Baseline build project progress.

Max TenEyck Woodbury mtew at
Tue Sep 17 05:23:01 GMT 2002

Sorry, for some reason this didn't get sent to the list. I must have hit the
wrong button. (My life's a bit stressful at the moment.) I didn't realize there
was a problem until no reply came and it didn't show up in the archive. 

-------- Original Message --------
Subject: Re: Baseline build project progress.
Date: Sun, 15 Sep 2002 17:18:00 -0400
From: Max TenEyck Woodbury <mtew at>
To: Andrew Bartlett <abartlet at>
References: <Pine.LNX.4.44.0209122215010.2200-100000 at garion>
<3D82064E.981F0B83 at> <3D8343BB.966A97B4 at>
<3D834621.5C4AE112 at> <3D837A75.F0F7DD2 at>
<3D8480CB.88AB1819 at>

Andrew Bartlett wrote:
> Max TenEyck Woodbury wrote:
>> I'm setting up a number of virtual machines with different OSs and OS
>> variations on them so builds can be done in those environments. I believe
>> that is one of (if not the main) functions of the build farm.
>> I've started with a virgin RH 7.2 and a patched RH 7.2 system. I have/can
>> get distributions and patches for 7.x and 6.2. With a little help, I should
>> be able to put up other Linux distributions and possibly a FreeBSD
>> distribution as well, but I need to establish a priority list since I can
>> not do them all at once. I also need to schedule the builds since having more
>> than one virtual machine up at a time is probably going to put more load
>> on the base system than I can handle, particularly since it has other uses
>> too.
> So, is the idea to create a large collection of machines on which to
> perform automated builds as per the build farm, on an hourly basis or
> so?

The exact rate will depend on the load they produce. I want to run each
virtual machine in the build group in turn with a script starting the next
virtual machine in the build set as soon as the previous one shuts down.
That is, each virtual machine will boot, do one build and then shutdown. The 
shutdown of one virtual machine will trigger the boot of the next on the list 
in a cycle. There may need to be a time delay added between boots to keep the 
load reasonable. If the load is light, I might try running two virtual 
machines at once. (The patch level machines will also see if there are any
new updates available from the distributor and apply them before doing the

> If so, then when you have them sucessfully building Samba and manually
> running the build farm scripts, give me a yell with machine names and
> I'll set them up.  We do however need these to be pretty well 'full
> time', otherwise just get lag issues.  It's also not generally worth
> duplicating (under VMware in particular) machines we already have, so
> check if we already have a particular setup building.

I tried the test script. It wanted a 'password'. Also, exactly which name
do you want? There are several -- the name on the internet interface -, the name of the base machine - vmware.localdomain, the
name of the baseline RH 7.2 virtual machine - currently rh72.localdomain, 
which may well be changed to sbfrh72.localdomain, and the name of the patched 
RH 7.2 virtual machine - currently rh72p.localdomain, which may be changed to 
sbfrh72p.localdomain. (I suspect you will want to set things up for sbfrh72
and sbfrh72p initially.)

Finally, do you have a preference for which distributions I should add to the
rotation first?

max at or mtew at

More information about the samba-technical mailing list