From: Curt Steinmetz (curt@cola.iges.org)
Date: Wed Apr 28 2004 - 12:31:17 EDT
Thanks to Chris Wincentsen, Tom Blinn, Kris Smith,
John Lanier, Michael Crowley and Eric Milligan.
The problem was that insufficient memory was available
for the restore process. This was resolved using the
sysconfigdb command to change the "proc" subsystem in the
file "/etc/sysconfigtab". It should be noted that there
was no "proc" subsystem in sysconfigtab when I started -
so it was using some kind of default values, I guess.
The procedure I followed was to (1) create a "stanza" file
with the new values for some kernel parameters (this
was a lot easier under 4.0F!). The "stanza" file I used
looks like this (do a man on "stanza" for more on this
abstruse topic):
proc:
per_proc_stack_size = 15719912800
max_per_proc_stack_size = 4294967296
per_proc_data_size = 4294967296
max_per_proc_data_size = 137438953472
(2) then I ran the command:
/sbin/sysconfigdb -rf <stanza filename> proc
followed by the command
/sbin/sysconfigdb -af <stanza filename> proc
(3) Note on the above: according to my reading of the
man page for sysconfigdb it is necessary to run the command
first with the -r option in order to avoid having to reboot the
system for the changes to take effect.
(4) After running the two command in step (2) above I ran
the retrieval procedure as originally described in the problem
report - and it worked fine!
Thanks again to all those who replied.
- Curt
On Apr 27, 2004, at 3:45 PM, Curt Steinmetz wrote:
> I have encountered an annoying problem trying to
> retrieve some files from a dump tape using restore.
> The tape was created with the following script:
>
> mt -f /dev/ntape/tape0c rewind
> /usr/sbin/dump -0uf /dev/ntape/tape0c /
> /usr/sbin/dump -0uf /dev/ntape/tape0c /usr
> /usr/sbin/dump -0uf /dev/ntape/tape0c /usr/local
> /usr/sbin/dump -0uf /dev/ntape/tape0c /homes/cola
> /usr/sbin/dump -0uf /dev/ntape/tape0c /var
> /usr/sbin/dump -0uf /dev/ntape/tape0c /var/spool/mail
> mt -f /dev/ntape/tape0c offline
>
> And now I am trying to restore a users files from
> the partition /homes/cola (the fourth partition in the
> dump). I use these three commands to start up an interactive
> restore from the command line:
> # mt -f /dev/ntape/tape0c rewind
> # mt -f /dev/ntape/tape0c fsf 3
> # /usr/sbin/restore if /dev/ntape/tape0c
>
> And I get this error message:
> Cannot malloc space for directory list, errno = 12
>
> I have done many restores using exactly this scenario but
> this is the first time I have tried since we upgraded from 4.0F
> to 5.1b, and also from DLT to SDLT. If I go to the partition prior
> to /homes/cola, ie, if I do "fsf 2" rather than "fsf 3" I can do a
> restore no problem (but I don't need to restore anything from
> /usr/local!!!!).
>
> I found a "solution" in the archives - to use vdump and vrestore,
> but that won't work for a tape that's already written using just
> plain old "dump" (I tried anyway).
>
> Any suggestions?
>
> Sincerely,
> Curt Steinmetz
> curt@cola.iges.org
> www.iges.org
>
>
This archive was generated by hypermail 2.1.7 : Sat Apr 12 2008 - 10:49:58 EDT