I've been backuping up by osf 4.0b system with /sbin/dump and it
/sbin/restore(s) fine to the 4.0b system.
I tried to /sbin/restore this on a 4.0f system, after about 10 minutes
the command errors off with:
Cannot calloc space for entry (addentry), errno=12
It creates 20,163 empty directories.
DEC Tech tells me to add more swap and if that didn't work use
vdump/vrestore.
. adding MORE swap didn't help
. dump/vrestore isn't really an option
tar takes WAYYYYYYYYYYYY too long.
Details:
The file system is about 8 gigabytes with about about a quarter
billion files (really).
We are backing up to a DAT tape -- but will try to a hard drive
later this week.
We have 2 gig of ram and 6 gig of swap.
to install vrestore/vdump we'd have to take a production machine
down and uninstall a patch, install the tools, install a
patchset, build a kernel and THEN start the transfer
we killed tar after 8 hours, normally the dump takes 3 hours.
breaking the /sbin/restore using the -i option and restoring
individual trees doesn't work 'cause there are simply too many
trees > 20,163 in any reasonable area.
Why is /sbin/restore doing this? How can we fix it?
How do we restore this /sbin/dump tape?
Is there a better way of doing this short of vdump/vrestore and tar?
TIA,
--
Jeffrey Hundstad
Minnesota State University, Mankato
Received on Tue Aug 24 1999 - 03:18:49 NZST