«

PSU power usage

October 18th, 2010 . Posted in Hardware, PowerSaving | 3 Comments »

When I was swapping my motherboard in for a different one, I found something interesting. I used my kill-a-watt to measure how much my old motherboard + CPU + memory was using compared to my new (used) one.

The hardware
I swapped out a Pentium D 930 (yes, I know, old hardware, but it served its purpose) with a Core2Duo T5500. Impressive! The motherboards are nothing impressive (read: I forgot the brand and model), and the only thing noting was the difference in memory: from 2 GB to 4 GB. Other than that, they are pretty similar in performance (I measured with 3DMark2006: 6500-ish vs 7100-ish, respectively).

Interesting!
The interesting thing wasn’t the power usage difference between them (well, I expected a drop in power usage and got it), but the difference in power usage when I used a 400 W PSU with the new setup compared to a 500 W PSU. For the record, the only thing I changed was the PSU. No settings changed, no other hardware changed. I didn’t even run windows updates in between them. Unfortunately, I don’t have the numbers for the D930 with the 500W PSU, because I already swapped it for the other motherboard when I thought about trying that.

The details
In the table below you can see power usage that I measured (it was oscillating a bit sometimes (a few percentage of the total), I took a number that looked average). The activities are: Busy (Battlefield: Bad Company 2, ingame), Idle (Windows desktop, no programs running), Windows Standby (as the name implies), Off (shutdown, cable plugged in).

Anyway, the measurements:

Activity Pentium D 930 400W C2D T5500 500W C2D T5500 400W
Busy 210 W 145 W 120 W
Idle 133 W 85 W 69 W
Windows Standby 121 W 6 W 4 W
Off 3 W 6 W 3 W

As expected, there is some difference between the Pentium D930 and the C2D T5500 (max 115 W, min 3 W), even forget that the D930 is terrible at being Standby (I mean: 121 W? really?) but look at the difference between the same C2D with a 400 W and 500 W PSU (max 25 W, min 2 W). I guess the 400 W PSU I have is more power efficient. By the way, I rechecked 3DMark2006 for both the 400 W and 500 W PSU, both are around 7100-ish, so no big change there.

Update: 400 W PSU is FSP Group Inc, 500 W PSU is Coolermaster

Debian packages: debugging postinst

June 2nd, 2010 . Posted in Programming, Sysadmin | Comments Off

When writing debian packages, sometimes things go wrong in the postinst stage. An error like “the postinstallation script returned an error code of 1″ will occur. This is not very informative.

If you want to debug postinst scripts, you can easily do this on the machine where you (half) installed the package. Normally you can execute /var/lib/dpkg/info/foo.postinst like this:

sh -x /var/lib/dpkg/info/foo.postinst configure

However, if you use debconf, the output will be like this:
+ . /usr/share/debconf/confmodule
+ [ ! ]
+ PERL_DL_NONLAZY=1
+ export PERL_DL_NONLAZY
+ [ ]
+ exec /usr/share/debconf/frontend /var/lib/dpkg/info/foo.postinst

Which is not helpful either. However, you can add “set -x” in /var/lib/dpkg/info/foo.postinst and then just execute it, it will give you the full output.


sh /var/lib/dpkg/info/foo.postinst configure

Now the debugging process can begin!

Debian packaging: multiple packages

November 25th, 2009 . Posted in Programming, Sysadmin | Comments Off

For different projects, I’ve made debian packages before, but I’ve never had a project where multiple packages were created from one control file. Today, I had to work on a project where I had to do this. Everything went as normal, except that I could not find out where to install the files so dh_builddeb could include them in their respective packages.

When only one package is made, this is easy: just install the files in debian/tmp as you would when installing on the system. For multiple packages, each package has its own directory: debian/<package-name>. How does dh_builddeb figure out which directory to use?

Apparently, the default behaviour is to let the first package use the debian/tmp dir, and the other packages use the debian/<package-name> directory.

I would rather have them all using the debian/<package-name> directory, but I could not find an option to do this.

Update:
It seems that setting the compatibility level to 7 fixes this behaviour (echo “7″ > debian/compat), see man debhelper (section “Compatibility levels” for more info.

Munin plugin bind9

November 1st, 2009 . Posted in Sysadmin | Comments Off

Today, I was trying to get the ‘bind9′ plugin to work, but it didn’t work out-of-the-box. So I debugged and I found out that the standard bind9 plugin does not consider ‘views’. My bind9 setup has an ‘internal’ as well as an ‘external’ view. In the bind9 query log, this adds a ‘view internal: ‘ for each line, so the plugin didn’t see it as a valid query line:

client 192.168.1.1#37242: view internal: query: www.apple.com IN A +

I changed the regexp and I archive it here in case anyone else (or future-me) wants to use this:

--- bind9.orig	2009-11-01 15:06:16.000000000 +0100
+++ bind9.new	2009-11-01 15:06:07.000000000 +0100
@@ -73,9 +73,9 @@
     open(Q,"< $QUERYLOG") or die "$!";
     while (<Q>) {
 	chomp;
-	if (/client \d+\.\d+.\d+.\d+\#\d+: query\: \S+ (\w+) (\w+)/) {
-	    if ($1 eq 'IN' and $2 !~ /^TYPE/) {
-		$IN{$2}++;
+	if (/client \d+\.\d+.\d+.\d+\#\d+: (view \w+: )?query\: \S+ (\w+) (\w+)/) {
+	    if ($2 eq 'IN' and $3 !~ /^TYPE/) {
+		$IN{$3}++;
 	    } else {
 		$OTHER++;
 	    }

By the way, I also found out that there is a program called 'munin-run' which can be used to run the plugin as munin would, this saves a lot of hassle with sudo and environment settings!

kjoradsatn. omtitra eod

October 8th, 2009 . Posted in Sysadmin | Comments Off

Today, I found this in the mail. I’m not sure if I should keep this kernel (it’s a stock Debian kernel: linux-image-2.6.26-2-vserver-amd64, 2.6.26-15). Google does not give any results, seems like a corruption of log messages.

Oct  8 03:32:08 hoshi kernel: [145430.434977]kjoradsatn. omtitra  eod

edit: Oct 9
Today, I got this one, this is getting scary:

Oct  9 03:32:08 hoshi kernel: >99873]koradsatn. omtitra  eod

The only link I could find was this one: http://forums.gentoo.org/viewtopic-t-245137-start-0.html But it does not help me a lot.

When comparing the output of dmesg to the one in /var/log/kern.log (which is the source of the garbled messages), the only thing that is missing in the kern.log is this:

[236989.877537] kjournald starting.  Commit interval 5 seconds

It looks like this is the source from the garbled message: If you start from ‘k’ and take only every other character, you’ll see. So apparently, this is not a kernel problem (because the message appears ungarbled in dmesg) but rather a problem with syslogd.

Now things get really strange, as I have an rsyslogd daemon running, but I could not find the binary on the disk. It seems that I accidentally started a syslog daemon in a chroot (I did play around with that certain chroot, because I was fixing a computer that was booting via nfs-root from this machine).

The funny thing is that the rsyslogd that was running, was actually a 32-bit executable (the nfsroot is a 32-bits install), while the host is a 64-bit system. So I’ve been running a 32-bit syslogd on a 64-bit for two days :)

I’ll see if I get another garbled message tomorrow, but I think the problem is solved now :)

« Older Entries