diskless clients restart problems

ph0enix · 07-15-2003, 05:06 AM

Hi all,
I'm building linux cluster with diskless nodes having their root at nfs. After first boot node obtain universal image with HW detection routine. Depending to detected values the server prepare special image with required modules and proper kernel for this node and modify DNS and DHCP configuration. When this is done node should reboot and start again properly initialized with it's own configuration. My problem is that node does not reboot after detection

Even it looks like problem with init. After manual reset node start only HW detection and no services can't be started.

I'm using selfcompiled kernel 2.4.21 + mosix + i2c + lm_sensors with devfs enabled. The system is based on SuSE Linux 8.2.

The last lines I get from serial console after HW detection are

Code:

DEBUG: list of mounted filesystems...
rootfs on / type rootfs (rw)
/dev/root on / type nfs (rw,v2,rsize=4096,wsize=4096,hard,udp,nolock,addr=192.168.2.1)
none on /dev type devfs (rw)
proc on /proc type proc (rw)
devpts on /dev/pts type devpts (rw)
LCC: Rebooting...
INIT: Switching to runlevel: 6
INIT: Sending processes the TERM signal
Boot logging started on /dev/tts/0(/dev/console) at Tue Jul 15 00:13:29 2003
Master Resource Control: previous runlevel: 4, switching to runlevel: 6
INIT: no more processes left in this runlevel

which does fit to executed script

Code:

[...]
  if [ "$RESULT" == "REBOOT" ]; then
    echo "DEBUG: list of mounted filesystems..."
    mount
    echo "LCC: Rebooting..."
    shutdown -r now
    break
  else
[...]

But after it it looks like system was halted(?) instead of rebooting.

rch · 07-15-2003, 09:28 AM

Quote:

Originally posted by ph0enix
Hi all,
I'm building linux cluster with diskless nodes having their root at nfs. After first boot node obtain universal image with HW detection routine. Depending to detected values the server prepare special image with required modules and proper kernel for this node and modify DNS and DHCP configuration. When this is done node should reboot and start again properly initialized with it's own configuration. My problem is that node does not reboot after detection Even it looks like problem with init. After manual reset node start only HW detection and no services can't be started.

I'm using selfcompiled kernel 2.4.21 + mosix + i2c + lm_sensors with devfs enabled. The system is based on SuSE Linux 8.2.

The last lines I get from serial console after HW detection are

Code:

DEBUG: list of mounted filesystems... rootfs on / type rootfs (rw) /dev/root on / type nfs (rw,v2,rsize=4096,wsize=4096,hard,udp,nolock,addr=192.168.2.1) none on /dev type devfs (rw) proc on /proc type proc (rw) devpts on /dev/pts type devpts (rw) LCC: Rebooting... INIT: Switching to runlevel: 6 INIT: Sending processes the TERM signal Boot logging started on /dev/tts/0(/dev/console) at Tue Jul 15 00:13:29 2003 Master Resource Control: previous runlevel: 4, switching to runlevel: 6 INIT: no more processes left in this runlevel

which does fit to executed script

Code:

[...] if [ "$RESULT" == "REBOOT" ]; then echo "DEBUG: list of mounted filesystems..." mount echo "LCC: Rebooting..." shutdown -r now break else [...]

But after it it looks like system was halted(?) instead of rebooting.

Well I think that the problem is because of the runlevel that you are in.You are using systemV init, that means that your mounted root(from nfs) should contain the folder <nfs root>/etc/rc6.d and
<nfs root>/etc/init.d as would be specified by the <nfs root>/etc/rc.sysinit or any sysinit file that you use.So I think that you should check the sysinit file first,and check whether any service that would be required are in the relavent directory or not.