[Cialug] Filesystems!

[Jeffrey Ollie]

In a cloud setup like AWS (or really any cloud provider for that matter) I
wouldn't use software RAID to try and get some redundancy. First, the
default storage for EC2 instances is EBS which provides its own redundancy
and high availability so running software RAID on top of that seems like a
waste. Second, trying to achieve redundancy with software RAID in the cloud
means that you're not really taking advantage of the cloud (i.e. you're
doing it wrong). The cloud is different, and you shouldn't be running your
software in the cloud in the same way that you run software on bare metal
servers that you've stuck in the hall closet next to the bathrooms.

Personally, if this was a money-making site I'd strongly consider using RDS
and letting Amazon figure out redundancy and high availability for my
database.

If you use a different cloud provider they may not provide EBS-like storage
options that provide their own redundancy. Even in that case I wouldn't run
software RAID because I doubt that you could get a guarantee that two block
devices were actually on different physical drives. So if one physical
drive failed it might take out both of your virtual drives.


On Fri, Jun 21, 2019 at 2:04 PM Todd Walton <tdwalton at gmail.com> wrote:

> I learned today that a davfs mount can cause your computer not to boot
> properly. I created the mount yesterday and it worked just fine. But
> upon rebooting my computer I could not get past "emergency" mode.
> (Single-user, essentially.) And I couldn't log into emergency mode
> because root doesn't have a password! I appended "init=/bin/bash" to
> the linux16 line in the grub entry and set a root password. Eventually
> figured out that it was the mount. Rebooted and got to the graphical
> screen.... and... it hung. I held down the shift key and whacked at
> the numbers on the top row of my keyboard, but it didn't help. Then
> figured out that if you change the root password that way you have to
> also relabel the entire filesystem with security context
> thing-a-muh-bobbers for SELinux. I'm all good now. Phew.
>
> So I'm working on cloning a database server today, for another
> emergency, and I discover a bit of weirdness. Tell me if I'm wrong,
> but it seems a little silly to have multiple disks that are software
> raided -- just raid 1, just a simple mirror -- and then LVM on top of
> that raid array. And all of this on an AWS EC2 instance, so Amazon
> *already* has the disks on some sort of high availability setup. I
> understand using LVM for the flexibility of moving disks around,
> growing, shrinking, whatever. But raid under that? In fact, I wouldn't
> even use LVM's raid features. How likely is it that one of AWS's disks
> is going to fail? And we're taking backups of the server.
>
> So complicated.
>
> --
> Todd
> _______________________________________________
> Cialug mailing list
> Cialug at cialug.org
> https://www.cialug.org/cgi-bin/mailman/listinfo/cialug
>


-- 
Jeff Ollie
The majestik møøse is one of the mäni interesting furry animals in Sweden.