Piotr Wilkin <pwilkin@astercity.net>:
>Yesterday I was working on my system doing some standard work (running
>netscape, downloading mail via fetchmail etc.), when suddenly I got a
>syslogd message on the console from smail stating that there was no space
>left on device. I was pretty sure that just yesterday there was 600MB
>avail. besides my root-space (5% of 6GB = 300MB), but now df showed that
>there was no space available. I deleted some files, fetched the mail, but
>then just in a while got another message trying to save a file in
>WordPerfect. I checked the used space using du - there was an about 1GB
>gap between what du showed and what df did. I then found the perpetrator -
>it was an .xsession-errors file in my /root directory that had risen to
>1.1GB in size. I deleted the file, but lsof showed the file still opened,
>although with status (deleted). Without thinking much, I rebooted the
>machine (1.1GB was released almost immediately without e2fsck), then
>remounted r/o and ran fsck, which released a further 400MB via "free
>groups count wrong for #...". I also got the following system messages at
>boot:
>Jun 6 23:43:48 debian kernel: EXT2-fs error (device ide1(22,1)):
>ext2_check_blocks_bitmap: Wrong free blocks count for group 18, stored =
>2087, counted = 1950
>Jun 6 23:43:48 debian kernel: EXT2-fs error (device ide1(22,1)):
>ext2_check_blocks_bitmap: Wrong free blocks count in super block, stored =
>392229, counted = 392093
>The moment in which this happened was probably displayed by this fragment
>in the logs:
>Jun 6 22:22:06 debian -- MARK --
>Jun 6 22:42:00 debian kernel: Process accounting paused
>Jun 6 22:44:58 debian mcedit: /dev/tty3: Permission denied
>
>My question is: what could have caused this behavior?
I ran across something similar using "reboot" to boot the system -
It looked like (since I had just deleted some large tar files) the
system shutdown did not fully wait for the disk buffers to flush before
halting/rebooting. A delete of a large file may take a while to deallocate
the blocks.
This may have bit you since yours was still open by X - the large error
file. This file wasn't closed until X had exited (that is when the delete
actually started being processed), which by my reckoning would have occured
about 5 seconds before the BIOS reboot started.
I won't swear this is what got you, but I think there is a timeout in
the halt/reboot that ends up having the reboot start the BIOS before the disk
flush completes.
I usually get logged out before doing a ctrl-alt-del or reboot command from
a bare console VT.
-------------------------------------------------------------------------
Jesse I Pollard, II
Email: pollard@navo.hpc.mil
Any opinions expressed are solely my own.
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/
This archive was generated by hypermail 2b29 : Thu Jun 15 2000 - 21:00:14 EST