#12 severe crash while dumping
It's too strange:
I have 3 linux machines with same configuration and the
same big problem. There were upgraded from Redhat 5.2 to
7.1 for 1 year now. Since I installed RH7.1, severe
crashes appear repeatdly, most often (99,9% of cases)
while dump is running. There are severe because when
they appear, the only thing I can do is switch off the
power, restart, and hope there are no filesystem
corruptions. Crashes are not systematic: sometime I can
do many dumps successly.
I did the following without eliminating the problem,
either I think I know less crashes now than before:
- upgrade the kernel and patch it.
kernel is now 2.4.17 with
linux-aic7xxx-6.2.5-2.4.17.patch
(from http://people.freebsd.org/~gibbs/linux/\)
- ugrade do dump-0.4b29-1
- upgrade the scsi card BIOS (aic2940UW)
I use one HP dat devices on all machines, plus an HP DLT
on one machine. Processors are AMD-K6 266 MHz.Those
machines are servers only (X not installed), running
principally netatalk, samba, apache, ftpd...
other informations:
e2fsprogs is 1.19-4
glibc is 2.2.2-10
The only thing I can do now is to upgrade to RH7.2?
Any other better idea ??
Thanks for any help,
Bernard Paris
UCL - Universite de Louvain-la-Neuve
BELGIUM
bernard.paris@psp.ucl.ac.be
Logged In: YES
user_id=5513
When you say crashes, do you mean dump crashes or kernel
crashes. In the first case, please give details on what
fails and how. In the second case, this is not a dump
problem and you should contact the kernel developers in
order to find out what the problem is and how to correct it.
Stelian.
Logged In: NO
Well, when I say the system craches, I mean I have to switch
the power off and reboot the machine. It could be a problem
in the kernel, but it's strange since there is only "dump"
giving problems (and not all the time).
I've submitted a message in the linux.dev.kernel list.
Someone there tell me "in fact, linus have said dump is
broken and does not want to fix it.". What that means ??
Logged In: YES
user_id=5513
Ok, these kind of crashes indicates a problem in the kernel
(probably caused by a hardware malfunction).
Dump can trigger these problems because it is one of the few
programs which accesses the disk directly without going
through the kernel filesystem layer. It also causes massive
accesses to the SCSI tape drive, which can trigger a problem
in the SCSI hardware.
The person who told you that "dump is broken and Linus does
not want to fix it" has no clue about the problem. Just
ignore him, you have a real kernel and/or hardware problem.
I'm closing this support request since it is definately not
a dump issue.
Stelian.