Hi,

I need to fix the following error in my perl program: 

Out of memory during "large" request for 16781312 bytes, total sbrk()
is
1247642432 bytes at /spfs/reports/bin//cobanalyzer.pl line 4948, <JSP>
line 5319224.

The line 4948 is:   $query{$key} = $val;

I only have time to change a few lines, so, is there any way I can
identify which variables / hashes consume the most memory so that I
can only optimize a coupld of places?

Thanks,
Matt

You probably want to tie your hash to an on-disk hash.  BerkeleyDB's
Perl interface is a good solution to a problem such as this.  Here is
some sample code to get you started:

<sample>

#!/usr/bin/perl

use warnings;
use strict;

use BerkeleyDB;

my %query;

my $db = tie %query, 'BerkeleyDB::Btree',
    -Flags    => BerkeleyDB::DB_CREATE,
    -Filename => 'query.db' or die "$!, $BerkeleyDB::Error";


$query{ "key" } = "val";

# and later ...

warn "key has a value of: ", $query{ "key" };

</sample>

If you really want to see the size of your Perl datastructures, you
could use Devel::Size (also available from the CPAN).

Hope that helps a bit.

-- Douglas

Douglas,

Thanks for the help. If any of the following 3 directions works, I'm
done.

1) I installed Devel::Size. But I did not find info about how to view
the hash size such as: $spc_info_{yp}{$mem}{$spc}{$spc_t}

"total_size(\%spc_info_)" seems suspicious.

2) BerkeleyDB is interesting. How can I install it without root
privilege?

3) I have Oracle 8 installed, how can I use Oracle instead of
BerkleyDB here?

Thanks,
Matt