Continuation

It looks like the fortune has turned to me and I’ve got a job at BBC World Service Future Media. Doing Unix systems only, no MS.

So far, I love it!

Installing OpenSolaris on HP SmartArray

Solaris and OpenSolaris does not include SmartArray drivers, these need to be download from HP Website

After that, extract the .pkg directory onto a USB stick

Then start the LiveCD, open terminal and type :

 pfexec touch /ADD_DRV_IGNORE_ROOT_BASEDIR

after that install the driver :

 pfexec pkgadd -d path/to/pkgdir

Now you can start installation, it should see the array.

I was waiting whole Monday to get my hands on a new release and at the end of it I got it! Very , very long wait it was. Pity I don’t have time to play with it until I’m back from holiday. Even then I will not have much time strait away…

I am quite clued on Solaris, thanks to a friend of mine and now thanks to OpenSolaris project I am starting to like it.

Solaris suffers from one big problem … docs.sun.com is not that clear to get through for newbies. However I’ve got an OpenSolaris Bible for my PFY. It looks like more approachable paper version of docs. Handy.

It looks like S.M.A.R.T has told me that drive in my server is failing. Actually I’ve noticed that when my NFS server just stopped sharing data from it. Still, it was a bad sign, quick look at logs and clear failure messages. Cool.

Short trip to a highstreet shop – Maplin and I’ve grabbed a 1TB Hitachi ( I will never buy Seagate again ). I’ve picked up an Air Duster and 12cm fan as well just as I was there. Fan is for my new PC to keep disks cool.

Open the server, loads and loads of dust, Air Duster gets to work ( someone will kill me as it all came out and dispersed in a nice clean living room ). Swapping the drives. Starting the server.

It takes a bit longer for the server to startup as it was not restarted in ages and all timers told it to rescan all systems properly this time.

Starting Solaris … and my volume with missing disk is … OK … weird. Started zfs scrub on it , just to make sure. Finally it found out that it’s missing a copy, volume is degraded. zfs replace and the rebuild starts, tells me it will be an hour to mirror 750GB back on a new drive.

Now this failure got me thinking if I should replace my VM Host , it’s currently OpenSuse 10 with ancient version of Xen.

I’m wondering if I should put an ESX hypervisor or something else. Certainly my experience with ESX is that it’s rubbish at storage management. I am leaning towards OpenSolaris as it has a selection of hypervisors and storage management just proven to be a breeze.

Will see …

A glimpse of VirtualBox and Sun Unified Storage Simulator.

SUSS looks good, Only problem I had was with IPv6, donno if it was due to VBox or SUSS but when I’ve set a static IPv6 it was crashing the interface. Other than that, good stuff, will need to test both products more.

After doing converting one of my servers (baptop) from a laptop to Virtual Machine on Xen with full virtualization it didn’t work too fast. Virtual environment was too penalizing. It was less reactive and when calling ( it hosts Asterisk for VoIP )  sometime you we could hear each other for a few seconds.

Now, after failed play with Nexenta Core and Xen DomU, I decided to have a go at para virtualizing the baptop. After a few kernel versions and config files … it works. Still without console but works. Great!

I really wanna try that Debian on Solaris kernel called Nexenta. Unfortunately it looks that it can’t do DomU and ZFS_ROOT at the same time. I have filled a bug report with Nexenta.

After, not much over 1 year, I am retiring my Media Vault .

Having a network attached storage is a briliant idea. I am staying with it. It’s just the Media Vault product.

Slow death. After half a year, a small fan has broken and since then it became loud, I have contacted HP to get it replaced, but they asked me to ship the whole unit to them, even when I was perfectly fine to replace it my self. I said ˇNo, thank you˝. That was my first experience with HP Support for dummies (home user). Months later, I have moved to a new place with a dedicated computer closet and I bought a proper server, the ML110. From there, I started experimenting with OpenSolaris, which with ZFS and NFS – SMB sharing, it’s soo much better and extremely flexible solution.

Two 750GB drives later I have my first Sol(aris)NAS. Very nice creature and actually faster than the MV. Last week I started migrating data on to the SolNas, and today I’ve finished migrating clients to it. Nice!

MV is switched off, my energy consumption lost a 35W sucker and I have no idea what to do with it now. Just gonna leave it where it is. Maybe you want it ?

If you ever heard about monitoring systems it was usually some expensive closed source software or the open source package. I have little experience with the first type.

For some thoughts about the second type, read more

Read more…

One word: Nightmare!

How on Earth, can I install SVN/Subversion on that bloody operating system , so it will include bindings for either Perl or Ruby ?! fighting with it for last two days… finally got source to at least compile  … but even if I have Ruby ( from CSW ) and Perl ( by Sun ) I just can’t get binding for them …

Why it can’t just work… like on Linux distributions ( Ubuntu/Gentoo , you name it … ) …

The other thing, this Solaris is running inside Microsoft Virtual Server… on a decent machine … but it is slow like Dell PE2400 which I use for concept testing … and no, VS machine is not taken over by other VM …

I need to setup a system which uses SVN for storing ‘website’ revisions and automatically updates live/review/development systems whenever there is commit. Adding a little bit of complication each server is updated from different branch.

If website serving boxes would be any kind of UNIX, this post would not exist… way too easy.

Web servers are Windows boxes  and they are located in different networks, so no SAMBA.

My ideas so far :

Naturally using hooks in repository…

1. Using rsync
   This idea was to : whenever content of a branch is changed, branch is exported locally on SVN server, rsync it to Web server.
2. Using FTP
   similar to rsync , instead of using rsync for file transport we would use FTP
3. Using SVN+HTTP
   Here it gets lil bit more complicated ,  after branch changed, svn server sends HTTP request to web server, which fires svn up .

Each of an ideas has its minuses, 1 uses precious svn server disk space for second copy, requires web servers to run rsync servers. 2 uses disk space on svn and abuses a lot of bandwidth every time when something changes. 3 requires web servers to have svn clients installed and custom CGI/ASP/Whatever which will trigger svn updates on local copies.

There is one more, using scripting language to write FTP/SVN integration : something like ftp client which instead of uploading from File System it would upload from SVN … that looks good! will have to research on which tools I can use to do that!