[Drbl-user] very slow boot of DRBL clients

[Rodolphe B. <rod...@ep...>]

Hi,

DRBL clients are extremely slow to boot, I'm not sure about the cause.
The boot sequence stalls at

"creating one-time-use dir /root /etc /var mount of var done"

This sequence is located in /opt/drbl/setup/files/misc/init.drbl

I've added some "echo" debug and it seems that the script stalls at the
untar of /var. I don't understand why since I can run the same untar
command on the server and it's done in less than 1 sec. The script can
mount the rest of the folders without problems. After ~20 minutes the
script goes on and the machine is ready to use.

It was not always the case. I've installed a default client, tuned it a
bit and after 3 or 4 reboot the stalls began. Before, the script was
executing these steps very quickly.

The modifications I've made are related to /etc/nsswitch.conf,
/etc/rc.local, /etc/fstab (to remove the mount of /home, I want to do it
manually).

I see no reason for the script to slow down to this point. I've checked
/etc/resolv.conf and it's ok. Could it be because of the many reboot
that leave the NFS server and the connection in a inconsistent state ?

Once booted, the logs show nothing special. Is there a way to debug the
boot sequence ?

Thanks,

OS is centos 5.5 64 bits.

drbl-chntpw-0.0.20040818-7
gpxe-1.0.0-1drbl
drbl-lzop-1.02-0.8drbl
udpcast-20100130-1drbl
freedos-1.0-15drbl
pigz-2.1.6-1drbl
drbl-partimage-0.6.8-1drbl
drbl-etherboot-5.4.3-2
drbl-1.9.6-35
drbl-ntfsprogs-2.0.0-4
partclone-0.2.11-1drbl
pbzip2-1.1.1-1drbl



Rodolphe Buret