Appears that it cascaded across all of their nameservers. Seems like it could be an attack (remote or local DoS condition bug, since all users can create their own zones), or maybe just a random bug - hopefully they'll update us.
We have linodes in London, Newark, Atlanta, Dallas and Tokyo. All monitored by pingdom and none reported downtime... Our DNS is however hosted using Route53 and not Linode.
I'm a bit turned off by their (Digital Ocean) features page, in the SSD section, where they implied that SATA means traditional spindle-based HDDs. I kind of expect a cloud server host to know the proper terminology for these things. It's probably an editor's mistake but still...
Quote from Digital Ocean Features page: "With our SSD hard drives, you can expect much faster disk i/o performance when compared to a traditional storage medium (e.g. SATA)"
Can't really judge a service by their marketing. A company might have kick-ass benchmarks but a shitty site. Look at Linode. Their site design looks like it's still 2005.
I didn't even get to create a "droplet", the instance kept failing. I contacted support about a refund (I created the account to work on a small personal project for the weekend) and I was given the run-around, "there were issues and they were working on it", then they told me it was probably my issue (I forgot to "fill in a field"), etc. etc. Took 3 days to get my money back.
I'm sure it was an outlier since I've seen a decent amount of positive reviews, but support was terrible, and I think the way a company handles situations like this pretty telling.
13:53:14 caker@ : They operate completely independently, other than loading from the same zones. This looks to be a segfault in bind itself