From: Jeff Graves (
Date: Thu Nov 03 1994 - 09:10:20 CST


Original question:

> On one file system on one of our file servers, rdump has started giving
> the error message DUMP: SIGSEGV() ABORTING! on any dump level other
> than 0. Level 0 dumps are not having this problem, and none of the other
> file systems on this server are having this problem, and no other servers
> are having this problem. The server is a Sparc 10 running 4.1.3. Any ideas
> would be greatly appreciated.
> Example:
> lab1{jgraves}34: /usr/etc/rdump 5csf 23000 test.dump /files
> DUMP: Date of this level 5 dump: Tue Oct 11 14:05:45 1994
> DUMP: Date of last level 0 dump: Sun Oct 9 19:32:34 1994
> DUMP: Dumping /dev/rsd2c (/files) to test.dump
> DUMP: mapping (Pass I) [regular files]
> DUMP: mapping (Pass II) [directories]


Thanks to: (Perry Hutchison)
Nate Itkin <>


   Both Perry Hutchison and Nate Itkin suggested running fsck. On rebooting
the machine, the file system check did not find any problems, but a manual
fsck did (nothing that seemed unusual, though), and after that the dumps
worked fine.


> From: (Perry Hutchison)
> The two differences between a level 0 and a level-nonzero dump are that
> the nonzero must consult /etc/dumpdates to determine its "since when"
> (cutoff) date/time, and that the nonzero may have to run Pass II
> (directory mapping) more than once in order to pick up unchanged
> parents of changed directories and files.
> /etc/dumpdates seems unlikely to be the source of this problem, as the
> cutoff has already been determined and reported well before it dies,
> but it would not hurt to have a look at it anyway. Assuming this is
> OK, the remaining suspect is the filesystem itself: evidently Pass II
> is finding something confusing while scanning inodes and directories.
> I am guessing that descriptor 3 in the trace is reading the filesystem.
> Does this filesystem pass muster with fsck?


> From: Nate Itkin <>
> umount /files
> fsck /dev/rsd2c
> mount /files
> dump 5f - /files > /dev/null
> If this doesn't work, then something more exotic is in order.
> --
> - Nate Itkin
> - Portland Technology Development, Intel Corporation Aloha, Oregon
> - E-mail:

This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:09:13 CDT