Re: [SLUG] Managing software across a cluster

[Amos Shapira]

Go ahead and check Spacewalk for your particular situation, but personally
when I looked at it, it seemed to be much too complex to setup and use.

I just came across Katello (http://katello.org/), which might interest you
as well if you are eyeing Spacewalk.

--Amos

On 15 August 2012 11:45, Steven Tucker  wrote:

> Hi all,
>
> thanks for the suggestions, they were awesome.
> Since writing that email Puppet has been suggested to me as well, though I
> will go look into all the suggestions.
>
> Another option I am looking at is Spacewalk, which I may end up going with
> as I just found out that some other Redhat servers around here are managed
> with Satellite.
>
> Thanks again, always get great suggestions here.
>
>
> Tuxta
> --
> SLUG - Sydney Linux User's Group Mailing List - http://slug.org.au/
> Subscription info and FAQs: 
> http://slug.org.au/faq/**mailinglists.html
>



-- 
 [image: View my profile on LinkedIn]

-- 
SLUG - Sydney Linux User's Group Mailing List - http://slug.org.au/
Subscription info and FAQs: http://slug.org.au/faq/mailinglists.html

Re: [SLUG] Managing software across a cluster

[Nick Andrew]

On Wed, Aug 15, 2012 at 11:45:07AM +1000, Steven Tucker wrote:
> thanks for the suggestions, they were awesome.
> Since writing that email Puppet has been suggested to me as well,
> though I will go look into all the suggestions.

Another vote for puppet. The more you use it, the more useful
it gets.

Nick.
-- 
SLUG - Sydney Linux User's Group Mailing List - http://slug.org.au/
Subscription info and FAQs: http://slug.org.au/faq/mailinglists.html

Re: [SLUG] Managing software across a cluster

[Steven Tucker]

Hi all,

thanks for the suggestions, they were awesome.
Since writing that email Puppet has been suggested to me as well, though 
I will go look into all the suggestions.


Another option I am looking at is Spacewalk, which I may end up going 
with as I just found out that some other Redhat servers around here are 
managed with Satellite.


Thanks again, always get great suggestions here.

Tuxta
--
SLUG - Sydney Linux User's Group Mailing List - http://slug.org.au/
Subscription info and FAQs: http://slug.org.au/faq/mailinglists.html

Re: [SLUG] Managing software across a cluster

[Jason Ball]

Oddly I actually boot a large cluster off a diskless image, with a local HD
for file caching/work space. The boot image is relatively small
(500MB).   Yes this uses a little ram for the root file system on the
'diskless' node, but provides a single point of change and a reboot to
update the cluster software.

xcat is used to manage the cluster and I highly recommend it.

J.


On Wed, Aug 15, 2012 at 8:41 AM, Amos Shapira wrote:

> Just get it all under puppet (puppetlabs.com).
> On Aug 14, 2012 9:55 PM, "Steven Tucker"  wrote:
>
> > Hi all,
> >
> > I have just rolled out a Scientific Linux (Redhat clone) cluster of 20
> > nodes plus controller that runs torque for batching and uses a lot of
> MPI.
> > My problem is that I periodically get asked to include some library or
> > program to help out with some ones research code, and often it is
> something
> > that needs to be installed across all nodes.
> >
> > On my Mageia cluster I simply run urpmi parallel, issue the one command
> > and it installs across all nodes. For those not familiar with urpmi
> > parallel, it queries all the nodes asking what rpms are needed to install
> > the given software (including all dependencies), it then downloads all
> the
> > rpms to the machine running urpmi, it then pushes the appropriate rpms to
> > each node and installs them.
> >
> > Is there anything like this for Scientific Linux (RHEL) and Yum ??
> > The closest thing I have seen so far is clusterssh, but its not really
> the
> > same as each node has to download from the repository. clusterssh is a
> > general tool for parallel shells.
> >
> > I am using a Redhat clone for a number of reasons, the largest being I
> was
> > told to and don't really have choice in the matter, so no chance of
> > changing distros. Scientific Linux was easy to configure and everything
> > seems to work quite well, so I am quite happy with the distro except for
> > the lack of a urpmi parallel type tool.
> >
> > Any suggestions ??
> >
> > cheers
> >
> > Tuxta
> > --
> > SLUG - Sydney Linux User's Group Mailing List - http://slug.org.au/
> > Subscription info and FAQs: http://slug.org.au/faq/**mailinglists.html<
> http://slug.org.au/faq/mailinglists.html>
> >
> --
> SLUG - Sydney Linux User's Group Mailing List - http://slug.org.au/
> Subscription info and FAQs: http://slug.org.au/faq/mailinglists.html
>



-- 
--
Teach your kids Science, or somebody else will :/

ja...@ball.net
vk2...@google.com 
callsign: vk2vjb
-- 
SLUG - Sydney Linux User's Group Mailing List - http://slug.org.au/
Subscription info and FAQs: http://slug.org.au/faq/mailinglists.html

Re: [SLUG] Managing software across a cluster

[Amos Shapira]

Just get it all under puppet (puppetlabs.com).
On Aug 14, 2012 9:55 PM, "Steven Tucker"  wrote:

> Hi all,
>
> I have just rolled out a Scientific Linux (Redhat clone) cluster of 20
> nodes plus controller that runs torque for batching and uses a lot of MPI.
> My problem is that I periodically get asked to include some library or
> program to help out with some ones research code, and often it is something
> that needs to be installed across all nodes.
>
> On my Mageia cluster I simply run urpmi parallel, issue the one command
> and it installs across all nodes. For those not familiar with urpmi
> parallel, it queries all the nodes asking what rpms are needed to install
> the given software (including all dependencies), it then downloads all the
> rpms to the machine running urpmi, it then pushes the appropriate rpms to
> each node and installs them.
>
> Is there anything like this for Scientific Linux (RHEL) and Yum ??
> The closest thing I have seen so far is clusterssh, but its not really the
> same as each node has to download from the repository. clusterssh is a
> general tool for parallel shells.
>
> I am using a Redhat clone for a number of reasons, the largest being I was
> told to and don't really have choice in the matter, so no chance of
> changing distros. Scientific Linux was easy to configure and everything
> seems to work quite well, so I am quite happy with the distro except for
> the lack of a urpmi parallel type tool.
>
> Any suggestions ??
>
> cheers
>
> Tuxta
> --
> SLUG - Sydney Linux User's Group Mailing List - http://slug.org.au/
> Subscription info and FAQs: 
> http://slug.org.au/faq/**mailinglists.html
>
-- 
SLUG - Sydney Linux User's Group Mailing List - http://slug.org.au/
Subscription info and FAQs: http://slug.org.au/faq/mailinglists.html

Re: [SLUG] Managing software across a cluster

[Richard Ibbotson]

On Tuesday 14 August 2012 21:59:00 Steven Tucker wrote:
 
> I have just rolled out a Scientific Linux (Redhat clone) cluster of
> 20 nodes plus controller that runs torque for batching and uses a
> lot of MPI. My problem is that I periodically get asked to include
> some library or program to help out with some ones research code,
> and often it is something that needs to be installed across all
> nodes.
> 
> On my Mageia cluster I simply run urpmi parallel, issue the one
> command and it installs across all nodes. For those not familiar
> with urpmi parallel, it queries all the nodes asking what rpms are
> needed to install the given software (including all dependencies),
> it then downloads all the rpms to the machine running urpmi, it
> then pushes the appropriate rpms to each node and installs them.

If you don't get anywhere with this try subscribing to the Fermilab 
list and send mail to - linux-us...@fnal.gov.  Linux Users - 
http://fermilinux.fnal.gov/maillists/

-- 
Richard

https://twitter.com/SleepyPenguin1



-- 
SLUG - Sydney Linux User's Group Mailing List - http://slug.org.au/
Subscription info and FAQs: http://slug.org.au/faq/mailinglists.html

Re: [SLUG] Managing software across a cluster

[miloska]

I used cfengine to do such things (and much more).

Let me know if you need an example configuration as a starting point.

As far as I know cfengine is not part of RHEL, so you need to install
that third party RPM first :)

Puppet and other similar tools can do this for you as well.

Regards,

On Tue, Aug 14, 2012 at 9:59 PM, Steven Tucker  wrote:
> Hi all,
>
> I have just rolled out a Scientific Linux (Redhat clone) cluster of 20 nodes
> plus controller that runs torque for batching and uses a lot of MPI.
> My problem is that I periodically get asked to include some library or
> program to help out with some ones research code, and often it is something
> that needs to be installed across all nodes.
>
> On my Mageia cluster I simply run urpmi parallel, issue the one command and
> it installs across all nodes. For those not familiar with urpmi parallel, it
> queries all the nodes asking what rpms are needed to install the given
> software (including all dependencies), it then downloads all the rpms to the
> machine running urpmi, it then pushes the appropriate rpms to each node and
> installs them.
>
> Is there anything like this for Scientific Linux (RHEL) and Yum ??
> The closest thing I have seen so far is clusterssh, but its not really the
> same as each node has to download from the repository. clusterssh is a
> general tool for parallel shells.
>
> I am using a Redhat clone for a number of reasons, the largest being I was
> told to and don't really have choice in the matter, so no chance of changing
> distros. Scientific Linux was easy to configure and everything seems to work
> quite well, so I am quite happy with the distro except for the lack of a
> urpmi parallel type tool.
>
> Any suggestions ??
>
> cheers
>
> Tuxta
> --
> SLUG - Sydney Linux User's Group Mailing List - http://slug.org.au/
> Subscription info and FAQs: http://slug.org.au/faq/mailinglists.html
-- 
SLUG - Sydney Linux User's Group Mailing List - http://slug.org.au/
Subscription info and FAQs: http://slug.org.au/faq/mailinglists.html

[SLUG] Managing software across a cluster

[Steven Tucker]

Hi all,

I have just rolled out a Scientific Linux (Redhat clone) cluster of 20 
nodes plus controller that runs torque for batching and uses a lot of MPI.
My problem is that I periodically get asked to include some library or 
program to help out with some ones research code, and often it is 
something that needs to be installed across all nodes.


On my Mageia cluster I simply run urpmi parallel, issue the one command 
and it installs across all nodes. For those not familiar with urpmi 
parallel, it queries all the nodes asking what rpms are needed to 
install the given software (including all dependencies), it then 
downloads all the rpms to the machine running urpmi, it then pushes the 
appropriate rpms to each node and installs them.


Is there anything like this for Scientific Linux (RHEL) and Yum ??
The closest thing I have seen so far is clusterssh, but its not really 
the same as each node has to download from the repository. clusterssh is 
a general tool for parallel shells.


I am using a Redhat clone for a number of reasons, the largest being I 
was told to and don't really have choice in the matter, so no chance of 
changing distros. Scientific Linux was easy to configure and everything 
seems to work quite well, so I am quite happy with the distro except for 
the lack of a urpmi parallel type tool.


Any suggestions ??

cheers

Tuxta
--
SLUG - Sydney Linux User's Group Mailing List - http://slug.org.au/
Subscription info and FAQs: http://slug.org.au/faq/mailinglists.html