Linode was down | Hacker News Reader

Linode was down

linode.com

31 pointsjemhoff25 comments

From the #linode IRC channel, caker (CEO) states it was a segfault in bind.

13:53:14 caker@ : They operate completely independently, other than loading from the same zones. This looks to be a segfault in bind itself

Isn't having two independent name-servers supposed to prevent this kind of thing?

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.

Weird!

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.

The most interesting thing in this thread is frankly how many people seem to be using Pingdom.

My existing SSH connections are still up, must be a DNS issue.

That explains the pingdom notification.

So, Digital Ocean is looking pretty good right now.

Edit: New Jersey data center.

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)"

They likely just called it out like that here because so many other hosting providers use that as standard terminology.

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.

FYI I created a test account last Friday and got some of the worst service I've had in 10 years of hosting stuff online.

I've had no problems with my tiny node in the few days I've had it - what problems did you run into, so I can be on the lookout?

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.

Didn't experience any loss of connectivity from my VPS hosted in their London DC.

Back up now. Linode has definitely been doing all they can to scare me off!

My external http monitoring had no issues accessing my websites on linode. I do not user their DNS though.

[Edit: My nodes are in the Dallas Datacenter]

I use Linode DNS and I saw no issues from Pingdom either (also in dallas).

fwiw I use the dallas datacenter, linode dns, and pingdom said I was down for 11 minutes.

Must have been very short, I can access my sites, login to admin, and see all of Linode's pages as of 3:49 CDT

Pingdom detected no downtime for beeminder.com, hosted in Linode's New Jersey datacenter (I think).

Confirmed. I wasn't able to access my sites for like 5 minutes.

Haven't experienced any downtime for my servers...

DNS issues seem to be resolved now.

No downtime from London

Reply on news.ycombinator.com