Opensolaris & VirtualBox 3.1

Keeps locking up my system. Appending snooping=1 to /etc/system enables the solaris dead man timer and might yield a crash dump. We'll see ;)

clouds, agile infrastructure and tools

We're currently using puppet to setup our physical and virtual (zones & vmware vms) solaris boxen. Puppet is a step into the right direction but only part of the solution - a handy tool to describe, setup and manage single systems, replacing the trusty old shell scripts every administrator has in his tool belt.

With puppet, you can describe how your application server should look, how your database server should look but not how this two interact.

You can't describe relationships between two or more puppetized hosts.

You can't tell puppet that the application server service should not start until the database is ready.

There are tools which are able to handle this cases. Capistrano and ControlTier come to mind.

clean shit

http://grulos.blogspot.com/

zoned, centralized solaris container management

Did some design work, speced the overall architecture and wrote a proof of concept / throw away prototype which you can find at:

git://github.com/robertgstoehl/narfd.git

My ruby is quite rusty, good to have something to work on again. Why ruby you may ask. Well puppet, my favorite tool at the moment, is written in ruby, I like the smell of the language more than that of my code and last but not least - sex sells.

Feedback as always welcome.

Puppet gets an internal DSL

http://groups.google.com/group/puppet-dev/browse_thread/thread/842957b5c61ccbc3/731b0dcf07d23ad1?lnk=gst&q=shadowpuppet#731b0dcf07d23ad1

crossbow network virtualization

dladm create-vnic -l e1000g0 vnic1
dladm create-vnic -l e1000g0 vnic2
dladm show-vnic
zonecfg -z zone1
zonecfg:zone1> create
zonecfg:zone1> set zonepath=/export/home/zone1
zonecfg:zone1> set autoboot=true
zonecfg:zone1> set ip-type=exclusive
zonecfg:zone1> add net
zonecfg:zone1:net> set physical=vnic1
zonecfg:zone1:net> end
zonecfg:zone1> verify

Works as expected. Good stuff better late than never...

OpenSolaris /dev repository

Upgrading right now. Can't wait to get my little dirty hands on the crossbow bits. If all goes well I'm back in a few minutes.

osol 2008.11 ips network timeout

I'm running OpenSolaris 2008.11 in VirtualBox at work. My workstation only has access to the internet and the ips authorities through a proxy. So setting the following environment variables initially worked:

export HTTP_PROXY="http://username:password@proxy:port"

But trying to install larger packages (ruby) resulted in network timeouts and didn't really work. Today i stumbled upon two undocumented variables which finally solved my issues.

PKG_CLIENT_TIMEOUT number of seconds to wait before socket gets timed out

PKG_TIMEOUT_MAX number of attempts

Ready to do some puppet source code hacking (ok building solaris packages more up to date then the ones in blastwave).

Slappy New Year

Hello World,

After an exhausting cutover week at my workplace i've got 2 days off and with it some time to blog. Since 2 years I'm a systems engineer at a financial institution working with big sun iron and large oracle deployments. Since i graduatet with bachelor degree in software engineering i learned a hell of lot of things in the last two years. Some examples:

• Enteprise infrastructure is expensive
• Solaris has bugs, lots of it
• Oracle has bugs, lots of it
• SOA isn't quite what i learned in school about it
• Java VMs core dump too

More of this and giant robots later.