On 11/24/2012 07:53 AM, Adam Williamson wrote:
On Sat, 2012-11-24 at 05:20 +0100, Ralf Corsepius wrote:
d) hard and difficult to reproduce
you're _really_ going to have to include logs.
No luck, yet. So far, I have always found myself with a shredded
partition, but haven't found indication for what could have caused the
shredding in any log, yet.
Some details to report (No idea whether they are related to my sporadic
issue):
[Preliminary remark: The exteral HD has several different Linuxes
distros installed in parallel. It is the drive being booted from.]
1. The external disk seems to have issues with it's id (Here Fedora 16):
# ls -l /dev/disk/by-id/ | grep sdb11
lrwxrwxrwx. 1 root root 11 Nov 27 14:29 usb-TOSHIBA_MK8025GAS_†††††娠㈵
㝇㠳匹-0:0-part11 -> ../../sdb11
2. dmesg reports this (Here: Fedora 16):
[ 26.904380] ata_id[762]: HDIO_GET_IDENTITY failed for '/dev/sdb':
Invalid argument
3. Fedora 16 and 18 seem to order the drives this way:
sda -> internal drive
sdb -> external drive
For reasons unknown to me, CentOS6 has this order reversed:
sdb -> internal drive
sda -> external drive
4. When additionally having plugged-in a USB-stick while booting,
afterwards, different distros see the HD under different device names:
Fedora 18:
# df | grep /dev/sd
/dev/sdc6 12385456 6773000 4983312 58% /
/dev/sdc5 495844 42823 427421 10% /boot
Fedora 16:
# df | grep sd
/dev/sdb12 12385456 3789344 7966968 33% /
/dev/sdb11 495844 34269 435975 8% /boot
CentOS 6
# df | grep sd
/dev/sda9 12385456 3418456 8337856 30% /
/dev/sda8 495844 31598 438646 7% /boot
[The disk and the machine are the same in all cases.
Of cause, the partitions are different for each distro.]
4. Is something in Fedora using partions "by-label"?
There is a possiblity, the shredded partion's label was identical to a
partition label on the external HD.
(Since the incident, both disks have been partially repartioned and
reformated.)
Together, storage.log, program.log and syslog ought to record all
relevant actions. storage.log and program.log should record formatting
and bootloader writing actions, and syslog should ID the disks.
Thanks for the pointer, but so far nothing has caught my eye.
There is
no possible way we can figure out what happened without the install
logs.
Correct, that's why I haven't BZ'ed it, yet and am telling you about it,
here.
This doesn't invalidate the fact, I am experiencing issues of this kind
with this particular machine and this particular external HD every now
and then (non deterministic, sporatic). However, like I said before, I
have not been able to identify the cause and am only "wild guessing".
is it possible the BIOS is changing the drive order for some reason and
you're not noticing?
I would not want to exclude this possiblity, but haven't yet found any
way to provoke such behavior.
Ralf
--
test mailing list
test@xxxxxxxxxxxxxxxxxxxxxxx
To unsubscribe:
https://admin.fedoraproject.org/mailman/listinfo/test
