|
|
OpenBSD is an off-shoot of the NetBSD project that focuses on security, portability, correctness, and standards conformance. The OpenBSD team has committed to a continuous audit of their entire source tree, and have found and fixed numerous bugs. Most of the other BSD derived systems now follow the OpenBSD team's software changes. More information on it can be found at: http://www.openbsd.org/
2 responses total.
The EeeBox hypervisor experiment
It's the July 4th holiday 2019, time on my hands, a perfect to
investigate the NOT pressing matter of my test OpenBSD VM not surviving
a suspend/wake cycle by the Linux host (which somehow magically works
for Linux guests).
With no idea how that might work for Linux guests, I thought why not
just move the OpenBSD VM to an always-on box. I try to keep my energy
foot print small so I had one always-on choice, my EeeBox 1007 nettop
used as my home file/media server. Could this work?
It did work - details below. The end result is an OpenBSD VM which I
think may be similar in performance to the Sun 3/4 systems used by Grex
in the early 1990's.
https://greatgreenroom.org/item_19.html
"Grex has continued to thrive, upgrading first to a Sun 3, and then to a
Sun 4/260 and a 4/670. It open up an internet connection early in 1993
and gradually more and more of its users came in over the internet. By
1999, it had over 25,000 users ..."
My OpenBSD VM running on the EeeBox produced a UnixBench score of 5.7.
Again 5.7, which is lower than the baseline UnixBench score of 10
derived from a SPARCstation 20-61.
https://en.wikipedia.org/wiki/Sun-4
https://en.wikipedia.org/wiki/SPARCstation_20
/usr/local/src/byte-unixbench-master/README.md
"The tests compare Unix systems by comparing their results to a set of
scores set by running the code on a benchmark system, which is a
SPARCstation 20-61 (rated at 10.0)."
For comparison my $5 Raspberry Pi Zero with the stock Raspbian OS had an
UnixBench Index Score of 122.
https://www.raspberrypi.org/products/raspberry-pi-zero
Unexpectedly, the performance of the OpenBSD VM on the EeeBox ain't bad
... so long as one is mindful of it's limits. That means text
applications written (and re-written) for speed. For me it brings back
memories from the 1990's of timing my database queries and re-writing
until the performance was acceptable and/or pushing some parts of the
process to 'night-processing'. If I could time travel back to the early
1990's I would bring the magical Raspberry Pi Zero and SQLite.
The ingredients:
(1) EeeBox nettop PC from 2010:
- Atom D410 (x86_64, 1.67GHz, 2 threads, no VM support)
- 1G ram
- Ubuntu 18.04.2 LTS
(2) Qemu 4.0.0 (built from source) This was a detour
as the 18.04 qemu package had a show stopping bug.
Install gcc, etc:
# aptitude install build-essential
# aptitude install pkg-config
# aptitude install glibc-source
# aptitude install libglib2.0-dev
# aptitude install libpixman-1-dev
Get qemu from qemu.org and extract to /usr/local/src
# tar xJvf qemu-4.0.0.tar.xz
# view README
./confugure --help
Follow the README directions:
# mkdir build
# cd build
# ../configure --target-list=x86_64-softmmu
# make
Note !!
Do not build all targets. Not needed.
Takes hours on the EB1007 - RTF README :)
Build with x86_64 target only took 33 minutes.
(3) A Qemu x86_64 OpenBSD VM.
Copy to your nettop:
Mine:
OpenBSD 6.5 installed with sets:
bsd bsd.rd base65.tgz comp65.tgz man65.tgz game65.tgz
(excluded all X sets)
The result:
(1) Start-up:
$ qemu-system-x86_64 \
-m 500 \
-display none \
-nic user,ipv6=off,hostfwd=tcp::$port-:22 \
-hda ${distro}.qcow2 &
Note the resource allocations:
500MB RAM - the EeeBox only has 1G
The CPU has 2 threads so the VM effectively gets 1
No hardware acceleration - not support by the Atom CPU
> boot time on EB1007 is ~8 minutes
> long running boot tasks: reordering libs, ctfconv
(2) UnixBench test on the EeeBox net top host (abridged):
BYTE UNIX Benchmarks (Version 5.1.3)
System: EB1007: GNU/Linux
OS: GNU/Linux -- 4.15.0-54-generic -- #58-Ubuntu SMP
Machine: x86_64 (x86_64)
CPU 0: Intel(R) Atom(TM) CPU D410 @ 1.66GHz (3333.0 bogomips)
Hyper-Threading, x86-64, MMX ...
CPU 1: Intel(R) Atom(TM) CPU D410 @ 1.66GHz (3333.0 bogomips)
Hyper-Threading, x86-64, MMX ...
-------------------------------------------------------------
Benchmark Run: Thu Jul 04 2019 21:16:37 - 21:44:45
2 CPUs in system; running 1 parallel copy of tests
... details omitted ...
System Benchmarks Index score 264.5
-------------------------------------------------------------
Benchmark Run: Thu Jul 04 2019 21:44:45 - 22:12:57
2 CPUs in system; running 2 parallel copies of tests
... details omitted ...
System Benchmarks Index Score 335.7
(3) UnixBench test on the OpenBSD VM guest (abridged):
BYTE UNIX Benchmarks (Version 5.1.3)
System: obsd.my.domain: OpenBSD
OS: OpenBSD -- 6.5 -- GENERIC#3
Machine: amd64 (unknown)
CPU: no details available
-------------------------------------------------------------
Benchmark Run: Thu Jul 04 2019 20:12:15 - 20:44:50
1 CPU in system; running 1 parallel copy of tests
... details omitted ...
System Benchmarks Index Score 5.7
(Qemu only, no kvm support.)
Note:
For UnixBench on OpenBSD, add the following:
'# pkg_add unzip' to extract files
'# pkg_add gmake' as make had missing dependencies
Create gRun from Run substituting gmake for make
A few OpenBSD pieces I think are worth reading especially if (like me) your *nix experience is mainly/entirely Linux: OpenBSD from a veteran Linux user perspective: https://cfenollosa.com/blog/openbsd-from-a-veteran-linux-user-perspective .html What are the differences between OpenBSD and Linux? https://cfenollosa.com/blog/what-are-the-differences-between-openbsd-and- linux.html
Response not possible - You must register and login before posting.
|
|
- Backtalk version 1.3.30 - Copyright 1996-2006, Jan Wolter and Steve Weiss