Improve the Debian boot process - blog

With the Post-etch frenzy, I installed Debian 4.0 stable as soon as I could. I used the same system used a few months ago for the google Summer of Code. Under the same conditions, the system required 32 seconds until KDE started. Compared with the previous stable releases:

• Woody - 32 seconds
• Sarge - 44 seconds
• Etch - 32 seconds

Nevertheless, Etch has significantly more process started at boot time and the CPU usage is more intensive --with still some room for improvement --. Follow this link for the bootcharts and more detailed information.

As I starting point, using dash instead of bash, the boot time is reduced to 30 seconds and it could be further improved but there is some "hald" process (maybe part of dbus) that doesn't seem to want to run in parallel. See bootcharts. Any ideas?

Labels: linux debian boot

It is funny to see that when I remove discover (/etc/rcS.d/S36discover) from the boot, a disk utilisation and cpu usage peak appear at the end of loading KDE. Would it be that discover is loaded by KDE if it wasn't loaded? Or a discover ghost :p. Well, the bootcharts for the two different cases are available here.

I've tried prelinking with the aim of making the boot faster. I've used the package of prelink from debian in unstable and we get no significant time improvement. Prelinking was added after readahead in a custom script run during the boot in /etc/rcS.d/S70prelink. What the script does is to prelink: kdeinit, Xorg, kwin, kdesktop, kicker, artsd, kaccess, ktip, klipper and korgac. The bootcharts are availabe here.

By removing useless init-scripts from the symbolic link pool (/etc/rc?.d) we got a further second improvement in the boot time. The two init-scripts removed were hotkey-setup and gdm. The former can be removed as I'm not using a laptop and the latter because I'm using KDE. The boot time was reduced from 46 to 45 seconds. See the new bootchart and compare with the reference bootchart.

I've been playing a bit with Ubuntu's readahead to get a better boot time on debian. At least they seem to get it. I've used strace to get the list of files for different init-scripts and generate the required /etc/readahead/boot file - containing the files to load to memory. The advantage of readahead over preload so far is that it does not affect the boot time when it is not active. I consider readahead inactive when the boot file is empty and preload when the preload script is removed from the symbolic link pools (/etc/rcX.d).

By using readahead shortly after the system starts to boot (near udev), the boot time is even increased by 2 seconds (e.g. S03readahead in /etc/rcS.d). On the other side, by using some of the files used by KDE during startup there is almost a one second improvement as may be seen in the bootchart. We use readahead in /etc/rcS.d position S42. The list of files and the bootcharts may be found here.

Besides, I've corrected the deliverable 5 including the fact that lintian incorporates now a check for lsb compliance based on our patch! The deliverable is available here

With the alioth sever down, it seems our project webpage is down as well. Anyway, we are currently on a 7 second improvement of the boot time by using reordering, dash as /bin/sh and hwclock in the background. The bootchart is available here while that of the original system here. No time was won using preload 0.4 with varied configuration combinations. Curious to see that the time is 2 seconds more even without preload in the symbolic link farm (/etc/rcX.d) when preload is installed.

The new deliverables 3 and 5 were published. In the third deliverable we mention the effect of most hotspots, the procedure used for testing them and their bootcharts. Besides, it contains a group of combinations done with the hotspots and their effect. Deliverable 5 simply consists of the LSB guide for maintainers and the lintian patch to check LSB compliance. Both issues were published some weeks ago and this deliverable was just published to comply with the original deliverable list. Still two deliverables are missing: Deliverable 4: a proposal to change the boot system in debian according to the results from Deliverable 3 and further research, and Deliverable 6: the final report of the project. They are available in the deliverables webpage

Some time ago I posted having obtained a 2 second time decrease in the boot time using insserv reordering. In that occasion, I had already deleted some repeated init scripts from the insserv modified /etc/rc2.d directory. Recently, I had problems using insserv as I thought the major change was just to change the order of some of the init-scripts (stop-bootlogd, sysklogd, klogd). Now, from a system starting in 53 seconds I got a 2 second improvement by using the same script order used some time ago. Besides, I thought I shouldn't bother to delete the repeated init-scripts as /sbin/init is supposed to ignore them. Nevertheless, by removing the remaining repeated scripts I got a further 2 second time improvement, i.e., a 4 seconds improvement! The bootcharts and init order are available at the project webpage.

Currently recovering from a disk failure with lots of data lost (i'll backup more often now), I'm trying to tackle with the inconsistencies sketched in the previous blog: preload and parallel booting hotspots don't perform as before. I guess the discrepancies came from playing too much with the boot process.
The new system has originally a boot time of 52 seconds and the installation of preload 0.4 on a freshly installed system gave 1 second longer boot time during the first reboot and came back to the original time after 4 reboots and goes back to 53 seconds afterwards.
The bootchart is available on the bootcharts section of the project webpage for two preload configurations.

Besides, unverbose booting was tested. Well, simply verbose was changed to "VERBOSE=no" and "VERBOSE=quiet" just making the system slower! What could this be? Manual tweaking may be probably needed. The bootcharts for the different cases are available on the bootcharts section of the project webpage

Finally, a laptop testing system was prepared for comparison. It has also a fresh debian sid installation on an external USB hard drive. The results using the laptop test bed were similar for unverbose booting although the boot time seem to be unconsistent! Around 2 second changes without networking and higher with networking. I _guess_ a possible reason is by using an usb hard drive.

I've tried preload 0.2 and 0.4 to compare its effects. Unfortunately my results don't hold with the results obtained some blogs ago: preload 0.2 reduced the boot time by 2 seconds. In this occasion, time was even lost (making me remember my first tests). This may be due to some upgrades had been made to sid since.
First I tried preload 0.2 with default with 1 second lost and a modified configuration with no boot time change.
Then I tried preload 0.4 with default and modified configuration both with no time change.
The bootcharts and log file are availabe at the project webpage.
Yesterday I presented a table with hotspots combinates and unlike it was expected, preload may not be helping anymore (preload 0.4 was used).