What a week it's been. Primarily holed up inside 24/7 as we reached record lows with windchill. Our lowest temp reached -33°F, and with the wind we had would've made it feel like closer to -55°F. Dangerously cold. "Historically" cold. Things (schools, businesses, etc.) wound up being more or less shut down from about Tuesday morning through Thursday afternoon. It's certainly the coldest I could remember for the better part of 25 years (it was the mid-1990's). And it's the only time I ever remember campus being closed for more than a day (two and a half, all told). Crazy stuff.
But we survived. Everything here was in good working order (save for the snowblower which stopped "driving" ~36 hours before the brutal cold started). Since I was more or less done with it anyway I just pushed it back to its home for consideration later...which I did just yesterday to "fix" with a couple of new shear pins.
Also a good time to catch up with the backlog of Drupal- and site-related configuration that I'd referenced in the last post. Which I did, for the most part...at least to a satisfactory place where it was time to figure out the cutover/migration plan.
And that's where the good stuff stopped. For as much testing and whatnot I'd done with regard to the template I've mashed together and all of the various special Drupal blocks and configuration I've made...it didn't actually get a new production site up and running properly. I'd become relatively 'pro-level' at making theme and configuration modifications, "installing/uninstalling" them, changing necessary uuids and porting all of those bits to and from a couple of dev/test instances.
But I'd never actually done a content migration...namely as there wasn't much content to migrate (or so I thought). Most of the "core" pages were minor edits from the old site, as expected (though I really do need to refresh those pages), so it couldn't be too bad, right?
Those of you who know me understand that I tend to toss around the "dumpster fire" metaphor. But this was one of those situations I felt in the midst of such a thing. Absolutely nothing I tried during a "production" cutover was working...and I became pretty good at spinning up new instances of Drupal in short order, testing a few minor things, and throwing it all out. Oh yeah, and I'd felt comfortable enough to configure the Apache VirtualHost for the primary domain...which I didn't want to undo (I'll get it working in a few minutes anyway...).
So after an hour (or so) of repeated catastrophic failures, I decided to punt. Drupal itself was stable on a new production instance, and the theme and configuration was generally happy, so into site maintenance mode we went...to be continued later.
My initial postmortem indicates there were a couple of key failures in my particular environment:
- Cloning my primary test site was simply not succeeding for some reason. A site clone, even if successful/manual/automatic, would cause some bizarre and frustratingly unhelpful error to pop up (presented to the user as "The website has encountered an error..."). The logs indicate a problem with underlying core libraries baked into Drupal...but it was in several cases literally a copy with all of the rights and responsibilities. And not the kind of stuff that can be fixed with settings.php.
- Database content porting. I didn't spend a substantial amount of time working on porting the content (pages, blocks, etc.) over...and certainly not much time trying to troubleshoot it...namely due to the cloning problem.
- Spinning up a vanilla production site was always successful in that I could make it, configure it, theme it, and import my custom stuff, but I shouldn't necessarily need to go that far. A clean vanilla install also caused serious issues in trying to import database content.
Ultimately I gave up and went with a new, vanilla install with my configuration. But the content and its bits will need to be ported over manually...so we're delayed again. I'm hoping to have it all out of maintenance mode by the middle of this upcoming week. Fingers crossed!