www.riscos.info down?

Chrome says www.riscos.info is unreachable.

https://downforeveryoneorjustme.com/riscos.info
(give it a minute to try to connect)

Still down

Still down.
Anyone have a contact point.

I think it’s been down for a couple of days at least.

When riscos.info goes down it seems to be down for days/weeks :-(

Theo has been hassling the hosting company for the past week, but it seems they did a firmware upgrade that has bricked a large (100+) number of their servers and they are waiting for replacement hardware. Yes this does seem a touch careless of them! (I’m assuming that number is virtual servers not physical, but even so it is a bit of a fubar)

It’s several hundred physical servers. New hardware has apparently been shipped in, but I think the size of the task is overwhelming the manpower they have and certainly their ability to respond to tickets. They were never great at responding to tickets in the first place.

I’ve ordered a new server with a different hosting company which has been provisioned, but a power cut during the restore means I now need to repair the backup server. It never rains but it pours :(

There’s a lot to be said for cloud storage.

There’s a lot to be said for cloud storage.

Walk in to my office and say that. There may be some cranial denting without me doing a thing.
A colleague has been assisting the FM company of the PFI partners to process a migration from set of hardware on our site to a “cloud solution” estimated time for data extract, anonymising and restore in the cloud 12 hours. Started 08:00 Monday. They’ve had the data at the cloud end for two days now and the lack of a “ding” for incoming mail on my iPhone¹ says they don’t have the system working yet.

So yes, the phrase you need is actually “there’s a lot to be said about cloud storage and plenty of it not rosey”

¹ I somehow ended up in the circulation for the mails

There’s a lot to be said for cloud storage.

Given that “cloud” storage also requires some sort of physical server somewhere (how long has it been since AWS US-EAST-1 last fell over? hint: less than a month), I read this as nemo being ironic.

There are indeed attributes to cloud storage, however the current trendy idea of dumping everything in the cloud and arriving at a total dependency on a third party service (especially if as a mere user and not a paying customer) is reckless.
More Rick’s rules:

• Sensitive data does not go on any cloud. No if/but/maybe. Understand that anything uploaded to a third party is no longer in your control. You only have the weak illusion of that.
• Nothing should exist in any cloud that you don’t have a local copy of. If the cloud fails, service goes bust, gets hacked, etc; nobody is going to cry you a river if all your important company documents are “lost”. You should always have local copies and, yes, a local backup.
• Related to the above: what do you mean nobody can do any work today because the cloud service is down!?!?!?

The cloud can be a useful aid. Not a replacement, an aid.

Distributed servers can be frustrating. For example, when something happens to be on another Akamai server and the upstream DNS hasn’t refreshed. I mean I get around DNS server issues to a degree by having a caching DNS server on my LAN which is connected to a bunch of other DNS servers.

I’m not wowed by “the cloud” but it can be useful when uptime of a web facing server is important.

edit:

If autobuilder had a switch to tell it to use cached copies of files instead of blindly seeking out fresh ones, that’d be fantastic and probably take a decent amount of load off riscos.info.

http://www.riscos.info/index.php?title=Special:AWCforum&action=st/id330/riscos.info_back_up_after_datacenter_meltdown

(ie it works again. I’ve still had no reply to more than two weeks worth of tickets at the old ISP…)

I’ve thought about cloud storage, unfortunately with the CI infrastructure it starts eating the GB rather rapidly (the new machine now has 2x 2TB which is a bit more than the old one). I’ve also thought about failover, but a) it requires time to set up and get right b) it needs paying for more servers. And since riscos.info doesn’t actually have any revenue streams…

Something I’d like to do is set up something to nightly mirror the SVN repos to github for general resilience, but it’s a bit tricky the way some of them are structured (especially GCCSDK). If anyone is interested in helping with this I’d be happy to give them access to the github and send them the raw repos to experiment with.

Seems to work well. Probably my imagination but there seem to be less lumps in it too. !PackMan seemed to be a lot more responsive.