All was looking well on a pair of nice new dell 1550 1ru dual cpu
servers. Heartbeat was fine, drbd compiled and appeared to load. Then I
added the last line to /etc/fstab, "/dev/nb0 /mail ext2 noauto 0 0"
After which the following kernel panic occurred, some 30 to 60 seconds
after the login prompt appeared. This is with only one machine in the
cluster running, and answering YES to drbd's load time question about not
waiting for the other cluster member.
Unable to handle kernel NULL pointer dereference at virtual address 0000008
printing eip:
e089488f
*pde = 0
Oops: 0000
CPU 0
EIP: 0010:[<e089488f>]
EFLAGS: 00010246
etc
Prosess kupdated (pid:8,stackpage=c18a3000)
stack ...
Calltrace ...
Code: 8b 40 08 89 f3 89 f9 89 45 f0 d3 eb 8b 45 08 f0 fe 48 1c 0f
The only other interesting thing is that at "make install" time, depmod
complained about an unresolved symbol.
"depmod: *** unresolved symbols in
/lib/modules/2.4.9/kernel/drivers/block/drbd.o"
The output of depmod -e was substantial and contains things like
fget,printk,kernel_thread. Which I find odd, not understanding the whole
kernel verson/kernel module mine field all that well. Errm not at all
actually.
The box is installed from RedHat 7.1, with kernel 2.4.9 compiled after. All
of hartbeat,drbd,and the apps were compiled and installed. Then I used
ghost to generate partition images, which I put on the second box. After a
few edits of hostnames, and IP addresses I had the cluster partner up.
Any comments would be most welcome this is beginning to exceed my Linux
skill set :-)
Alan
servers. Heartbeat was fine, drbd compiled and appeared to load. Then I
added the last line to /etc/fstab, "/dev/nb0 /mail ext2 noauto 0 0"
After which the following kernel panic occurred, some 30 to 60 seconds
after the login prompt appeared. This is with only one machine in the
cluster running, and answering YES to drbd's load time question about not
waiting for the other cluster member.
Unable to handle kernel NULL pointer dereference at virtual address 0000008
printing eip:
e089488f
*pde = 0
Oops: 0000
CPU 0
EIP: 0010:[<e089488f>]
EFLAGS: 00010246
etc
Prosess kupdated (pid:8,stackpage=c18a3000)
stack ...
Calltrace ...
Code: 8b 40 08 89 f3 89 f9 89 45 f0 d3 eb 8b 45 08 f0 fe 48 1c 0f
The only other interesting thing is that at "make install" time, depmod
complained about an unresolved symbol.
"depmod: *** unresolved symbols in
/lib/modules/2.4.9/kernel/drivers/block/drbd.o"
The output of depmod -e was substantial and contains things like
fget,printk,kernel_thread. Which I find odd, not understanding the whole
kernel verson/kernel module mine field all that well. Errm not at all
actually.
The box is installed from RedHat 7.1, with kernel 2.4.9 compiled after. All
of hartbeat,drbd,and the apps were compiled and installed. Then I used
ghost to generate partition images, which I put on the second box. After a
few edits of hostnames, and IP addresses I had the cluster partner up.
Any comments would be most welcome this is beginning to exceed my Linux
skill set :-)
Alan