Bug#405098: Many errors and eventual failure when "Copy data from partition" used on partition with dir_index
Package: debian-installer
Severity: important
i386 using RC1 netinst image
I am copying my Debian installation onto a new hard drive with LVM and
encrypted swap and /tmp. I tried to use the installer's option to
"Copy data from partition," but after I selected the source partition,
it complained about parted not being able to resize a partition that
had dir_index enabled, saying that parted can only resize it if it
disables dir_index first. I don't know why it would need to resize
any of the partitions just to copy data from one partition to another,
and I'm not sure if it was referring to the source partition, or the
new, blank partition.
The options in the dialog were "Ignore," "Cancel," and "Go Back." The
first time I selected Ignore, and it started copying the data. At the
end of the copy process, it started giving a series of error dialogs
saying, "Block 2490887 shouldn't have been marked (1, 0)!" Choosing
"Continue" caused it to pop up several more times, but the first
number in the parentheses would double each time, and the block number
would increase as well, though only by a few.
After three or four of those, it would pop up a larger dialog saying,
"Assertion (block < EXT2_SUPER_BLOCKS_COUNT)fs->sb)) at
../../../../libparted/fs/ext2/ext2.h:226 in function
ext2_is_data_block() failed. \n\n A bug has been discovered! \n\n
Ignore \n Cancel" (added "\n" myself to indicate newlines in error
dialog) Choosing Ignore caused more of the "block shouldn't have been
marked" errors, and after a few more of those, I'd get another one of
this same larger error.
IIRC, after getting into this situation, choosing Cancel or Go Back
did the same as choosing Ignore, and the only way to get out of it was
to reboot. But if I chose Cancel or Go Back when the first warning
about resizing with dir_index appeared, all I got was a blank debconf
screen, blue with a gray bay at the bottom, and no text. I rebooted
then too.
I've described this to the best of my recollection. Apologies for any
errors. I will follow up with a few photos I took of some errors
after I get my system running on the new disk.
In the end, I used the rescue mode to use the Guided Partitioning for
LVM and crypto setup, but then switched to a console and am copying
over the data manually after making symlinks to the necessary libs so
I could use GNU cp rather than busybox's.
P.S. Is there no room for the GNU utils on the netinst iso? Busybox
is nice but, for example, has no --help for cp, and no -v for cp
either. I don't know about you, but when I'm copying tens of gigs of
data, I'd like -v so I can see that it's working. :)
Reply to: