These new servers are very fast. Applications that usually take their time to respond on hotrod or wheeljack are instantaneously served now. Cacti, which generates a large range of images on the fly, and lambie.org, which makes use of a remote database – both are are blindingly fast.

Click, pow! Click, pow!

There appears to be a problem with virtualizaiton in general that sees host systems’ clocks slip, sometimes quite drastically.

I saw our time adjust to 1987, then 2259 in the last few days. I’ve also experienced some issues with prime and jazz where they wouldn’t let me in. Releated? Definitely. 

These two machines are the only servers we’ve deployed with multiple virtual CPUs. As such, they’re using a different clocksource to the single CPU machines – tcs vs. acpi_pm.

mlambie@prime:~$ sudo cat /sys/devices/system/clocksource/clocksource0/current_clocksource
tcs

I have forced out multi-CPU virtual machines to use acpi_pm instead of tcs by altering the kernel parameters in the menu.lst file found in /etc/grub. The relevant parts are:

## ## Start Default Options ##
## default kernel options
## default kernel options for automagic boot options
## If you want special options for specific kernels use kopt_x_y_z
## where x.y.z is kernel version. Minor versions can be omitted.
## e.g. kopt=root=/dev/hda1 ro
##      kopt_2_6_8=root=/dev/hdc1 ro
##      kopt_2_6_8_2_686=root=/dev/hdc2 ro
# kopt=root=UUID=0fac0e90-1edf-420c-a0e6-c8b0e85ea6eb ro clocksource=acpi_pm

Note that you’r root UUID will be different, and also the first hash is necessary.

Regenerate your boot config with:

mlambie@prime:~$ sudo update-grub
Searching for GRUB installation directory ... found: /boot/grub
Searching for default file ... found: /boot/grub/default
Testing for an existing GRUB menu.lst file ... found: /boot/grub/menu.lst
Searching for splash image ... none found, skipping ...
Found kernel: /boot/vmlinuz-2.6.24-19-server
Found kernel: /boot/memtest86+.bin
Updating /boot/grub/menu.lst ... done

I hope and expect this will fix the clock drift, and will be forcing our single-CPU machines to stay with acpi_pm, even though they’re using it at the moment.

Update:

It appears to have solved the issue – 8 hours and no driftin’.

I’ve moved lambie.org over to prime and prowl, our new Apache and MySQL servers. It all appears to be working properly.

Let me know if anything is broken or missing and I’ll sort it out.

With our new server deployments we’ve had some unexpected, positive, consequences.

One such side effect is that because we now share our home directories over an NFS mount, when we copy our SSH key (to allow password-less login) to the first server, all the other servers automatically know how to authenticate us too.

Update:

Don’t forget to combine this with ssh-agent so that once you’ve connected to a server that can authenticate you you’ll never enter your password again. In this case, I never enter passwords – full stop.

As most of you are probably aware, I quite like children and tend to give them my time and attention. It’s no different on the jiu-jitsu mats, which has the possibility of being misconstrued by someone that doesn’t know me as well. What would you think about a 28 year old man that is always happy to train with the kids?

I wonder if it’s worthwhile me completing a Working with Children Check, or if that sends an even worse message. Would I be considered a “volunteer who works with children ages 0-12 years” when I’m training at the club with children I wonder?