OpenSolaris 2010.03

I’m looking forward to the release of OpenSolaris 2010.03. Among other reasons – a stable dedup engine, a simpler automated installer process, and most importantly for me, built in drivers for the dual gigabit NICs on my fileserver’s motherboard. This means I can either leave the Intel NIC in place, and have three dladm aware NICs, or move one over to my development box, Thor, which is currently running on 10/100.

I really hope Oracle doesn’t decide to kill off the OpenSolaris project. It’s doing me wonders. But with recent moves they have been making, I’m not so sure.

skipfish

Here’s another tool to add in, along side nikto and nmap – skipfish, an internal Google tool, hosted on Google Code. It’s written entirely in C++, and compiles cleanly on Snow Leopard (after libdin-1.18, which also installs cleanly).

Running the full tests against my Nagios VM took about an hour – and pushed about 4GB of data over the network to the VM. The report is comprehensive, and found two places where I’d forgot to validate my inputs for SQL inserts – which would allow for SQL injection attacks. This is very handy, and I’m going to keep it in my arsenal of security tools.

The Nagios Project

I’ve been working on a massive home project recently – writing an Ajax frontend to Nagios and NDOUtils. This similar to groundworks, zenoss and op5, but I have specific needs to address, and a lot of what I learn in doing this can be used for my work projects. The important part of this is getting a handle on AJAX – it’s been all the rage for the last few years, but I avoided it because javascript implementations differs between browsers, and it wasn’t worth the hassle for what I needed to get done.

However, I’ve stumbled across jQuery – a mature, lightweight AJAX javascript library designed for cross-browser compatibility. I’m also working to achieve graceful degradation – so if a user doesn’t have javascript enabled, or their browser only understands a subset of the javascript calls, they still get full access the tools I write.

So, all of this lives in a subversion tree, and I’m madly hacking away on it. Once I get a stable, clean demo that validates user input, I may well post a link to it.

Nagios Check Plugin For Google App Dashboard

I’ve written a Nagios check plugin for the Google App Status Dashboard. It’s written in perl, and has two dependancies: the JSON perl module (I’m calling JSON::PP, the pure-perl version), as well as a copy of the open source application curl to pull the JSON data file from Google’s servers. Curl needs to be in your default path – which it is on OSX, RedHat, Ubuntu and OpenSolaris – which are the distributions I care about these days.

If you run the check command with no arguments, it returns an UNKNOWN state. Using the -s "App Name" argument, you can specify what you want to check. If you don’t supply data with the -s flag, or misspell the name of the App, you instead get a list of available applications back – queried live from the JSON feed. This way, as more Apps are added, it’s easy to get the status of them. (Note that all of the app names have spaces in them, so you will need to use quotes around the name.)


bwdezend@godzilla:[~/] $ ./check_google_apps.pl -s "Google Mail"
OK - Google Mail is operating normally | warnings=0


The perfdata returned is the number of messages for a service that aren’t resolved. This means that each follow up message about a service issue increases the count. Once the serivce issue is cleared, the count drops back to 0, and the service returns “OK” again. The check only ever returns UNKNOWN, OK, and WARNING states. A later version will distinguish between a service disruption and a service outage, but this was all I needed at the moment.

Download check_google_apps.pl

Drive Throughput Improvement

I’ve gotta hand it to the OpenSolairs guys – they’ve done a lot of good work. When I was running an (admittedly old) build of Solaris Nevada, my network throughput would top out around 22MB/s – which is pretty respectable for a RAIDZ1 pool on consumer-level SATA disks. I was moving some VM disks around, and noticed this:

root@vault:~# zpool iostat tank 30
               capacity     operations    bandwidth
pool         used  avail   read  write   read  write
----------  -----  -----  -----  -----  -----  -----
tank        1.95T  1.46T     36      0  3.68M  17.8K
tank        1.95T  1.46T      0    324      0  39.5M
tank        1.96T  1.45T      0    511      0  62.9M
tank        1.96T  1.45T      0    501      0  61.7M
tank        1.96T  1.45T      0    485      0  59.6M
tank        1.96T  1.45T      0    482      0  59.3M
tank        1.96T  1.45T      0    498      0  61.4M
tank        1.96T  1.45T      0    509      0  62.6M
tank        1.96T  1.44T      0    495      0  61.0M

Wow. Just wow.