Dear Sun-Managers:
You guys are just fantatic, coming to my rescue once again!
If I've missed mentioning all the Sun-Managers who helped me, please excuse
me.
- Don Williams
- Glenn Satchell
- Peter Bestel
- Mark S. Anderson
- Louis C. Liao
Original Question:
=================
I came across the following ufsdump errors during my disk to disk backup
on an x86 running Solaris 2.1:
taft# ufsdump 0f - /usr/openwin | (cd /new_openwin && ufsrestore xf -)
...
Directories already exist, set modes anyway? [yn] n
DUMP: error flushing command pipe: Error 0
DUMP: Write error on standard output
DUMP: Cannot recover
DUMP: The ENTIRE dump is aborted.
I did backup on 7 file systems. /usr/openwin and /opt were giving me the
above errors and the other file systems completed OK. I could start
OW with no problem and visually the old and the new file systems seemed
to be identical, except for the group ownership of any newly created
links and directories.
I did a search on ufsdump exit codes through the man pages:
Error
0 Normal Exit
1 Startup errors encountered
3 Abort - no checkpoint attempted
Do I have to worry about the abort dump message since the error exit code
is 0?
============================================================================
Responses:
=========
Don't worry, it's perfectly normal, your dump should be fine.
You usually get a pipe problem when doing disk-to-disk dumps.
This is either because the dump or the restore ends first,
closing its end of the pipe, upsetting the other process,
or it's because the restore is waiting for an EOF marker
on the "tape" which doesn't exist, or ...
(I've heard various explainations, but the message is normal)
Just out of curiousity, why don't you use "find" with "cpio" to do
disk-to-disk
copies? For example:
cd /usr/openwin; find . -print -mount | dd bs=32k | cpio -pdumv /new_openwin
should give you same result with less head-aches and problems. Also, you can
increase the blocking size of dd to optimize your transfer speed.
[I did use cpio at first. Somehow, the ownership was not copied correctly
so I used ufsdump instead.]
taft# cd /new_openwin
taft# ufsdump 0f - /usr/openwin | ufsrestore rf -
Try running fsck on the problem filesystems, then re-running ufsdump.
The message you're getting comes from ufsrestore. When it
fails, because you answer 'n' to set the modes anyway,
ufsdump gets the error code back.
All ufsrestore is saying is that it's about to set inode
permissions and ownership all files/directories restored.
Since you've had a few goes at this, or had already made a
copy of the openwindows distribution, you get a warning
that the directories already exist. Just answer 'y' to the
query on setting modes anyway.
============================================================================
Once again, thank you for putting my worries at rest.
Vicky Lau
vlau@msmail2.hac.com
This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:11:09 CDT