Hello Ziga, sorry for the late reply.
Ziga Mahkovec schrieb am Wed 18. Apr, 13:23 (+0200): > On Wed, 2007-04-18 at 11:23 +0200, Jörg Sommer wrote: > > Ziga Mahkovec schrieb am Sun 15. Apr, 02:30 (+0200): > > > How about we avoid the udev problem (missing /dev/null) by starting the > > > loggers before udev is started? > > > > Yes, this might be another way to go. But didn't we get errors because > > the not yet mounted /proc? I've checked the script that mounts /proc in > > Debian. We can mount /proc in bootchartd without breaking this mount > > script and I guess other distributions do also check for an already > > mounted filesystem before mounting it. > > I don't expect this to be problem. I've tested bootchartd with an > unmounted /proc and the log_output functions would simply not log > anything. As soon as /proc becomes available, the functions start > logging. But then you loose the informations until /proc is available. On Debian these are at least two init scripts. > Also, as far as Fedora is concerned, /proc is already mounted in initrd > phase (before bootchart is started). Is this not the case in Debian? Yes, if you _use_ an initrd. What about mounting the proc filesystem in bootchartd? I've checked Debian's init scripts and they are aware of a mounted procfs—at least, because this _could_ happen on an initrd. So, I would suggest to check for a mounted procfs with “mountpoint /proc” and mount it accordingly. And do not quit the master process until all childs are spawned. Bye, Jörg. -- Damit das Mögliche entsteht, muß immer wieder das Unmögliche versucht werden. (Hermann Hesse)
pgpsJ4sHNyWrN.pgp
Description: PGP signature
