-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathsqsh_layers_with_ram_upper_why_it_is_awesome.txt
41 lines (34 loc) · 2.11 KB
/
sqsh_layers_with_ram_upper_why_it_is_awesome.txt
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
-*- mode: outline -*-
So, why is using a non-retentive layered rootfs so awesome? Good question.
The more I use it, the more I like it. You might find you like it, too.
Especially if you're constantly compiling things and messing up your nice clean
installation.
1. It's a magic undo button. First and foremost, it gives you the ability
to just go ahead and install whatever you want w/out any fear of
breaking your system. So, you just installed a new version of some
package and it completely broke the universe? No biggee, just reboot
and you'll have the system back up and running in no time. Oh, the new
version works greate and you want it to be a permanent change? Just
make a new sqsh_layer out of /mnt/root-true/upper/upper and add it to
your bootloader config.
2. Quickly see what you've changed since the last time you baselined your
rootfs. This makes keeping track of changes very easy.
3. Installing rapidly. You can build up multiple machines using the same
sqsh_layers very rapidly and just tweak a few things (e.g., hostname)
via upper layer.
4. Super easy upgrades. Just stick new sqsh_layers on your boot device,
maybe clean things out of your upper layer, and reboot.
5. Just go ahead and kill power w/out doing a safe reboot. I dare you. As
long as your sqsh_layerdev is mounted read-only (the default) (or not
even mounted if you loaded the layers into RAM), you don't ever have to
wory about power outage (and an itchy power button finger) corrupting
your root filesystem. Yup, "safe shutdown" just got reduced to "remove
power".
6. Shared rootfs images accross multiple virtual machines. This advanced
technique involves using the same device containing a colloction of
sqsh_layers as the sqsh_layerdev for multiple virtual machines
(read-only!) that each have their own independent root devices (i.e.,
for upper layer). This makes it incredibly easy to quickly go from
having no virtual machines to having a whole fleet of them. And you can
upgrade the sqsh_layers once and upgrade all of the virtual machines at
once.