every 15 minutes.

While this is normally acceptable, if a user is anxiously awaiting an email, or, as has frequently happened to me, you’re on the phone with someone and they email you a file so you can collaborate with them on it, waiting up to 15 minutes for the file can be a real time-waster.
The second issue, which I’ve never personally cared about, was the inability to leave an email which was downloaded on the server for a period of time. Many of my clients do this as a backup option (in case something gets corrupted or deleted after being downloaded), or as a poor substitute for IMAP, to allow them to receive their mail on multiple computers. Of course, given that they have an Exchange server, this is relatively pointless, but I digress.

Last week, I found the main reason why the POP Connector for Microsoft Exchange should be avoided at all costs. It is summarized in the following Microsoft Knowledgebase article: http://support.microsoft.com/kb/842293/

SYMPTOMS

When the Microsoft Windows Small Business Server (SBS) 2003 Connector for POP3 Mailboxes component downloads e-mail messages from an Internet Service Provider (ISP) all the following symptoms may occur:

E-mail messages that are downloaded by the SBS 2003 Connector for POP3 Mailboxes are not successfully delivered to the intended recipients.
The Microsoft Exchange 2003 Server-based computer removes the e-mail messages that it was not able to deliver.
The senders of these e-mail messages do not receive non-delivery reports from the Exchange 2003 Server-based computer as they typically do if their e-mail messages do not reach the intended recipients.

To summarize, the POP Connector gets the message, determines that it doesn’t know what to do with the message, and silently throws it away.

Without any warning to the user. Or the server administrator. Or the sender of the original message.

Who was the rocket scientist who came up with this plan of action?!

I’ve been using IGetMail from the folks at Lockstep. It costs $69, and has a free trial. The program allows you to customize the polling schedule to your liking (including poll freqencies of less than every 15 minutes!). It also has more flexible handling of emails than the built in POP Connector, and can be configured to simply delete emails that are corrupted on the server. This feature has saved my clients a lot of troubleshooting time, especially in cases when certain Asian-language spam or a corrupted attachment has managed to really confuse the client and server, which would have caused a client to be unable to receive any email in the past.

If you’re using POP to pull your mail down to an Exchange server, I highly recommend IGetMail as an alternative to the built in POP Connector.

Is anyone else experiencing multiple hard drive failures lately?

In the last month, I’ve had the following run-ins with failing hard drives:

  • A local credit union had two teller stations in different offices die, and a third started to act up, but running SpinRite, chkdsk, and defrag seems to have fixed it.
  • A local law office lost a drive in their server, as well as an external drive attached to their server. Today, about two weeks later, one of the lawyer’s workstation drives died as well.
  • A client’s home PC had a drive failure.
  • A client’s main Citrix server lost not one, but two drives in its RAID array.

All of these happened in separate towns (Bethel, Randolph, Rochester,  Rutland, and Woodstock, Vermont), in Dell computers, running a mix Maxtor, Seagate, and Western Digital hard drives. I’m no stranger to hard drive failures, but having so many fail all in such a short timeframe seems a bit odd. Anyone else experiencing this sort of behavior lately? If so, leave a comment. I’d love to hear from you.