On Wed, Aug 1, 2018 at 2:51 AM, Kenneth Peiruza <[email protected]> wrote:
> PS: depending on which hardware do you emulate, it can be really difficult
> to determine.
> Imagine a KVM emulating regular hardware and CPUs instead of using default
> qemu-cpu and virtio NICs and HDDs. It's gonna be tricky to get aware of the
> platform and it will make no difference because you will not tweak anything
> inside that VM...

That's completely untrue, especially for virtualization.  There are
many "standards" to such now.

While I do _not_ expect a junior sysadmin to know the various, common
locations in /sys (that just comes with long experience), I do expect
them to be able to use some of the commands/locations -- such as those
found in this StackExchange article, for example. [1]

Several of those, _even_ when not using QEMU-VirtIO paravirtualization
drivers, _always_ still show up.  It's one of the very first things I
teach any new (to Linux) sysadmin in every datacenter.  They need to
know what kinda of system they are on for many operations.

I.e., some datacenter policies and considerations are different
between baremetal and virtualization -- especially if they screw up
something and have to get on the console.

E.g., I've yet to be in a datacenter where I had physical access to a
server, and it was a major PITA to get such.  You have to know what
you're on, and how to get to a console "in the worst case."  That's
101 to me.

That said ...

_Unlike_ virtualization, I'm agreeable to the fact that
containerization is still more complex of a concept for level 1.  Even
this StackExchange article does detail that, although "cgroups" are
the common denominator of most. [2]

Although with systemd, I'm at the point that "cgroups" should be level
1 at some point, even if not the current revision (considerate for the
next).  Even Red Hat was using cgroups (along with namespace and
SELinux MCS) for OpenShift _before_ Linux Containers (LXC) and Docker
were even a thing.

Most of these facilities have been "standard" in other UNIX flavors,
and "required knowledge" for junior admins of those systems, and are
now commonplace in Linux.

As always ... just my experience, not telling anyone what should or
should be in.  But I consider virtualization "identification" (from
inside the guest) to be mandatory knowledge for any junior sysadmin in
a datacenter.

- bjs

[1] 
https://unix.stackexchange.com/questions/89714/easy-way-to-determine-virtualization-technology

[2] 
https://stackoverflow.com/questions/20010199/how-to-determine-if-a-process-runs-inside-lxc-docker
_______________________________________________
lpi-examdev mailing list
[email protected]
http://list.lpi.org/cgi-bin/mailman/listinfo/lpi-examdev

Reply via email to