www.gregd.org

One Million Websites Infected during the second quarter of 2010! It pays to keep your software up to date!

I saw a Dilbert today that is well, funny and true… odd how dilbert always seems to bring those two elements together….

At a recent meeting… If you start at zero, then we’re number one!

Your reasoning is excellent — it’s only your basic assumptions that are wrong.

Yay, the new PCIe card is in and working great – running at over 850Mbps in both directions and a 9K MTU in linux In windows, the performance is around 750Mbps…  Overall, that’s ~4x faster than the first gigabit cards i tried, and a full ~8x improvement over my old 100Mbps (fast ethernet) network.

Now I’ve just got to get my raid array under control… it’s rebuiding again… i accidentally put a md raid super block on sdb a couple days ago. This confused udev very much. After I created the raid there, i removed it and recreated it on the first partition… (Where i meant to put the raid in the first place). Even though the disk had been partitioned, i still needed to zero the super block on the device itsself before udev would find the first partition and realize that the raid was there instead. Of course, I didn’t notice this problem until i rebooted to replace the network card So once the rebuild finishes we’ll get to see how fast SAMBA runs.

Hmm, looks like as long as the PCIe slot is big enough, any card will work. For example, I just ordered a EXPI9300PTBLK which is a PCIe x1 card – and it is going into the one (and only) PCIe x16 slot in my server!

That aside, I spent the past week running a very real DR drill – and things are looking good, but there were a few hiccups. The drill was caused by me playing with my raid 1 array. I converted it to a linux style “10 f2″ array which basically doubles read performance at the expense of write performance. It’s still a 2 disk mirror, but now it is striped so we can read two disks at once.

Of course, I did the raid “upgrade” the silly way and didn’t bother to bring one disk over at a time into a new array – instead I just dropped the existing md0 array and recreated it. Bye bye data. The worst part was restoring the vmware server, but it really wasn’t that bad since I had a full list of packages and a copy of the /etc folder Otherwise, it was just waiting a few days for the backup to copy from my parents place back to my apartment.

How does this all relate? Well, I upgraded to a gigabit lan and after trying some pci cards (Linksys EG1032 and a Netgear GA311) based on the Realtek RTL8169S-32 chipset that seems to be limited to ~218Mb/sec upstream instead of getting closer to the gigabit speed of 1000Mb/sec (actual should be around 700Mb/sec maybe a bit better or worse). Not to mention that this chip is limited to a 7200 byte MTU (instead of the standard 9000). I thought I was reaching a HD physical limit – so improving the raid performance seemed to be the answer (and still is). However, that wasn’t the only limit – when i measured with iperf, I found that the network card could stream around 218Mb/sec up and take in around 420Mb/sec down…

Psalm 4:7,8
You have filled my heart with greater joy than when their grain and new wine abound. I will lie down and sleep in peace, for you alone, O LORD, make me dwell in safety.

I overclocked my WRT54GS v2.0 to 300 mhz! This poor router has been customized in just about every way possible. In addition to the OpenWRT Kamikaze firmware, it also sports two high gain antennas. Yes, this is the router that once covered the whole chapel (barely). I suspect that the people across the street at the DMV are now seeing “R2D2″ in their list of access points

This router has a stock speed of 200 mhz and the 50% speed increase is quite noticeable when I log into the router via ssh. Yup, still waiting for it to crash…

Read the rest of this entry »