[prev in list] [next in list] [prev in thread] [next in thread]
List: bioclusters
Subject: Re: [Bioclusters] Debian cluster
From: Nox <pheusion () snet ! net>
Date: 2003-09-19 20:51:07
Message-ID: 1064004666.782.588.camel () shaolin ! gridlocksecurity ! com
[Download RAW message or body]
Wow, nice info, thanks,
Ill pass this onto my developer.
Thanks again
On Fri, 2003-09-19 at 16:25, Joseph Landman wrote:
> Having written automated installation tools for IRIX (Autoinst, the
> predecessor to Roboinst, used for a number of years), and looking at
> things like SI and other imaging tools, I personally prefer the
> non-image based solutions, as they do not introduce this new layer of
> management you have to worry about (image management) and debug.
> Imaging is great for some things (laptop loads, etc).
>
> FAI looks like they do some things right. It looks like they do not
> have an automatic method of BOOTP mac address harvesting though. That
> could be quite painful on a reasonable sized cluster. There are a
> number of ways to handle that, including using File::Tail on the
> /var/log/messages for your dhcp server. From this, you can have a
> script automatically build your dhcpd.conf file, populate your
> /etc/hosts, etc.
>
> Note: Autoinst booted IRIX into a diskless mode, handled all the disk
> partitioning and file system formatting, package installation, etc.
> Rules built in to it (much like the FAI classes) allowed for multiple
> mixed hardware types with package/layout variations. Pissed the network
> people off though... it was faster to use this than to do a fresh
> install via CD. Imaging over the network was marginally faster to get
> the image onto the disk, but then you had to do the post image
> processing to make the machine unique, and sometimes (in those days) the
> image transfer was not successful, so you had to do it again over poor
> networks. The package method returned a code as to whether or not the
> package install succeeded, so that inst could handle problems. All in
> all the package method turned out to be significantly better in terms of
> ease of use and reliability.
>
> Please let us know how it goes.
>
> On Fri, 2003-09-19 at 00:52, Nox wrote:
> > We have begun to investigate FAI and Debian for our clustering
> > solutions.
> >
> > I have always been partial to Debian, its my OS of choice,
> > has been for yrs.
> >
> > So when I discovered the clustering tools, I was excited.
> > (FAI, Fully Automated Install)
> >
> > Does anyone have experience with this?
> > Pitfalls, things to watch for,
> > success stories etc.
> >
> > from my perspective:
> >
> > Debian offers ease of use
> > Updates are a snap (apt-get) --Including security
> > Ability to be a very minimalist system
> > Quick access to Biotech tools (bioperl as an example)
> >
> > We currently use debian as our Halted Firewall as well.
> >
> > Thanks in advance
> >
> > Nox
> > GenMicro Systems
> >
> > _______________________________________________
> > Bioclusters maillist - Bioclusters@bioinformatics.org
> > https://bioinformatics.org/mailman/listinfo/bioclusters
[prev in list] [next in list] [prev in thread] [next in thread]
Configure |
About |
News |
Add a list |
Sponsored by KoreLogic