Doomshammer's Weblog

• Comments (0)
• Trackbacks (0)

Just btw... I quit my current job and will start at another company on 1st of december. Looking forward to it If you are interessted in a UNIX administrator job in cologne- have a look at the job announcement for my current position.

Before you ask.. I'd like to keep the new company still secret- so as I know that some of my co-workers are reading my blog, I won't tell about it yet - will come soon

• Comments (2)
• Trackbacks (0)

I am currently monitoring all my servers with Nagios and I use Cacti to make some fancy graphs of these hosts. As I soon have to disable the server, from which I do the monitoring, I was looking for a new solution. I already read some good articles about Zabbix, so I installed Zabbix on a new server.

The installation was fairly simple. Apache + PHP5 + PostgreSQL - quickly compiled and setup. Now download the Zabbix sources from the Zabbix website and compile the server. Copy the frontend into your webroot and perform the rest of the installtion steps - that's it.

The concept of Zabbix is a bit different from Nagios. It it mainly based on it's own agent daemon (which has to be installed on every host that you wanna monitor), where nagios is rather based on some checking tools that are run against the host over snmp or other service ports (check_http for tcp/80/443, check_ftp for tcp/21, and so on).

At the first glance Zabbix is a bit confusing, but after some reading in the manual you quickly get the point about triggers, actions and stuff. What I like about Zabbix is, that it combines Nagios + Cacti + Content Validation. So I can easily montitor all my hosts (Windows, Linux, Solaris ...), generate some graphs and perform website validation. The frontend concept of Zabbix is very mature. I'll show you some brief screenshots to get some overview of it.



The overview page. It lists all configured hosts (currently only 3 as I didn't had the time last night to add all of them). It shows all configured triggers together with a little status box for each host. If the status box is green, everything is fine- is it red... go ahead and check your host


The web content validation screen. You can configure content validation suites for every host. You can even go through a i. e. complete order process on a website and if one of the steps fails, you'll get informed by Zabbix.


The latest data screen. Thanks to the agent that is running on every host, you can easily access lots of host informations w/o having to configure ugly SNMP suites. The client provides lots of data about the host like performance, OS, cpu, network, and so on.


On-the-fly graphs. For nearly every data you can create graphs - on-the-fly. You can define the time period, you can define the look of the graph. You can also create graph-suites which a building graphs automatically. With this tools you can easily graph the cpu load of your hosts, the network traffic and so one.

There are lots of more features with Zabbix. If you want an easy to use but powerful distributed monitoring system, you should give Zabbix a try.

• Comments (4)
• Trackbacks (0)

Not much to say... the topic says it all. Read more here.

• Comments (0)
• Trackbacks (0)

Today I got two new hard discs for my local file server (thanks to my mate Tommy for the discs ). So I quickly mounted them into the case of my server and booted the box.

As expected the discs were dected as well by the BIOS as by SunOS. So all I had to do, was adding them to my current ZFS data pool- which was done in less than two minutes.

Verify the OS detected the discs:

There we are... c1d1 and c2d1 are the new discs (currently w/o label, that's why the drive type is unknow)

Check the pool status:

Everything's fine- as expected

Add the new discs to the data pool:


Done! Now have a quick look, if everything has been added and that's it!


Great! Isn't it?!

• Comment (1)
• Trackbacks (0)

In his blog, Joerg Moellenkamp points to an interessting Benchmark. ZFS vs. Veritas' VxFS. Have a look at it- it's worth reading.

• Comments (0)
• Trackbacks (0)

I found an interessting article at the MySQL Performance blog, pointing me to these benchmarks:

• MySQL vs. PostgreSQL
• RDBMS on Solaris vs. Linux

It's pretty interessting to see, that PostgreSQL is performing very well, where MySQL is suffering very quickly with increasing concurrency. Also the 2nd benchmark shows pretty good, that PostgreSQL and Solaris is a good team, and Linux/MySQL combination is also scaling very bad.

Another confirmation that I'm going the right way with my SunOS/PostgreSQL combination

• Comments (0)
• Trackbacks (0)

Here Joerg Moellenkamp developed an idea for using ZFS as kind of "version control system" e. g. for a website. IMHO a pretty neat idea!

• Comments (0)
• Trackbacks (0)

Even if the data I used in this example isn't reasonable, the result is pretty impressive though:
Creation of a 1 GiB file w/o zfs compression/filesystem cache:



After doing a "zfs set compression=on data/postgresql":



More CPU usage (by the filesystem cache) but the write operation was 6 times faster as it was performed in memory before the physical write operation has been performed.

• Comments (0)
• Trackbacks (0)

I am currently playing around with ZFS on my SunOS File Server. ZFS is a new developed file system (initially called: Zettabyte File System). ZFS is an easy to administer but mighty file system. Find the basic administration guide here.

I just created a ZFS for my /data directory. It is pretty easy, as the zfs commands do everything needed by theirselfs. Just do s. th. like this to create a pool: "zpool create <name> mirror c1d0s5 c2d0s5" (I created the zfs on a slice, as I don't have two additional disks in that server to accomplish the recommended creation over a entire disk). And there you are. Zpool checks the disks, creates a filesystem and mounts the newly created file system. No I created a sub-filesystem for my user with a quota of 35GiB which is pretty simple as well. "zfs create <name>/doomy; zfs set quota=35G <name>/doomy" and you are ready to go. Impressive! I'm going to play around a bit more with it and will let you know about my experiences (if I'll get some ).

• Comments (0)
• Trackbacks (0)

As PostgreSQL sadly doesn't brings a smf manifest with it, so that you can include it as SMF service on your Solaris 10 box, I've written one by myself. I've uploaded it to my blog, so if you are running PostgreSQL on a Solaris >= 10 box, feel free to download and use it. Simply untar the file, adjust the paths and run:



Find the manifest (including a start script) here.

• Comments (0)
• Trackbacks (0)

• Comments (0)
• Trackbacks (0)

In this article on his blog, AK shows some screenshots of his new rss feedreader "noos". To me it already looks pretty nice and I'm looking forward to give it a try soon.

• Comments (0)
• Trackbacks (0)

As you could read here, I set up my old server as SunOS file server. So as this box is running fine now, I wanted to "backup" my photos from my Laptop (Windows XP) to the fileserver but without moving the photos and without copying them completelly every week. So I quickly installed ssh and rsync for my cygwin installation, created an SSH key and put it to the SunOS box. Now I can easily keep my photos in sync with this simple line:

• Comments (0)
• Trackbacks (0)

[root@suppe:/etc] # rm * /var/log/
zsh: sure you want to delete all the files in /etc [yn]? y

Thanks echox! You made my day!

• Comments (0)
• Trackback (1)