[Beowulf] anyone using SALT on your clusters?
jonathan.barber at gmail.com
Mon Jul 1 07:15:18 PDT 2013
On 28 June 2013 21:56, Joe Landman <landman at scalableinformatics.com> wrote:
> On 06/28/2013 04:45 AM, Jonathan Barber wrote:
>> On 27 June 2013 23:53, Joe Landman <landman at scalableinformatics.**com<landman at scalableinformatics.com>
>> <mailto:landman@**scalableinformatics.com<landman at scalableinformatics.com>>>
>> On 06/25/2013 04:55 PM, Paul English wrote:
>> The problem with SSH based approaches is when you have failed nodes -
>> normally they cause the entire command to hang until the attempted
>> connection times out.
> This isn't an issue for things like pdsh. Also theres a nifty utility
> called whatsup that handles all this for you. It makes determining what is
> up, well, fairly painless.
I will investigate pdsh to see how it avoids some of the problems I
described in my other response.
>> The message based systems typically don't have this problem because they
>> use a pub-sub model where the clients subscribe to hear the commands
>> from the server. If the client is down, the server doesn't wait on them
> This isn't so much of an issue.
<shrugs/> It's always been the problem I've had when trying to run commands
over a large number of hosts.
>> Honestly configuration management is largely a moot point for
>> image/remote boot, and an annoying necessity for local
>> I would like to point out that configuration management isn't something
>> you only need to use if you have webscale sites. IMHO it's useful
>> whenever you need to manage some configuration - including the
>> configuration of binary images. How do you manage the creation of that
>> image if not with some kind of configuration management tool (even if
>> your "tool" is a set of shell scripts)?
> I think you are conflating too many things here.
> 1) image management: This is for OS config as you are talking about
> above. Configuration management is usually associated with this to some
> degree, this is where yum and many other tools come into play at a low
By other tools do you mean tools such as apt?
> 2) configuration management atop a base image. Many distros try to mix
> these together, and often do a terrible job of it (rpm.saves anyone?).
For me, this is a distinction based upon tools and it suggests that we are
talking about different things when we refer to configuration management.
I would say that configuration management is the process by which you reach
the final operating system state, and therefore includes the "base OS" (or
JEOS to use a particularly hideous acronym) configuration (disk layout,
network configuration [bonds], etc.) to provide a platform to bootstrap the
rest of the system onto and then installing additional software and
configuring it all via script/puppet/chef/whatever and dpkg/rpm/tar or
So, moving the configuration to the point of installation (for me) doesn't
stop it from being configuration. Indeed, if your root is via NFS or a
binary image, then that configuration is still being done - just at a
single point instead of over many machines.
Please note that I'm not saying one approach is superior to the other, but
I would argue there is configuration management going on either way.
A big chunk of what you write about are best handled by a monitoring
>> system as compared to a configuration management system.
>> Yes, monitoring is not the same as configuration.
> I think you may have not grasped what I wrote.
Perhaps not, would you be kind enough to explain?
>> If you could completely eliminate the "install OS, run configure
>> on it" section of startup, would you? This isn't a sales pitch, its a
>> genuine question.
>> I don't understand your question, how can you eliminate configuration?
>> At some point you have to tell the system what it's supposed to do.
> Its done once. Then you don't have to install it again. Its installed.
> Its done.
> And if you have to change it afterwards, how do you manage the change in
configuration? Presumably you have to apply the changes to some base image
via some process (even if it's just a script or manual alteration with
$EDITOR) and then push the changes out to the nodes? If so, then I would
still call this configuration (management).
Joseph Landman, Ph.D
Founder and CEO
Scalable Informatics, Inc.
email: landman at scalableinformatics.**com <landman at scalableinformatics.com>
web : http://scalableinformatics.com
phone: +1 734 786 8423 x121
fax : +1 866 888 3112
cell : +1 734 612 4615
Jonathan Barber <jonathan.barber at gmail.com>
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the Beowulf