A message describing the problem was put up on host9 (in fact, it would appear to still be there), and www.linode.com was pointed there... but if you had the address for the normal www.linode.com in a DNS cache somewhere along the line, you probably wouldn't be able to see it.

Personally, I'd like to see something like a "status.linode.com" in DNS with a short TTL, so it can be pointed to a working machine in the event of a failure. Though hopefully it'll be a while before we'd have to test it...

Yaas indeed,

That message appeared to be timed at 9:24 EDT, which makes it 2:24AM here in New Zealand I was in bed well before that, happy in the belief that when I got up ThePlanet would all be sorted!

Yes, that is a very good idea.

Same here, I'd rather any money was put to a better use than wasting time doing refunds

Perhaps a better use of the money would be to send management team of ThePlanet on a Disaster Recovery course! or at least to get them to read that part that says "test your systems under load".

The following quote has been edited to make it easier to understand the real reason for the outage:


Do they not do monthly tests on their equipment?

They can do all the tests they like, but in my opinion and experience ThePlanet has a typical (for big data centre) power systems, and these are designed at birth to fail.

I have a Linode and a server in London, and both have lost power within the last year.

My employer owns several systems hosted in France by Tiscali (old C&W) and we never had a power failure in 3 years. But our contract guarantee a 2x pro-rata refund for the hosting fees, I'm not sure of the minimum downtime / month, IIRC it should be around one hour.
Obviously the costs and needs for this kind of hosting aren't the same.

This is exactly my point. I've worked for 3 different ISP's in the UK over the last 7 years - the last one being Genuity - and we had not had any power problems.

We tested our equipment each month by tripping out the live power feed.....and it worked every time.

It did help that we had two power feeds to each server, two different UPS systems and 2 diesel Generators outside. Testing one power feed at a time made sure that should something fail we were still up and running on the other power feed.

It's just a question of design, so I guess were right back at what dbuckley said about the system being designed to fail from birth!


EDIT: Just noticed that my post count has dropped from 45 to 4 and I registered in 2004 and not in 2003 like I had done.

I've got two linodes, on hosts 40 and 42. This isn't the first non-trivial outage for those machines in recent memory, just the longest. I don't want to formally demand a refund, or any sort of compensation, but I do want caker etc to take the effort to make sure this is less likely to happen again.

Additionally, I strongly support the status.linode.com idea from above, but I say forget the short TTL... just permanently host it in another data center. It was several hours into the outage yesterday before www.linode.com was redirecting and giving status info. Until that time, I was afraid that linode had forgotten to pay a colo bill and the equipment was being reposessed or something. "Next time" something breaks, I'd like to know whether the ETA for fixing is within a few hours or if I should start the process of moving domains for the duration of the outage.

Thanks.

If you think a linode is a highly available solution, you're just plain wrong.Forget what linode.com says, forget what theplanet says (they say 100 uptime sla). I don't get to slap down my 20 bucks and expect an HA soltuion.

If you're going to be using UML, might be better to have 2 vm's, from different providers, and implement round robin dns with a very low TTL - at leat you have the option to take out the failed node. Kinda like a poor mans load balancer.

I've been pretty happy with the uptime - it's a least as good as any hosting company out there...

ericcron: If you think a linode is a highly available solution, you're just plain wrong. Forget what linode.com says, forget what theplanet says (they say 100 uptime sla). I don't get to slap down my 20 bucks and expect an HA soltuion.

That's true, but I also wouldn't expect to slap down my 20 bucks and not get the advertised uptime. Arguing over a few minutes of downtime/unreachability would be silly but some of us are talking about nearly 8 hours. That's an insane amount of time to be down and not expect some sort of compensation. Add to that the fact that it took a few hours before info was posted/available to all on what was going on...

I took a look at the linode.com pages. They do mention a 100% uptime SLA for network and power between TP and Linode.com.

No mention of any guarantees between Linode.com and the individual customers.

I'm not bothered by the outage since it was just my personal mail with backup DNS and MX elsewhere.

I'm just pointing out the 100% thing was between TP and the company (Linode.com), not TP and us or Linode.com and us.

Linode service is pretty good on the whole. There's the occasional UML bug or rare network burp, sure. So at best, Linode.com's uptime is actually somewhere in the ballpark of 99.990-99.993% on an individual host level.

I don't get the impression that this service was designed for a 24x7 100% availability setup; it was more targeted towards the 99.xx% crowd.

Could Linode.com clarify the 100% figure better on the web site? Certainly. Could add some wording indicating that this was between TP and Linode.com, and not between Linode.com and the individual Linode.com customers. Better yet, add some explicit wording mentioning actual availability between Linode.com and the customers were on a best-effort basis.

AIUI, since Linode.com never guaranteed 100% availability to its customers, nobody was really paying money for 100%. They were paying for what Linode.com could deliver -- occasional burp and all. Could the advertising be more clear and explicit? Yes.

Big 'effing deal people, get over it, move on. Every DC has or will have their problems.

_________________
/ Peter

Ahh, that's right. Mea culpa -- and thanks for the correction.

This is the kind of post I truly despise. People who come in and think they have a right to tell everyone else what to talk about and not to talk about.

By your logic you should not be reading this thread or posting in it. You should be saying to yourself, "people are upset about the downtime and are discussing it on the forums - big effing deal, I'll get over it and move on". Then you should get over it and move on rather than trivializing other people's concerns.