Today a totally ridiculous coincidence happened, which I want to document somehow since it was fun to fix and will probably never happen again.
When Mind launched their new website back in November 2012, we hired a keen team of digital volunteers to set about the tedious task of migrating content from one site to the other. Mind has a huge site, it was an endless and dispiriting task which we eventually decided to shelve, once the main content was over.
What we left behind were several hundred old news stories and blogs, back as far as 2010. Since then we’ve had a fair amount of internal and external pressure to bring the rest across, none more demanding than Moz.com and its hard to ignore report of 404 errors resulting from dead inbound links. Google is our master and we’ve been willing anarchists causing SEO chaos all over the internet, it was time for action.
A few weeks ago we decided to revisit the monster; exported the dead links in priority order and hired someone to migrate them. He’s been doing his job excellently for most of every day since he started, all’s been fine.
Today, we had an announcement to make, a lot of work went into getting as far as launching the story and the last bit was the news article – no big deal. The media team added the news story, naming it “Mind Media Awards: shortlist announced”, which naturally inherited the standard url format taking into account its structural position and title. All fine.
Elsewhere in the office, not 20 seconds before this happened, our patient content migrator had published a story, from the hundreds he was working through, called “Mind Media Awards shortlist announced”, a defunct page from back in 2011. Not a risk, right? Our CMS is smart enough not to let us name two pages the same thing, that would be silly. Except that he didn’t, there was no colon in his news story, so the website says “fine, these are different things, you can have both.”
But back in the 1970s, a group of nerds at Ascii decided characters like colons in a url pathway would probably confuse or break things, so decided to omit them, which means servers don’t read that they were ever there, even if good CMSs like Umbraco are fine with the whole thing.
At 4pm we tweet our release and point to the news story – right before some sharp-eyed tweeters alert us to the fact we’ve sent out a story from 2011. I’ve been managing the migration guy while my boss is away so know straight away what must have happened (and what a ridiculous coincidence! Of all the hundreds of urls on that list!)
To solve it, I set up a 301 from his content node to ours. Only of course it doesn’t work, because Umbraco knows they’re different, but the server’s thinking “why are you making me serve this page in circles you strange person, what kind of digital officer are you?” And I’m thinking “how on earth did this all happen at the same time?!”
We delete the migrated page, change the url of the news story and set up a 301 from the old to the new. Sorted. 15 minutes of team-awe ensue, while the rest of social media is none the wiser. Ace.