One possible reason if this problem happens intermittently is the following: you may have a program such as ntpdate or rdate that starts in parallel with other init scripts such as KDM e.g., when some network interface is brought up, and steps the system clock by such an amount that the KDM timeout for X server startup (30 s by default) is exceeded.
In the case of ntpdate, one possible fix is to disable /etc/network/if-up.d/ntpdate and run ntpdate in the *foreground* from /etc/init.d/networking in the "start)" case after successful "ifup -a" call and the corresponding LSB logging message. Unfortunately, by default rsyslog is started later, in normal runlevels, therefore one cannot use the -s ntpdate option in this setup. Another little problem with this solution is that the clock may be stepped every time you do: invoke-rc.d networking start whereas it would be desirable to limit clock stepping to the (early) boot process. One might use a "flag" file for that, but I hope there are more elegant ways... -- To UNSUBSCRIBE, email to debian-qt-kde-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/87y5xagdkb....@frougon.dyndns.org