Comments for Myblog Nodeofcrash.com http://www.nodeofcrash.com Blog of Simon Brennan Tue, 10 Aug 2010 05:52:03 +0000 hourly 1 http://wordpress.org/?v=3.0.1 Comment on Puppet, xCAT and a big HPC Cluster by Simon http://www.nodeofcrash.com/?p=353&cpage=1#comment-6325 Simon Tue, 10 Aug 2010 05:52:03 +0000 http://www.nodeofcrash.com/?p=353#comment-6325 My plan is to use xCAT to do the basic provisioning of the nodes and part of that provisioning process is installing the puppet client and performing the initial configuration. Once that is done puppet will take over and complete the setup, configuring ldap, nfs mounts etc. I'm looking at using the xCAT databases and hooking that into puppet, for example generating the node groups etc. At this moment in time I'm not sure if integrating puppet heavily into xCAT is worth the effort. My plan is to use xCAT to do the basic provisioning of the nodes and part of that provisioning process is installing the puppet client and performing the initial configuration. Once that is done puppet will take over and complete the setup, configuring ldap, nfs mounts etc.

I’m looking at using the xCAT databases and hooking that into puppet, for example generating the node groups etc.
At this moment in time I’m not sure if integrating puppet heavily into xCAT is worth the effort.

]]> Comment on Puppet, xCAT and a big HPC Cluster by Vallard http://www.nodeofcrash.com/?p=353&cpage=1#comment-6256 Vallard Fri, 06 Aug 2010 23:39:19 +0000 http://www.nodeofcrash.com/?p=353#comment-6256 I'd be interested to see how you use Puppet in your xCAT environment and if you knew of any ways the two could be more integrated, or if it makes any sense to do it or not. I’d be interested to see how you use Puppet in your xCAT environment and if you knew of any ways the two could be more integrated, or if it makes any sense to do it or not.

]]> Comment on lm-sensors w83627ehf problem on Centos 5 by lm sensors http://www.nodeofcrash.com/?p=71&cpage=1#comment-3751 lm sensors Tue, 16 Mar 2010 13:53:24 +0000 http://www.nodeofcrash.com/?p=71#comment-3751 [...] version 2.5.4 and edited the config.php file to enable lm-sensors hardware monitoring. ...lm-sensors w83627ehf problem on Centos 5 Myblog Nodeofcrash.comlm-sensors w83627ehf problem on Centos 5. I've got a ASUS P5KPL-CM motherboard I've ... [...] [...] version 2.5.4 and edited the config.php file to enable lm-sensors hardware monitoring. …lm-sensors w83627ehf problem on Centos 5 Myblog Nodeofcrash.comlm-sensors w83627ehf problem on Centos 5. I've got a ASUS P5KPL-CM motherboard I've … [...]

]]> Comment on My head in the clouds by Simon http://www.nodeofcrash.com/?p=307&cpage=1#comment-3731 Simon Sun, 14 Mar 2010 20:53:02 +0000 http://www.nodeofcrash.com/?p=307#comment-3731 Yes a private cload.... ummm... okay it's a spelling mistake, you spelling Nazi :p I guess it's down to what we actually want to run. Alot of the researchers use code that is not particularly well written, is a pain in the rear to install, especially on our Novell SLES based systems. An application called "meep" which is a free finite-difference time-domain simulator, took nearly two months to get working on SLES, but about 5 minutes on Ubuntu. We use SLES because that is what the vendor supports. So it would be easier to create an ubuntu virtual machine and run it on the virtual infrastructure. We are actually in theory running a "platform as a service" cloud effectively at the moment on our existing system. But maintaining it and compiling the code is a nightmare. So the idea behind switching to a VM approach was to allow researchers to use a distibution of Linux that has the code they want to use in precompiled debs rpms etc. I was looking for a framework to base the system on, and Amazon based cloud frameworks seem to be springing up everywhere. All I really want is an easy way to provision and manage the whole thing easily. There are other solutions, but most require a heap of money, like Vmware virtual infrastructure. That Azure service looked interesting, we don't run very much Windows code. Google App engine is good too. However it all needs to be done in house. Out sourcing things to Google would probably work, but the people who fund us would frown upon outsourcing to the US. Yes a private cload…. ummm… okay it’s a spelling mistake, you spelling Nazi :p

I guess it’s down to what we actually want to run. Alot of the researchers use code that is not particularly well written, is a pain in the rear to install, especially on our Novell SLES based systems. An application called “meep” which is a free finite-difference time-domain simulator, took nearly two months to get working on SLES, but about 5 minutes on Ubuntu.
We use SLES because that is what the vendor supports.
So it would be easier to create an ubuntu virtual machine and run it on the virtual infrastructure.
We are actually in theory running a “platform as a service” cloud effectively at the moment on our existing system. But maintaining it and compiling the code is a nightmare.
So the idea behind switching to a VM approach was to allow researchers to use a distibution of Linux that has the code they want to use in precompiled debs rpms etc.

I was looking for a framework to base the system on, and Amazon based cloud frameworks seem to be springing up everywhere. All I really want is an easy way to provision and manage the whole thing easily. There are other solutions, but most require a heap of money, like Vmware virtual infrastructure.

That Azure service looked interesting, we don’t run very much Windows code.
Google App engine is good too. However it all needs to be done in house. Out sourcing things to Google would probably work, but the people who fund us would frown upon outsourcing to the US.

]]> Comment on My head in the clouds by Robert MacLean http://www.nodeofcrash.com/?p=307&cpage=1#comment-3728 Robert MacLean Sun, 14 Mar 2010 15:38:05 +0000 http://www.nodeofcrash.com/?p=307#comment-3728 "Cluster into a private cload" - Is that the ozzie way to say cloud? Sounds a bit scottish to me ;) Why are you looking at the Amazon (host your VM here model)? Have you looked at the Azure or Google Appengine models? Not saying you have to run their systems, but rather the difference in design. Amazon is fairly low level, where Azure (currently) does not give you a VM, but rather an interface and the ability to run PHP, .NET or unmanaged code (think of it the next level of abstraction over Amazon), and AppEngine goes even further giving you their own API (another level of abstraction over Azure). No one is perfect for everyone in this space, but I would love to hear why you think the Amazon route would work best? “Cluster into a private cload” – Is that the ozzie way to say cloud? Sounds a bit scottish to me ;)

Why are you looking at the Amazon (host your VM here model)?
Have you looked at the Azure or Google Appengine models? Not saying you have to run their systems, but rather the difference in design.

Amazon is fairly low level, where Azure (currently) does not give you a VM, but rather an interface and the ability to run PHP, .NET or unmanaged code (think of it the next level of abstraction over Amazon), and AppEngine goes even further giving you their own API (another level of abstraction over Azure).

No one is perfect for everyone in this space, but I would love to hear why you think the Amazon route would work best?

]]> Comment on SLES 10 compatible nrpe init.d script by Simon http://www.nodeofcrash.com/?p=257&cpage=1#comment-2891 Simon Tue, 05 Jan 2010 11:24:40 +0000 http://www.nodeofcrash.com/?p=257#comment-2891 Sure thing. A text version is available here: http://www.nodeofcrash.com/wp-content/uploads/2009/11/nrpe.txt Sure thing.
A text version is available here: http://www.nodeofcrash.com/wp-content/uploads/2009/11/nrpe.txt

]]> Comment on lm-sensors w83627ehf problem on Centos 5 by Simon http://www.nodeofcrash.com/?p=71&cpage=1#comment-2890 Simon Tue, 05 Jan 2010 11:15:12 +0000 http://www.nodeofcrash.com/?p=71#comment-2890 You're welcome :) Glad it helped you out. You’re welcome :)
Glad it helped you out.

]]> Comment on SLES 10 compatible nrpe init.d script by Desi http://www.nodeofcrash.com/?p=257&cpage=1#comment-2768 Desi Fri, 18 Dec 2009 20:02:37 +0000 http://www.nodeofcrash.com/?p=257#comment-2768 Can you post a version that doesn't have fancy formatting (regular quotes instead of beginning and ending quotes for example) so I can copy and past this thing rather than retyping it? Or a link where I can download the file in plain text? Thanx Can you post a version that doesn’t have fancy formatting (regular quotes instead of beginning and ending quotes for example) so I can copy and past this thing rather than retyping it?
Or a link where I can download the file in plain text?
Thanx

]]> Comment on lm-sensors w83627ehf problem on Centos 5 by GottaLuvIt http://www.nodeofcrash.com/?p=71&cpage=1#comment-2680 GottaLuvIt Thu, 10 Dec 2009 19:49:45 +0000 http://www.nodeofcrash.com/?p=71#comment-2680 I compiled the kit tarball on 2.6.18-164.6.1.el5. When I tried to load it, got an Unknown symbol error. Ran depmod and everyone is happy! Thanks for your efforts! I compiled the kit tarball on 2.6.18-164.6.1.el5. When I tried to load it, got an Unknown symbol error. Ran depmod and everyone is happy!

Thanks for your efforts!

]]> Comment on IT Support and Home Renovations…. What?? by Robert MacLean http://www.nodeofcrash.com/?p=259&cpage=1#comment-2555 Robert MacLean Wed, 02 Dec 2009 06:36:56 +0000 http://www.nodeofcrash.com/?p=259#comment-2555 Were you driving and using your mobile at the same time? Breaking the law again huh? Were you driving and using your mobile at the same time? Breaking the law again huh?

]]>