Hi, Sun Managers:
_|(Solaris 2.3, SC1000E, 768 MB of RAM, patch 101318-70 and so on...)
_|Last Saturday, I was copying datas from one disk to another, using
_| # ufsdump 0f - /dev/rdsk/c1t3d0s4 | (cd /mount_point ; ufsrestore rf -)
_|Everything was nice :-) until "Broken pipe" and "The ENTIRE dump is
_|This is the trace:
_| Sat Jun 17 13:54:16 MET DST 1995
_| mount /dev/dsk/c6t0d2s7 /mnt_prov
_| ufsdump 0f - /dev/rdsk/c1t3d0s4 |(cd /mnt_prov; ufsrestore rf -)
_| Cannot open("/dev/tty"): No such device or address
_| DUMP: Date of this level 0 dump: Sat Jun 17 13:54:17 1995
_| DUMP: Date of last level 0 dump: the epoch
_| DUMP: Dumping /dev/rdsk/c1t3d0s4 (/dump/SERV_IMAGE/dmp2) to standard
_| DUMP: mapping (Pass I) [regular files]
_| DUMP: mapping (Pass II) [directories]
_| DUMP: estimated 606238 blocks (296.01MB)
_| DUMP: Writing 32 Kilobyte records
_| DUMP: dumping (Pass III) [directories]
_| DUMP: dumping (Pass IV) [regular files]
_| Warning: ./lost+found: File exists
_| DUMP: 80.30% done, finished in 0:01
_| DUMP: Broken pipe
_| DUMP: The ENTIRE dump is aborted.
_| umount /mnt_prov
_| Sat Jun 17 14:00:41 MET DST 1995
_|PS: For information, the message
_| "Cannot open("/dev/tty"): No such device or address"
_| appeared because it was a batch (non interactive)
_|I've tried again "manually" without success.
_|I've done a
_| # fcsk /dev/rdsk/c1t3d0s4
_|but all was good.
_|In fact, copy seemed to be good (same number of files and same size, good
_|Any ideas ?
Original answers :-)
_|No, sorry, but basically I have a "me too". Except for me it was
_|Solaris 1.1.1B, SPARC5, 112MB, recommended patches, etc., using
_|a dump | restore command. I tried it twice (interactively), then
_|tried it by dumping to a (compressed) disk file, and restoring
_|from the file, which worked perfectly. I noticed, like you, that
_|the dump seemed to have completed OK. Weird.
_|It happened to be an active /usr partition for me, but I can't
_| see how that would affect anything, since I have /usr set up to
_|be effectively read-only.
_|Hope this helps,
_|Ray Brownrigg <firstname.lastname@example.org> http://www.isor.vuw.ac.nz/~ray
_| You will always see this message when doing a disk to
_|disk dump like you did. The dump is fine. The issue is that
_|the restore is sitting on a pipe from another process, when
_|the dump is done, it closes it's end of the pipe, breaking the
_|pipe. Since there is not EOT on a pipe (what restore is
_|looking for), restore is still taking input from the defunct
_|dump. Since pipe breaks, restore bails. Your files are fine.
_|Chris Swanson [C=us; A= ; P=Internet; DDA.TYPE=RFC-822;
I realized that I have forgotten to say...
On that Saturday, 13 file-systems have been copied.
-> 11 with "DUMP IS DONE"
-> 2 with "The ENTIRE dump is aborted."
The conclusion is that ALL IS GOOD but by some random, dump and restore are
not exactly on phase. If the "restore" or "ufsrestore" command ends before
the "dump" or "ufsdump" command, then you'll see a "Broken Pipe". Don't
worry about that !
Thanks for Chris and Ray for their post.
Thanks for you for your reading !
[Rem: I really don't understand why there are so many "subscribe" and
especially "unsubscribe" on my mailbox every day !]
This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:10:27 CDT