Microsoft Outlook to Mozilla Thunderbird: Making Backup Times (And Sizes) Less Absurd

It’s been four days since I culled my RSS subscription list, and Mozilla Thunderbird is still behaving itself, so I’m inclined to call this particular hiccup resolved. Yes, I know I’ve just jinxed myself by typing those words…

On to my next Thunderbird gripe. I rely on OS 10.5’s Time Machine facility to automatically back up my MacBook Air to a network drive each hour; to a Netgear ReadyNAS NV+ NAS (and formerly to an external HDD USB-tethered to my Airport Extreme N router) here at the home office, and to a Time Capsule router at my friend’s place. Early on, I realized I’d need to exclude my VMware Fusion virtual machine images (Windows XP and the Google Chrome O/S) from the backup process because of their size; each image dynamically updated whenever a virtual machine was running, therefore making it an incremental backup candidate, and backing up a multi-GByte file structure even over IEEE 802.11n takes quite a long time. This, in fact, was one motivation to migrate from virtualized Microsoft Outlook to native Mozilla Thunderbird; my email archive would be regularly archived as a result.

However, I quickly noticed that my every-hour backups had increased in size from a few dozen MBytes to a few hundred MBytes in size. Even at their former bulk they were slightly annoying; the predominant factor in their girth was Firefox’s history file, which I hadn’t bothered manually excluding from the backup file-and-folder suite. But Thunderbird had obviously become a substantial factor in the per-hour incremental-backup payload, and the limited number of hour-to-hour updates to MBOX and MSF (i.e. per-MBOX index) files didn’t explain the uptick.

A bit of Google research uncovered the culprit; a file called global-messages-db.sqlite (yes, another SQL database) in my Thunderbird profile folder. It’s the master index for all messages, and ironically I suspect its corruption is what caused my initial Thunderbird issues. Backing it up isn’t strictly necessary, since if it’s not present Thunderbird will recreate it (although the re-indexing algorithm will take time and consume incremental system resources in the process). By excluding global-messages-db.sqlite and companion global-messages-db.sqlite.journal within my Time Machine configuration settings, backups have returned to a semblance of their former size and latency.

The other key to minimizing Thunderbird profile backup delays (along with, for that matter, Thunderbird launch delays and run-time perceived speed) is to keep the MBOX and MSF files as small as possible. This aspiration brings to the forefront another minor nit that I have with Thunderbird. Each account’s trash folder must be separately manual-emptied (although, admittedly, the option to automatically empty trash on a per-account basis is available in the program settings), and each account’s folders must similarly be manually compacted. In my case, the three accounts that get regular use are ‘bdipert@gmail.com’, ‘Blogs & News Feeds’ and ‘Local Folders’ (i.e. my imported Outlook archive). Global ‘empty trash’ and ‘compact folder’ options would be a welcome addition to a future Thunderbird iteration.

p.s…regarding my earlier grumble about Thunderbird’s Address Book only providing access to a maximum of two of the email addresses for each of the contacts in OS X’s Address Book, I’ve just confirmed that Thunderbird uses the first two email addresses in each contact’s OS X Address Book database, and that it offers them to the Thunderbird user in priority order that matches their listing order in the OS X Address Book database.