1 |
dpavlin |
201 |
#!/usr/bin/perl |
2 |
|
|
|
3 |
|
|
use warnings; |
4 |
|
|
use strict; |
5 |
|
|
|
6 |
|
|
use IO::Socket::INET; |
7 |
|
|
use Storable qw(); |
8 |
|
|
use JSON; |
9 |
|
|
use Data::Dump qw(dump); |
10 |
dpavlin |
206 |
use Time::HiRes qw(time); |
11 |
dpavlin |
201 |
|
12 |
|
|
my $name = 'pxelator'; |
13 |
dpavlin |
206 |
my $shard_size = 5000; |
14 |
dpavlin |
201 |
|
15 |
|
|
|
16 |
|
|
my $path = '/tmp/sack'; |
17 |
|
|
mkdir $path; |
18 |
|
|
$path .= '/' . $name; |
19 |
|
|
mkdir $path; |
20 |
|
|
|
21 |
|
|
|
22 |
|
|
sub couchdb_socket { |
23 |
|
|
IO::Socket::INET->new( |
24 |
|
|
PeerAddr => '10.60.0.91', |
25 |
|
|
PeerPort => 5984, |
26 |
|
|
Proto => 'tcp', |
27 |
|
|
) || die $!; |
28 |
|
|
} |
29 |
|
|
|
30 |
|
|
sub get_chunk { |
31 |
|
|
my $sock = shift; |
32 |
|
|
my $chunk; |
33 |
|
|
while(<$sock>) { |
34 |
|
|
$chunk .= $_; |
35 |
|
|
last if /^[\n\r]+$/; |
36 |
|
|
} |
37 |
|
|
# warn "# $sock\n$chunk\n"; |
38 |
|
|
return $chunk; |
39 |
|
|
} |
40 |
|
|
|
41 |
|
|
my $sock = couchdb_socket; |
42 |
|
|
my $docs = couchdb_socket; |
43 |
|
|
|
44 |
|
|
print $sock "GET /$name/_all_docs HTTP/1.0\r\n\r\n"; |
45 |
|
|
|
46 |
|
|
get_chunk($sock); |
47 |
|
|
|
48 |
|
|
my $total = <$sock>; |
49 |
dpavlin |
206 |
$total =~ s{^.*total_rows\D+(\d+).+$}{$1}; |
50 |
dpavlin |
201 |
warn "# skip total: $total\n"; |
51 |
|
|
|
52 |
|
|
our $shard; |
53 |
|
|
our $shard_nr = 0; |
54 |
dpavlin |
206 |
my $t_start = time(); |
55 |
dpavlin |
201 |
|
56 |
|
|
sub save_shard { |
57 |
|
|
my $shard_path = sprintf("%s/%06d.%d", $path, $shard_nr++ * $shard_size, $shard_size); |
58 |
|
|
Storable::store( $shard, $shard_path ); |
59 |
dpavlin |
206 |
|
60 |
|
|
my $dt = time() - $t_start; |
61 |
|
|
my $pos = $shard_nr * $shard_size; |
62 |
|
|
my $rec_s = $pos / $dt; |
63 |
|
|
my $total_t = $total / $rec_s; |
64 |
|
|
|
65 |
|
|
warn sprintf "shard %s %d bytes\t%3.2f%% %.2f/s ETA %d:%02ds\n" |
66 |
|
|
, $shard_path, -s $shard_path |
67 |
|
|
, $pos * 100 / $total |
68 |
|
|
, $rec_s, |
69 |
|
|
, $total_t / 60, $total_t % 60 |
70 |
|
|
; |
71 |
|
|
|
72 |
dpavlin |
201 |
$shard = []; |
73 |
|
|
} |
74 |
|
|
|
75 |
|
|
while(<$sock>) { |
76 |
|
|
if ( /"id":"([^"]+)"/ ) { |
77 |
|
|
# warn "get $1\n"; |
78 |
|
|
print $docs "GET /$name/$1 HTTP/1.0\r\nConnection: Keep-Alive\r\n\r\n"; |
79 |
|
|
get_chunk($docs); |
80 |
|
|
my $json = from_json( <$docs> ); |
81 |
|
|
push @$shard, $json; |
82 |
|
|
save_shard if $#{ $shard } == $shard_size; |
83 |
|
|
} else { |
84 |
|
|
warn "UNKNOWN: $_"; |
85 |
|
|
} |
86 |
|
|
} |
87 |
|
|
save_shard; |