SliceHost has never failed to introduce unique features that impress me. They gave you console access via Ajaxterm, which allows you to rescue badly configured server over nothing but HTTPS. They have also recently introduced “Rescue Mode” for their VPS — another feature that you wish you never would need to use, but is super useful when you need it. It shuts down your VPS, reboots in a separate partition, and enables you to mount your VPS partition as /dev/sda1 for 90 minutes.
90 minutes might not be enough to do Gentoo install from stage 1, but there are still many other uses for it. Why is it useful?
- You accidentally deleted all your init scripts and your Xen VPS just refused to boot.
- Physical server got a power outage, and your VPS came back with a corrupted file system that requires
fsck. - You really want to run Slackware on your slice.
- You want to use a file system other than the default Ext3.
On the topic of changing file systems, one feature I used to love when I hosted with Unixshell# is, that it allows you choose the root file system before dumping the Linux distribution of your choice on top. Upon moving to SliceHost, I was a little bit disappointed that Ext3 is used by default. It is standard, it is well supported, and it is very stable — but everyone loves to have some fancy file systems powering their Linux boxes, don’t we all? :)
Over the years I have been a ReiserFS fan because of its performance over directories of many small files (which is common when you are putting cached data on FS without hierarchical keys). However, after a few non-recoverable crashes due to sub-standard power source at home and at work, I switched and am mainly running XFS on my servers. Maybe my next slice will also have XFS on it.
Delicious
Digg
Reddit
Comments
A french hosting company, dedibox, has had this feature for a long time: you go to your console, choose you server and reboot it through pxe (I guess) in rescue mode. very convenient. It is sad that their website is only in french, because their offer is pretty good…
Hi — I think it is booting up a dedicated server through PXE serial console, which is a bit different from SliceHost, which is rebooting a VPS from a spare partition.
I’ll say having a serial console would actually require more work by the hosting provider as everything needs to be wired together. But as you said, it has been done for a long time. Just that I have not seen anyone doing it for VPS yet.
And I wish I know how to read French :(
The dedibox page says “en le démarrant en réseau sur un mini système d’exploitation”, which means that your server will be rebooted, on the network, to a mini-os. I can’t test right now (I would need to reboot my server :), but I think this is exactly what happens: you pxe boot a small linux distro, and get access to all your partitions. And you don’t have any time limit :) If you need some help translating parts of their pages, just ask me.
Thanks for the writeup Scotty. We’ll be keeping an eye on usage over the next couple of weeks and if possible extend the 90-minute limit. Also - we’re planning on giving filesystem options during a rebuild.
Virtuozzo has had a similar rescue mode for a long time… it creates a new VPS for you and mounts your old root under /repair. You can take it in or out of repair mode via Virtuozzo Power Panel. I don’t consider this anything groundbreaking, sorry.
Matt — however the usefulness of this repair mode is limited under VZ. For one, you don’t get to install your own OS, or fsck your own partition, or change filesystem, do you? It is only useful when init process failed, or does it have any other use on VZ?
Post new comment