line |
stmt |
bran |
cond |
sub |
pod |
time |
code |
1
|
|
|
|
|
|
|
package PostgreSQL::PLPerl::NYTProf; |
2
|
|
|
|
|
|
|
BEGIN { |
3
|
1
|
|
|
1
|
|
734
|
$PostgreSQL::PLPerl::NYTProf::VERSION = '1.002'; |
4
|
|
|
|
|
|
|
} |
5
|
|
|
|
|
|
|
|
6
|
|
|
|
|
|
|
# vim: ts=8 sw=4 expandtab: |
7
|
|
|
|
|
|
|
|
8
|
|
|
|
|
|
|
=head1 NAME |
9
|
|
|
|
|
|
|
|
10
|
|
|
|
|
|
|
PostgreSQL::PLPerl::NYTProf - Profile PostgreSQL PL/Perl functions with Devel::NYTProf |
11
|
|
|
|
|
|
|
|
12
|
|
|
|
|
|
|
=head1 VERSION |
13
|
|
|
|
|
|
|
|
14
|
|
|
|
|
|
|
version 1.002 |
15
|
|
|
|
|
|
|
|
16
|
|
|
|
|
|
|
=head1 SYNOPSIS |
17
|
|
|
|
|
|
|
|
18
|
|
|
|
|
|
|
Load via the C environment variable: |
19
|
|
|
|
|
|
|
|
20
|
|
|
|
|
|
|
$ PERL5OPT='-MPostgreSQL::PLPerl::NYTProf' pg_ctl restart |
21
|
|
|
|
|
|
|
|
22
|
|
|
|
|
|
|
or load via your C file: |
23
|
|
|
|
|
|
|
|
24
|
|
|
|
|
|
|
custom_variable_classes = 'plperl' |
25
|
|
|
|
|
|
|
plperl.on_init = 'use PostgreSQL::PLPerl::NYTProf;' |
26
|
|
|
|
|
|
|
|
27
|
|
|
|
|
|
|
and restart the server. |
28
|
|
|
|
|
|
|
|
29
|
|
|
|
|
|
|
Then run some PL/Perl code: |
30
|
|
|
|
|
|
|
|
31
|
|
|
|
|
|
|
$ psql -c "do 'sub w { } w() for 1..100_000' language plperl" template1 |
32
|
|
|
|
|
|
|
|
33
|
|
|
|
|
|
|
which will create a nytprof.out.I file in the C<$PGDATA> directory, |
34
|
|
|
|
|
|
|
where I is the process id of the postgres backend. |
35
|
|
|
|
|
|
|
|
36
|
|
|
|
|
|
|
Finally, run C to generate a report, for example: |
37
|
|
|
|
|
|
|
|
38
|
|
|
|
|
|
|
$ nytprofhtml --file $PGDATA/nytprof.out.54321 --open |
39
|
|
|
|
|
|
|
|
40
|
|
|
|
|
|
|
=head1 DESCRIPTION |
41
|
|
|
|
|
|
|
|
42
|
|
|
|
|
|
|
Profile PL/Perl functions inside PostgreSQL database with L. |
43
|
|
|
|
|
|
|
|
44
|
|
|
|
|
|
|
PostgreSQL 9.0 or later is required. |
45
|
|
|
|
|
|
|
|
46
|
|
|
|
|
|
|
=head1 ENABLING |
47
|
|
|
|
|
|
|
|
48
|
|
|
|
|
|
|
In order to use this module you need to arrange for it to be loaded when |
49
|
|
|
|
|
|
|
PostgreSQL initializes a Perl interpreter. |
50
|
|
|
|
|
|
|
|
51
|
|
|
|
|
|
|
=head2 Quick Occasional Use |
52
|
|
|
|
|
|
|
|
53
|
|
|
|
|
|
|
The C environment variable can be used like this: |
54
|
|
|
|
|
|
|
|
55
|
|
|
|
|
|
|
$ PERL5OPT='-MPostgreSQL::PLPerl::NYTProf' pg_ctl restart |
56
|
|
|
|
|
|
|
|
57
|
|
|
|
|
|
|
This will be effective for any pg_ctl command that restarts the postmaster |
58
|
|
|
|
|
|
|
process, so C will work but C won't. |
59
|
|
|
|
|
|
|
|
60
|
|
|
|
|
|
|
The profiler will remain enabled until the the postmaster process is restarted. |
61
|
|
|
|
|
|
|
|
62
|
|
|
|
|
|
|
=head2 Via postgres.conf |
63
|
|
|
|
|
|
|
|
64
|
|
|
|
|
|
|
You can simply add a C |
65
|
|
|
|
|
|
|
|
66
|
|
|
|
|
|
|
plperl.on_init='use PostgreSQL::PLPerl::NYTProf;' |
67
|
|
|
|
|
|
|
|
68
|
|
|
|
|
|
|
though I'd recommend arranging for PostgreSQL to load a separate |
69
|
|
|
|
|
|
|
F file from same directory as your F file: |
70
|
|
|
|
|
|
|
|
71
|
|
|
|
|
|
|
plperl.on_init='require "plperloninit.pl";' |
72
|
|
|
|
|
|
|
|
73
|
|
|
|
|
|
|
then you can put whatever Perl statements you want in that file: |
74
|
|
|
|
|
|
|
|
75
|
|
|
|
|
|
|
use PostgreSQL::PLPerl::NYTProf; |
76
|
|
|
|
|
|
|
|
77
|
|
|
|
|
|
|
When it's no longer needed just comment it out by prefixing with a C<#>. |
78
|
|
|
|
|
|
|
|
79
|
|
|
|
|
|
|
=head1 USAGE |
80
|
|
|
|
|
|
|
|
81
|
|
|
|
|
|
|
By default the NYTProf profile data files will be written into the database |
82
|
|
|
|
|
|
|
directory, alongside your F, with the process id of the backend |
83
|
|
|
|
|
|
|
appended to the name. For example F. |
84
|
|
|
|
|
|
|
|
85
|
|
|
|
|
|
|
You'll get one profile data file for each database connection. You can use the |
86
|
|
|
|
|
|
|
L utility to merge multiple data files if needed. |
87
|
|
|
|
|
|
|
|
88
|
|
|
|
|
|
|
To generate a report from a data file, use a command like: |
89
|
|
|
|
|
|
|
|
90
|
|
|
|
|
|
|
nytprofhtml --file=$PGDATA/nytprof.out.54321 --open |
91
|
|
|
|
|
|
|
|
92
|
|
|
|
|
|
|
=head1 INTERPRETING REPORTS |
93
|
|
|
|
|
|
|
|
94
|
|
|
|
|
|
|
PL/Perl functions are given names in perl that include the OID of the PL/Perl |
95
|
|
|
|
|
|
|
function. So a function created by C would appear |
96
|
|
|
|
|
|
|
in the reports as something like C. |
97
|
|
|
|
|
|
|
|
98
|
|
|
|
|
|
|
=head1 PROFILE ON DEMAND |
99
|
|
|
|
|
|
|
|
100
|
|
|
|
|
|
|
The instructions above enable profiling for all database sessions that use PL/Perl. |
101
|
|
|
|
|
|
|
Instead of profiling all sessions it can be useful to have the profiler loaded |
102
|
|
|
|
|
|
|
into the server but only enable it for particular sessions. |
103
|
|
|
|
|
|
|
|
104
|
|
|
|
|
|
|
You can do this by loading setting the C environment variable to |
105
|
|
|
|
|
|
|
include the "C" option. Then, to enable profiling for a particular |
106
|
|
|
|
|
|
|
session you just need to call the C function. For example: |
107
|
|
|
|
|
|
|
|
108
|
|
|
|
|
|
|
do 'DB::enable_profile' language plperl; |
109
|
|
|
|
|
|
|
|
110
|
|
|
|
|
|
|
See L. |
111
|
|
|
|
|
|
|
|
112
|
|
|
|
|
|
|
The performance impact of loading but not enabling NYTProf should be I |
113
|
|
|
|
|
|
|
low (though I've not tried measuring it). So, while I wouldn't recommend doing |
114
|
|
|
|
|
|
|
that on a production instance, it would be fine on a development instance. |
115
|
|
|
|
|
|
|
|
116
|
|
|
|
|
|
|
=head1 LIMITATIONS |
117
|
|
|
|
|
|
|
|
118
|
|
|
|
|
|
|
=head2 Can't use plperl and plperlu at the same time |
119
|
|
|
|
|
|
|
|
120
|
|
|
|
|
|
|
Postgres uses separate Perl interpreters for the plperl and plperlu languages. |
121
|
|
|
|
|
|
|
NYTProf is not multiplicity safe (as of version 4.05). It should just profile |
122
|
|
|
|
|
|
|
whichever language was used first and ignore the other, but there may still be |
123
|
|
|
|
|
|
|
problems in this situation. Let me know if you encounter any odd behaviour. |
124
|
|
|
|
|
|
|
|
125
|
|
|
|
|
|
|
=head2 PL/Perl functions with unusual names are __ANON__ |
126
|
|
|
|
|
|
|
|
127
|
|
|
|
|
|
|
PL/Perl functions are created as anonymous subroutines in Perl. |
128
|
|
|
|
|
|
|
PostgreSQL::PLPerl::NYTProf arranges for them to be given names. |
129
|
|
|
|
|
|
|
The logic currently only works for names that match C^\w+$/>. |
130
|
|
|
|
|
|
|
|
131
|
|
|
|
|
|
|
=head1 SEE ALSO |
132
|
|
|
|
|
|
|
|
133
|
|
|
|
|
|
|
L |
134
|
|
|
|
|
|
|
|
135
|
|
|
|
|
|
|
=head1 AUTHOR |
136
|
|
|
|
|
|
|
|
137
|
|
|
|
|
|
|
B, L and L |
138
|
|
|
|
|
|
|
|
139
|
|
|
|
|
|
|
=head1 COPYRIGHT AND LICENSE |
140
|
|
|
|
|
|
|
|
141
|
|
|
|
|
|
|
Copyright (C) 2009-2010 by Tim Bunce. |
142
|
|
|
|
|
|
|
|
143
|
|
|
|
|
|
|
This library is free software; you can redistribute it and/or modify |
144
|
|
|
|
|
|
|
it under the same terms as Perl itself, either Perl version 5.8.8 or, |
145
|
|
|
|
|
|
|
at your option, any later version of Perl 5 you may have available. |
146
|
|
|
|
|
|
|
|
147
|
|
|
|
|
|
|
=cut |
148
|
|
|
|
|
|
|
|
149
|
1
|
|
|
1
|
|
8
|
use strict; |
|
1
|
|
|
|
|
1
|
|
|
1
|
|
|
|
|
28
|
|
150
|
|
|
|
|
|
|
|
151
|
1
|
|
|
1
|
|
800
|
use Devel::NYTProf::Core; |
|
1
|
|
|
|
|
1604
|
|
|
1
|
|
|
|
|
431
|
|
152
|
|
|
|
|
|
|
|
153
|
|
|
|
|
|
|
# set some default options (can be overridden via NYTPROF env var) |
154
|
|
|
|
|
|
|
DB::set_option("endatexit", 1); # for pg 8.4 |
155
|
|
|
|
|
|
|
DB::set_option("savesrc", 1); |
156
|
|
|
|
|
|
|
DB::set_option("addpid", 1); |
157
|
|
|
|
|
|
|
# file defaults to nytprof.out.$pid in $PGDATA directory |
158
|
|
|
|
|
|
|
|
159
|
|
|
|
|
|
|
my $trace = $ENV{PLPERL_NYTPROF_TRACE} || 0; |
160
|
|
|
|
|
|
|
my @on_init; |
161
|
|
|
|
|
|
|
my $mkfuncsrc = "PostgreSQL::InServer::mkfuncsrc"; |
162
|
|
|
|
|
|
|
|
163
|
|
|
|
|
|
|
if (not -f 'postgres.conf') { |
164
|
|
|
|
|
|
|
# there's no easy way to tell that we're being loaded into the |
165
|
|
|
|
|
|
|
# postgres server when we're loaded via PERL5OPT. This'll do: |
166
|
|
|
|
|
|
|
warn __PACKAGE__." not running in postgres server"; |
167
|
|
|
|
|
|
|
} |
168
|
|
|
|
|
|
|
elsif (defined &{$mkfuncsrc}) { |
169
|
|
|
|
|
|
|
# We were probably loaded via plperloninit.pl |
170
|
|
|
|
|
|
|
fix_mkfuncsrc(); |
171
|
|
|
|
|
|
|
} |
172
|
|
|
|
|
|
|
else { |
173
|
|
|
|
|
|
|
# We were probably loaded via PERL5OPT='-M...' and so we're executing very |
174
|
|
|
|
|
|
|
# early, before mkfuncsrc has even been defined. |
175
|
|
|
|
|
|
|
# So we need to defer wrapping it until later. |
176
|
|
|
|
|
|
|
# We do that by wrapping PostgreSQL::InServer::Util::bootstrap |
177
|
|
|
|
|
|
|
# But that doesn't exist yet either. Happily it will do a INIT time |
178
|
|
|
|
|
|
|
# so we arrange to wrap it then. Got that? |
179
|
|
|
|
|
|
|
push @on_init, sub { |
180
|
|
|
|
|
|
|
hook_after_sub("PostgreSQL::InServer::Util::bootstrap", \&fix_mkfuncsrc); |
181
|
|
|
|
|
|
|
}; |
182
|
|
|
|
|
|
|
} |
183
|
|
|
|
|
|
|
|
184
|
|
|
|
|
|
|
|
185
|
|
|
|
|
|
|
sub fix_mkfuncsrc { |
186
|
|
|
|
|
|
|
|
187
|
|
|
|
|
|
|
# wrap mkfuncsrc with code that edits the returned code string |
188
|
|
|
|
|
|
|
# such that the code will give a name to the subroutine it defines. |
189
|
|
|
|
|
|
|
|
190
|
|
|
|
|
|
|
hook_after_sub("PostgreSQL::InServer::mkfuncsrc", sub { |
191
|
0
|
|
|
0
|
|
|
my ($argref, $code) = @_; |
192
|
0
|
|
|
|
|
|
my ($name, $imports, $prolog, $src) = @$argref; |
193
|
|
|
|
|
|
|
|
194
|
|
|
|
|
|
|
# $code = qq[ package main; sub { $BEGIN $prolog $src } ]; |
195
|
|
|
|
|
|
|
# XXX escape $name or extract from $code and use single quotes |
196
|
0
|
0
|
0
|
|
|
|
$code =~ s/\b sub \s* {(.*)} \s* $/sub $name { $1 }; \\&$name/sx |
197
|
|
|
|
|
|
|
or warn "Failed to edit sub name in $code" |
198
|
|
|
|
|
|
|
if $name =~ /^\w+$/; # XXX just sane names for now |
199
|
|
|
|
|
|
|
|
200
|
0
|
|
|
|
|
|
return $code; |
201
|
0
|
|
|
0
|
0
|
|
}); |
202
|
|
|
|
|
|
|
} |
203
|
|
|
|
|
|
|
|
204
|
|
|
|
|
|
|
|
205
|
|
|
|
|
|
|
sub hook_after_sub { |
206
|
0
|
|
|
0
|
0
|
|
my ($sub, $code, $force) = @_; |
207
|
|
|
|
|
|
|
|
208
|
0
|
0
|
|
|
|
|
warn "Wrapping $sub\n" if $trace; |
209
|
0
|
|
0
|
|
|
|
my $orig_sub = (defined &{$sub}) && \&{$sub}; |
210
|
0
|
0
|
0
|
|
|
|
if (not $orig_sub and not $force) { |
211
|
0
|
|
|
|
|
|
warn "hook_after_sub: $sub isn't defined\n"; |
212
|
0
|
|
|
|
|
|
return; |
213
|
|
|
|
|
|
|
} |
214
|
|
|
|
|
|
|
|
215
|
|
|
|
|
|
|
my $wrapped = sub { |
216
|
0
|
0
|
|
0
|
|
|
warn "Wrapped $sub(@_) called\n" if $trace; |
217
|
0
|
|
|
|
|
|
my @ret; |
218
|
0
|
0
|
|
|
|
|
if ($orig_sub) { |
219
|
|
|
|
|
|
|
# XXX doesn't handle context |
220
|
|
|
|
|
|
|
# XXX the 'package main;' here is a hack to make |
221
|
|
|
|
|
|
|
# PostgreSQL::InServer::Util::bootstrap do the right thing |
222
|
0
|
|
|
|
|
|
@ret = do { package main; |
223
|
|
|
|
|
|
|
BEGIN { |
224
|
1
|
|
|
1
|
|
68
|
$main::VERSION = '1.002'; |
225
|
0
|
|
|
|
|
|
} $orig_sub->(@_) }; |
226
|
|
|
|
|
|
|
} |
227
|
0
|
|
|
|
|
|
return $code->( [ @_ ], @ret ); |
228
|
0
|
|
|
|
|
|
}; |
229
|
|
|
|
|
|
|
|
230
|
1
|
|
|
1
|
|
7
|
no warnings 'redefine'; |
|
1
|
|
|
|
|
1
|
|
|
1
|
|
|
|
|
41
|
|
231
|
1
|
|
|
1
|
|
4
|
no strict; |
|
1
|
|
|
|
|
2
|
|
|
1
|
|
|
|
|
102
|
|
232
|
0
|
|
|
|
|
|
*{$sub} = $wrapped; |
|
0
|
|
|
|
|
|
|
233
|
|
|
|
|
|
|
} |
234
|
|
|
|
|
|
|
|
235
|
|
|
|
|
|
|
|
236
|
|
|
|
|
|
|
# --- final initialization --- |
237
|
|
|
|
|
|
|
|
238
|
|
|
|
|
|
|
# give the 'application' a more user-friendly name |
239
|
|
|
|
|
|
|
$0 = "PostgreSQL Session" if $0 eq '-e'; |
240
|
|
|
|
|
|
|
|
241
|
|
|
|
|
|
|
eval q{ INIT { $_->() for @on_init }; 1 } or die |
242
|
|
|
|
|
|
|
if @on_init; |
243
|
|
|
|
|
|
|
|
244
|
|
|
|
|
|
|
require Devel::NYTProf; # init profiler - DO THIS LAST |
245
|
|
|
|
|
|
|
|
246
|
|
|
|
|
|
|
__END__ |