Re: [xcat-user] xCAT and Vagrant

Song BJ Yang Tue, 19 Feb 2019 18:51:02 -0800

Hi Daniel,

Great work!

I will look into the repo and forward this mail to the team to see whether we can build the automation of discovery and hardware control like this, currently all the such kind of test are triggered manually on bare metal servers.

one comment for your repo https://github.com/dhilst/qemu-ipmi, it will be better if you can add some description and steps in the README.md, maybe some information in this mail:)

------------------------------------------------------------------------------
YANG Song (杨嵩)
IBM China System Technology Laboratory
Tel: 86-10-82452903
Email: yang...@cn.ibm.com
Address: Building 28, ZhongGuanCun Software Park,
No.8, Dong Bei Wang West Road, Haidian District Beijing 100193, PRC

北京市海淀区东北旺西路8号中关村软件园28号楼
邮编: 100193

----- Original message -----
From: Daniel Hilst <dan...@versatushpc.com.br>
To: xCAT Users Mailing list <xcat-user@lists.sourceforge.net>
Cc:
Subject: Re: [xcat-user] xCAT and Vagrant
Date: Wed, Feb 20, 2019 9:45 AM

Hi everyone

I'm using OpenIPMI and QEMU to get xCAT working in a virtualized environment. I could test sequential discovery with such setup, and also use rpower an rcons commands transparently. I have everything on github. It's very handcrafted for my environment but it has been working ... I was just waiting for the right timing to improve it :)

First of all, you need OpenIPMI and OpenIPMI-lanserv installed, this is easy to install on most modern distributions, I'm using Fedora, but any distro would work. The network topology is the simplest one. There is a br0 at my host that receives the gateway IP and all BMC IPs, And there is three virtual machines, HN, CN1 and CN2, representing headnode and two computing nodes. Each machine has a single NIC for sake of simplicity. I'm using iptables to masquerade the network output to my wifi card.

For each VM there is a ipmisim process that acts as its BMC. For each process a address is attached to br0, this was the way I found to get host processes and VM communicating. Every MAC and IP was hard coded to get it working, there are a lot of space for improvements. Each ipmsim has its own configuration file, which has a start command. When you issue a power on command to ipmi it runs this start command that points to a script that power ups the virtual machine. SoL is working too, at last for kernel messages, as long as you redirect it to the serial console.

HN IP: 192.168.123.2

CN 1 IP 192.168.123.3 BMC 192.168.123.4

CN 2 IP 192.168.123.5 BMC 192.168.123.6 (I rarely use this one, my host machine lacks of memory)

Here it is: https://github.com/dhilst/qemu-ipmi

PS: I spend about a month on trying to get snmpsim [1] integrated with this environment for emulating a SNMPv3 switch in sake of testing switch based discovery without touching the xcat bits but failed miserably, there may be lost bits of this work on the scripts.

[1] https://github.com/etingof/snmpsim

Regards!

Em ter, 19 de fev de 2019 2:36 PM, Kevin Keane <kke...@sandiego.edu escreveu:

I've tried virtualizing xCAT for testing purposes. To some extent, it works, but the really interesting parts are very hard to virtualize. What tripped me up was UEFI booting and BMC setup/IPMI. Without getting these pieces, all you can test in xCAT is whether the tables are set up correctly. Even when you do get it to work, the virtualized version was different enough from actual hardware to be of limited use.

Also, even when you do get it to work, these things are very hypervisor-specific. I eventually got UEFI-booting to work in libvirt, but then had to switch to VirtualBox due to another project. And I never got to the point where I could have put it into Vagrant.

_______________________________________________________________________
Kevin Keane | Systems Architect | University of San Diego ITS | kke...@sandiego.edu
Maher Hall, 192 |5998 Alcalá Park | San Diego, CA 92110-2492 | 619.260.6859

REMEMBER! No one from IT at USD will ever ask to confirm or supply your password.
These messages are an attempt to steal your username and password. Please do not reply to, click the links within, or open the attachments of these messages. Delete them!

On Tue, Feb 19, 2019 at 7:38 AM Christopher Walker <c.j.wal...@qmul.ac.uk> wrote:

Is there a Vagrant model of an xCAT cluster?

If there were, it should be possible to build a test case for:
https://github.com/xcat2/xcat-core/issues/2633

While clearly it doesn't model the hardware, it would allow some sort of
testing of changes.

Chris
--
Dr Christopher J. Walker
ITS Research
Queen Mary University of London, E1 4NS

_______________________________________________
xCAT-user mailing list
xCAT-user@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/xcat-user
_______________________________________________
xCAT-user mailing list
xCAT-user@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/xcat-user

_______________________________________________
xCAT-user mailing list
xCAT-user@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/xcat-user

_______________________________________________
xCAT-user mailing list
xCAT-user@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/xcat-user

Re: [xcat-user] xCAT and Vagrant

Reply via email to