1 |
root |
1.1 |
=head1 NAME |
2 |
|
|
|
3 |
root |
1.2 |
BDB - Asynchronous Berkeley DB access |
4 |
root |
1.1 |
|
5 |
|
|
=head1 SYNOPSIS |
6 |
|
|
|
7 |
root |
1.2 |
use BDB; |
8 |
root |
1.1 |
|
9 |
root |
1.19 |
my $env = db_env_create; |
10 |
|
|
|
11 |
|
|
mkdir "bdtest", 0700; |
12 |
|
|
db_env_open |
13 |
|
|
$env, |
14 |
|
|
"bdtest", |
15 |
|
|
BDB::INIT_LOCK | BDB::INIT_LOG | BDB::INIT_MPOOL |
16 |
|
|
| BDB::INIT_TXN | BDB::RECOVER | BDB::USE_ENVIRON | BDB::CREATE, |
17 |
|
|
0600; |
18 |
|
|
|
19 |
|
|
$env->set_flags (BDB::AUTO_COMMIT | BDB::TXN_NOSYNC, 1); |
20 |
|
|
|
21 |
|
|
my $db = db_create $env; |
22 |
|
|
db_open $db, undef, "table", undef, BDB::BTREE, BDB::AUTO_COMMIT | BDB::CREATE |
23 |
|
|
| BDB::READ_UNCOMMITTED, 0600; |
24 |
|
|
db_put $db, undef, "key", "data", 0, sub { |
25 |
|
|
db_del $db, undef, "key"; |
26 |
|
|
}; |
27 |
|
|
db_sync $db; |
28 |
|
|
|
29 |
|
|
# automatic result processing with AnyEvent: |
30 |
|
|
our $FH; open $FH, "<&=" . BDB::poll_fileno; |
31 |
|
|
our $WATCHER = AnyEvent->io (fh => $FH, poll => 'r', cb => \&BDB::poll_cb); |
32 |
|
|
|
33 |
|
|
# automatic result processing with EV: |
34 |
|
|
my $WATCHER = EV::io BDB::poll_fileno, EV::READ, \&BDB::poll_cb; |
35 |
|
|
|
36 |
|
|
# with Glib: |
37 |
|
|
add_watch Glib::IO BDB::poll_fileno, |
38 |
|
|
in => sub { BDB::poll_cb; 1 }; |
39 |
|
|
|
40 |
|
|
# or simply flush manually |
41 |
|
|
BDB::flush; |
42 |
|
|
|
43 |
|
|
|
44 |
root |
1.1 |
=head1 DESCRIPTION |
45 |
|
|
|
46 |
root |
1.10 |
See the BerkeleyDB documentation (L<http://www.oracle.com/technology/documentation/berkeley-db/db/index.html>). |
47 |
root |
1.12 |
The BDB API is very similar to the C API (the translation has been very faithful). |
48 |
root |
1.10 |
|
49 |
|
|
See also the example sections in the document below and possibly the eg/ |
50 |
|
|
subdirectory of the BDB distribution. Last not least see the IO::AIO |
51 |
|
|
documentation, as that module uses almost the same asynchronous request |
52 |
|
|
model as this module. |
53 |
|
|
|
54 |
|
|
I know this is woefully inadequate documentation. Send a patch! |
55 |
root |
1.7 |
|
56 |
root |
1.1 |
|
57 |
|
|
=head1 REQUEST ANATOMY AND LIFETIME |
58 |
|
|
|
59 |
|
|
Every request method creates a request. which is a C data structure not |
60 |
|
|
directly visible to Perl. |
61 |
|
|
|
62 |
|
|
During their existance, bdb requests travel through the following states, |
63 |
|
|
in order: |
64 |
|
|
|
65 |
|
|
=over 4 |
66 |
|
|
|
67 |
|
|
=item ready |
68 |
|
|
|
69 |
|
|
Immediately after a request is created it is put into the ready state, |
70 |
|
|
waiting for a thread to execute it. |
71 |
|
|
|
72 |
|
|
=item execute |
73 |
|
|
|
74 |
|
|
A thread has accepted the request for processing and is currently |
75 |
|
|
executing it (e.g. blocking in read). |
76 |
|
|
|
77 |
|
|
=item pending |
78 |
|
|
|
79 |
|
|
The request has been executed and is waiting for result processing. |
80 |
|
|
|
81 |
|
|
While request submission and execution is fully asynchronous, result |
82 |
|
|
processing is not and relies on the perl interpreter calling C<poll_cb> |
83 |
|
|
(or another function with the same effect). |
84 |
|
|
|
85 |
|
|
=item result |
86 |
|
|
|
87 |
|
|
The request results are processed synchronously by C<poll_cb>. |
88 |
|
|
|
89 |
|
|
The C<poll_cb> function will process all outstanding aio requests by |
90 |
|
|
calling their callbacks, freeing memory associated with them and managing |
91 |
|
|
any groups they are contained in. |
92 |
|
|
|
93 |
|
|
=item done |
94 |
|
|
|
95 |
|
|
Request has reached the end of its lifetime and holds no resources anymore |
96 |
|
|
(except possibly for the Perl object, but its connection to the actual |
97 |
|
|
aio request is severed and calling its methods will either do nothing or |
98 |
|
|
result in a runtime error). |
99 |
|
|
|
100 |
|
|
=back |
101 |
|
|
|
102 |
|
|
=cut |
103 |
|
|
|
104 |
root |
1.2 |
package BDB; |
105 |
root |
1.1 |
|
106 |
|
|
no warnings; |
107 |
|
|
use strict 'vars'; |
108 |
|
|
|
109 |
|
|
use base 'Exporter'; |
110 |
|
|
|
111 |
|
|
BEGIN { |
112 |
root |
1.27 |
our $VERSION = '1.4'; |
113 |
root |
1.1 |
|
114 |
root |
1.3 |
our @BDB_REQ = qw( |
115 |
root |
1.6 |
db_env_open db_env_close db_env_txn_checkpoint db_env_lock_detect |
116 |
|
|
db_env_memp_sync db_env_memp_trickle |
117 |
root |
1.25 |
db_open db_close db_compact db_sync db_upgrade |
118 |
|
|
db_put db_get db_pget db_del db_key_range |
119 |
root |
1.15 |
db_txn_commit db_txn_abort db_txn_finish |
120 |
root |
1.5 |
db_c_close db_c_count db_c_put db_c_get db_c_pget db_c_del |
121 |
root |
1.6 |
db_sequence_open db_sequence_close |
122 |
|
|
db_sequence_get db_sequence_remove |
123 |
|
|
); |
124 |
|
|
our @EXPORT = (@BDB_REQ, qw(dbreq_pri dbreq_nice db_env_create db_create)); |
125 |
|
|
our @EXPORT_OK = qw( |
126 |
|
|
poll_fileno poll_cb poll_wait flush |
127 |
|
|
min_parallel max_parallel max_idle |
128 |
|
|
nreqs nready npending nthreads |
129 |
|
|
max_poll_time max_poll_reqs |
130 |
root |
1.3 |
); |
131 |
root |
1.1 |
|
132 |
|
|
require XSLoader; |
133 |
root |
1.2 |
XSLoader::load ("BDB", $VERSION); |
134 |
root |
1.1 |
} |
135 |
|
|
|
136 |
root |
1.10 |
=head2 BERKELEYDB FUNCTIONS |
137 |
|
|
|
138 |
|
|
All of these are functions. The create functions simply return a new |
139 |
|
|
object and never block. All the remaining functions all take an optional |
140 |
root |
1.24 |
callback as last argument. If it is missing, then the function will be |
141 |
|
|
executed synchronously. In both cases, C<$!> will reflect the return value |
142 |
|
|
of the function. |
143 |
root |
1.10 |
|
144 |
|
|
BDB functions that cannot block (mostly functions that manipulate |
145 |
|
|
settings) are method calls on the relevant objects, so the rule of thumb |
146 |
|
|
is: if its a method, its not blocking, if its a function, it takes a |
147 |
|
|
callback as last argument. |
148 |
|
|
|
149 |
|
|
In the following, C<$int> signifies an integer return value, |
150 |
|
|
C<octetstring> is a "binary string" (i.e. a perl string with no character |
151 |
|
|
indices >255), C<U32> is an unsigned 32 bit integer, C<int> is some |
152 |
|
|
integer, C<NV> is a floating point value. |
153 |
|
|
|
154 |
|
|
The C<SV *> types are generic perl scalars (for input and output of data |
155 |
|
|
values), and the C<SV *callback> is the optional callback function to call |
156 |
|
|
when the request is completed. |
157 |
|
|
|
158 |
root |
1.11 |
The various C<DB_ENV> etc. arguments are handles return by |
159 |
|
|
C<db_env_create>, C<db_create>, C<txn_begin> and so on. If they have an |
160 |
|
|
appended C<_ornull> this means they are optional and you can pass C<undef> |
161 |
|
|
for them, resulting a NULL pointer on the C level. |
162 |
root |
1.10 |
|
163 |
|
|
=head3 BDB functions |
164 |
|
|
|
165 |
|
|
Functions in the BDB namespace, exported by default: |
166 |
|
|
|
167 |
|
|
$env = db_env_create (U32 env_flags = 0) |
168 |
root |
1.14 |
flags: RPCCLIENT |
169 |
root |
1.10 |
|
170 |
|
|
db_env_open (DB_ENV *env, octetstring db_home, U32 open_flags, int mode, SV *callback = &PL_sv_undef) |
171 |
root |
1.14 |
open_flags: INIT_CDB INIT_LOCK INIT_LOG INIT_MPOOL INIT_REP INIT_TXN RECOVER RECOVER_FATAL USE_ENVIRON USE_ENVIRON_ROOT CREATE LOCKDOWN PRIVATE REGISTER SYSTEM_MEM |
172 |
root |
1.10 |
db_env_close (DB_ENV *env, U32 flags = 0, SV *callback = &PL_sv_undef) |
173 |
|
|
db_env_txn_checkpoint (DB_ENV *env, U32 kbyte = 0, U32 min = 0, U32 flags = 0, SV *callback = &PL_sv_undef) |
174 |
root |
1.14 |
flags: FORCE |
175 |
root |
1.10 |
db_env_lock_detect (DB_ENV *env, U32 flags = 0, U32 atype = DB_LOCK_DEFAULT, SV *dummy = 0, SV *callback = &PL_sv_undef) |
176 |
root |
1.14 |
atype: LOCK_DEFAULT LOCK_EXPIRE LOCK_MAXLOCKS LOCK_MAXWRITE LOCK_MINLOCKS LOCK_MINWRITE LOCK_OLDEST LOCK_RANDOM LOCK_YOUNGEST |
177 |
root |
1.10 |
db_env_memp_sync (DB_ENV *env, SV *dummy = 0, SV *callback = &PL_sv_undef) |
178 |
|
|
db_env_memp_trickle (DB_ENV *env, int percent, SV *dummy = 0, SV *callback = &PL_sv_undef) |
179 |
|
|
|
180 |
|
|
$db = db_create (DB_ENV *env = 0, U32 flags = 0) |
181 |
root |
1.14 |
flags: XA_CREATE |
182 |
root |
1.10 |
|
183 |
|
|
db_open (DB *db, DB_TXN_ornull *txnid, octetstring file, octetstring database, int type, U32 flags, int mode, SV *callback = &PL_sv_undef) |
184 |
root |
1.14 |
flags: AUTO_COMMIT CREATE EXCL MULTIVERSION NOMMAP RDONLY READ_UNCOMMITTED THREAD TRUNCATE |
185 |
root |
1.10 |
db_close (DB *db, U32 flags = 0, SV *callback = &PL_sv_undef) |
186 |
root |
1.14 |
flags: DB_NOSYNC |
187 |
root |
1.25 |
db_upgrade (DB *db, octetstring file, U32 flags = 0, SV *callback = &PL_sv_undef) |
188 |
root |
1.12 |
db_compact (DB *db, DB_TXN_ornull *txn = 0, SV *start = 0, SV *stop = 0, SV *unused1 = 0, U32 flags = DB_FREE_SPACE, SV *unused2 = 0, SV *callback = &PL_sv_undef) |
189 |
root |
1.14 |
flags: FREELIST_ONLY FREE_SPACE |
190 |
root |
1.10 |
db_sync (DB *db, U32 flags = 0, SV *callback = &PL_sv_undef) |
191 |
|
|
db_key_range (DB *db, DB_TXN_ornull *txn, SV *key, SV *key_range, U32 flags = 0, SV *callback = &PL_sv_undef) |
192 |
|
|
db_put (DB *db, DB_TXN_ornull *txn, SV *key, SV *data, U32 flags = 0, SV *callback = &PL_sv_undef) |
193 |
root |
1.14 |
flags: APPEND NODUPDATA NOOVERWRITE |
194 |
root |
1.10 |
db_get (DB *db, DB_TXN_ornull *txn, SV *key, SV *data, U32 flags = 0, SV *callback = &PL_sv_undef) |
195 |
root |
1.14 |
flags: CONSUME CONSUME_WAIT GET_BOTH SET_RECNO MULTIPLE READ_COMMITTED READ_UNCOMMITTED RMW |
196 |
root |
1.10 |
db_pget (DB *db, DB_TXN_ornull *txn, SV *key, SV *pkey, SV *data, U32 flags = 0, SV *callback = &PL_sv_undef) |
197 |
root |
1.14 |
flags: CONSUME CONSUME_WAIT GET_BOTH SET_RECNO MULTIPLE READ_COMMITTED READ_UNCOMMITTED RMW |
198 |
root |
1.10 |
db_del (DB *db, DB_TXN_ornull *txn, SV *key, U32 flags = 0, SV *callback = &PL_sv_undef) |
199 |
|
|
db_txn_commit (DB_TXN *txn, U32 flags = 0, SV *callback = &PL_sv_undef) |
200 |
root |
1.14 |
flags: TXN_NOSYNC TXN_SYNC |
201 |
root |
1.10 |
db_txn_abort (DB_TXN *txn, SV *callback = &PL_sv_undef) |
202 |
root |
1.14 |
|
203 |
root |
1.10 |
db_c_close (DBC *dbc, SV *callback = &PL_sv_undef) |
204 |
|
|
db_c_count (DBC *dbc, SV *count, U32 flags = 0, SV *callback = &PL_sv_undef) |
205 |
|
|
db_c_put (DBC *dbc, SV *key, SV *data, U32 flags = 0, SV *callback = &PL_sv_undef) |
206 |
root |
1.14 |
flags: AFTER BEFORE CURRENT KEYFIRST KEYLAST NODUPDATA |
207 |
root |
1.10 |
db_c_get (DBC *dbc, SV *key, SV *data, U32 flags = 0, SV *callback = &PL_sv_undef) |
208 |
root |
1.14 |
flags: CURRENT FIRST GET_BOTH GET_BOTH_RANGE GET_RECNO JOIN_ITEM LAST NEXT NEXT_DUP NEXT_NODUP PREV PREV_DUP PREV_NODUP SET SET_RANGE SET_RECNO READ_UNCOMMITTED MULTIPLE MULTIPLE_KEY RMW |
209 |
root |
1.10 |
db_c_pget (DBC *dbc, SV *key, SV *pkey, SV *data, U32 flags = 0, SV *callback = &PL_sv_undef) |
210 |
|
|
db_c_del (DBC *dbc, U32 flags = 0, SV *callback = &PL_sv_undef) |
211 |
|
|
|
212 |
|
|
db_sequence_open (DB_SEQUENCE *seq, DB_TXN_ornull *txnid, SV *key, U32 flags = 0, SV *callback = &PL_sv_undef) |
213 |
root |
1.14 |
flags: CREATE EXCL |
214 |
root |
1.10 |
db_sequence_close (DB_SEQUENCE *seq, U32 flags = 0, SV *callback = &PL_sv_undef) |
215 |
|
|
db_sequence_get (DB_SEQUENCE *seq, DB_TXN_ornull *txnid, int delta, SV *seq_value, U32 flags = DB_TXN_NOSYNC, SV *callback = &PL_sv_undef) |
216 |
root |
1.14 |
flags: TXN_NOSYNC |
217 |
root |
1.10 |
db_sequence_remove (DB_SEQUENCE *seq, DB_TXN_ornull *txnid = 0, U32 flags = 0, SV *callback = &PL_sv_undef) |
218 |
root |
1.14 |
flags: TXN_NOSYNC |
219 |
root |
1.10 |
|
220 |
root |
1.15 |
=head4 db_txn_finish (DB_TXN *txn, U32 flags = 0, SV *callback = &PL_sv_undef) |
221 |
|
|
|
222 |
root |
1.16 |
This is not actually a Berkeley DB function but a BDB module |
223 |
|
|
extension. The background for this exytension is: It is very annoying to |
224 |
|
|
have to check every single BDB function for error returns and provide a |
225 |
|
|
codepath out of your transaction. While the BDB module still makes this |
226 |
|
|
possible, it contains the following extensions: |
227 |
root |
1.15 |
|
228 |
|
|
When a transaction-protected function returns any operating system |
229 |
|
|
error (errno > 0), BDB will set the C<TXN_DEADLOCK> flag on the |
230 |
root |
1.16 |
transaction. This flag is also set by Berkeley DB functions themselves |
231 |
root |
1.15 |
when an operation fails with LOCK_DEADLOCK, and it causes all further |
232 |
|
|
operations on that transaction (including C<db_txn_commit>) to fail. |
233 |
|
|
|
234 |
|
|
The C<db_txn_finish> request will look at this flag, and, if it is set, |
235 |
|
|
will automatically call C<db_txn_abort> (setting errno to C<LOCK_DEADLOCK> |
236 |
root |
1.16 |
if it isn't set to something else yet). If it isn't set, it will call |
237 |
|
|
C<db_txn_commit> and return the error normally. |
238 |
root |
1.15 |
|
239 |
|
|
How to use this? Easy: just write your transaction normally: |
240 |
|
|
|
241 |
|
|
my $txn = $db_env->txn_begin; |
242 |
|
|
db_get $db, $txn, "key", my $data; |
243 |
|
|
db_put $db, $txn, "key", $data + 1 unless $! == BDB::NOTFOUND; |
244 |
|
|
db_txn_finish $txn; |
245 |
|
|
die "transaction failed" if $!; |
246 |
|
|
|
247 |
|
|
That is, handle only the expected errors. If something unexpected happens |
248 |
|
|
(EIO, LOCK_NOTGRANTED or a deadlock in either db_get or db_put), then the remaining |
249 |
|
|
requests (db_put in this case) will simply be skipped (they will fail with |
250 |
|
|
LOCK_DEADLOCK) and the transaction will be aborted. |
251 |
|
|
|
252 |
root |
1.16 |
You can use the C<< $txn->failed >> method to check wether a transaction |
253 |
root |
1.15 |
has failed in this way and abort further processing (excluding |
254 |
|
|
C<db_txn_finish>). |
255 |
|
|
|
256 |
root |
1.10 |
=head3 DB_ENV/database environment methods |
257 |
|
|
|
258 |
|
|
Methods available on DB_ENV/$env handles: |
259 |
|
|
|
260 |
|
|
DESTROY (DB_ENV_ornull *env) |
261 |
|
|
CODE: |
262 |
|
|
if (env) |
263 |
|
|
env->close (env, 0); |
264 |
|
|
|
265 |
|
|
$int = $env->set_data_dir (const char *dir) |
266 |
|
|
$int = $env->set_tmp_dir (const char *dir) |
267 |
|
|
$int = $env->set_lg_dir (const char *dir) |
268 |
|
|
$int = $env->set_shm_key (long shm_key) |
269 |
|
|
$int = $env->set_cachesize (U32 gbytes, U32 bytes, int ncache = 0) |
270 |
|
|
$int = $env->set_flags (U32 flags, int onoff) |
271 |
|
|
$env->set_errfile (FILE *errfile = 0) |
272 |
|
|
$env->set_msgfile (FILE *msgfile = 0) |
273 |
|
|
$int = $env->set_verbose (U32 which, int onoff = 1) |
274 |
|
|
$int = $env->set_encrypt (const char *password, U32 flags = 0) |
275 |
root |
1.15 |
$int = $env->set_timeout (NV timeout_seconds, U32 flags = SET_TXN_TIMEOUT) |
276 |
root |
1.10 |
$int = $env->set_mp_max_openfd (int maxopenfd); |
277 |
|
|
$int = $env->set_mp_max_write (int maxwrite, int maxwrite_sleep); |
278 |
|
|
$int = $env->set_mp_mmapsize (int mmapsize_mb) |
279 |
|
|
$int = $env->set_lk_detect (U32 detect = DB_LOCK_DEFAULT) |
280 |
|
|
$int = $env->set_lk_max_lockers (U32 max) |
281 |
|
|
$int = $env->set_lk_max_locks (U32 max) |
282 |
|
|
$int = $env->set_lk_max_objects (U32 max) |
283 |
|
|
$int = $env->set_lg_bsize (U32 max) |
284 |
|
|
$int = $env->set_lg_max (U32 max) |
285 |
root |
1.20 |
$int = $env->mutex_set_increment (U32 increment) |
286 |
|
|
$int = $env->mutex_set_tas_spins (U32 tas_spins) |
287 |
|
|
$int = $env->mutex_set_max (U32 max) |
288 |
|
|
$int = $env->mutex_set_align (U32 align) |
289 |
root |
1.10 |
|
290 |
|
|
$txn = $env->txn_begin (DB_TXN_ornull *parent = 0, U32 flags = 0) |
291 |
root |
1.14 |
flags: READ_COMMITTED READ_UNCOMMITTED TXN_NOSYNC TXN_NOWAIT TXN_SNAPSHOT TXN_SYNC TXN_WAIT TXN_WRITE_NOSYNC |
292 |
root |
1.10 |
|
293 |
root |
1.12 |
=head4 Example: |
294 |
root |
1.10 |
|
295 |
|
|
use AnyEvent; |
296 |
|
|
use BDB; |
297 |
|
|
|
298 |
|
|
our $FH; open $FH, "<&=" . BDB::poll_fileno; |
299 |
|
|
our $WATCHER = AnyEvent->io (fh => $FH, poll => 'r', cb => \&BDB::poll_cb); |
300 |
|
|
|
301 |
|
|
BDB::min_parallel 8; |
302 |
|
|
|
303 |
|
|
my $env = db_env_create; |
304 |
|
|
|
305 |
|
|
mkdir "bdtest", 0700; |
306 |
|
|
db_env_open |
307 |
|
|
$env, |
308 |
|
|
"bdtest", |
309 |
|
|
BDB::INIT_LOCK | BDB::INIT_LOG | BDB::INIT_MPOOL | BDB::INIT_TXN | BDB::RECOVER | BDB::USE_ENVIRON | BDB::CREATE, |
310 |
|
|
0600; |
311 |
|
|
|
312 |
|
|
$env->set_flags (BDB::AUTO_COMMIT | BDB::TXN_NOSYNC, 1); |
313 |
|
|
|
314 |
|
|
|
315 |
|
|
=head3 DB/database methods |
316 |
|
|
|
317 |
|
|
Methods available on DB/$db handles: |
318 |
|
|
|
319 |
|
|
DESTROY (DB_ornull *db) |
320 |
|
|
CODE: |
321 |
|
|
if (db) |
322 |
|
|
{ |
323 |
|
|
SV *env = (SV *)db->app_private; |
324 |
|
|
db->close (db, 0); |
325 |
|
|
SvREFCNT_dec (env); |
326 |
|
|
} |
327 |
|
|
|
328 |
|
|
$int = $db->set_cachesize (U32 gbytes, U32 bytes, int ncache = 0) |
329 |
|
|
$int = $db->set_flags (U32 flags) |
330 |
root |
1.14 |
flags: CHKSUM ENCRYPT TXN_NOT_DURABLE |
331 |
|
|
Btree: DUP DUPSORT RECNUM REVSPLITOFF |
332 |
|
|
Hash: DUP DUPSORT |
333 |
|
|
Queue: INORDER |
334 |
|
|
Recno: RENUMBER SNAPSHOT |
335 |
|
|
|
336 |
root |
1.10 |
$int = $db->set_encrypt (const char *password, U32 flags) |
337 |
|
|
$int = $db->set_lorder (int lorder) |
338 |
|
|
$int = $db->set_bt_minkey (U32 minkey) |
339 |
|
|
$int = $db->set_re_delim (int delim) |
340 |
|
|
$int = $db->set_re_pad (int re_pad) |
341 |
|
|
$int = $db->set_re_source (char *source) |
342 |
|
|
$int = $db->set_re_len (U32 re_len) |
343 |
|
|
$int = $db->set_h_ffactor (U32 h_ffactor) |
344 |
|
|
$int = $db->set_h_nelem (U32 h_nelem) |
345 |
|
|
$int = $db->set_q_extentsize (U32 extentsize) |
346 |
|
|
|
347 |
|
|
$dbc = $db->cursor (DB_TXN_ornull *txn = 0, U32 flags = 0) |
348 |
root |
1.14 |
flags: READ_COMMITTED READ_UNCOMMITTED WRITECURSOR TXN_SNAPSHOT |
349 |
root |
1.10 |
$seq = $db->sequence (U32 flags = 0) |
350 |
|
|
|
351 |
root |
1.12 |
=head4 Example: |
352 |
root |
1.10 |
|
353 |
|
|
my $db = db_create $env; |
354 |
|
|
db_open $db, undef, "table", undef, BDB::BTREE, BDB::AUTO_COMMIT | BDB::CREATE | BDB::READ_UNCOMMITTED, 0600; |
355 |
|
|
|
356 |
|
|
for (1..1000) { |
357 |
|
|
db_put $db, undef, "key $_", "data $_"; |
358 |
|
|
|
359 |
|
|
db_key_range $db, undef, "key $_", my $keyrange; |
360 |
|
|
my ($lt, $eq, $gt) = @$keyrange; |
361 |
|
|
} |
362 |
|
|
|
363 |
|
|
db_del $db, undef, "key $_" for 1..1000; |
364 |
|
|
|
365 |
|
|
db_sync $db; |
366 |
|
|
|
367 |
|
|
|
368 |
|
|
=head3 DB_TXN/transaction methods |
369 |
|
|
|
370 |
|
|
Methods available on DB_TXN/$txn handles: |
371 |
|
|
|
372 |
|
|
DESTROY (DB_TXN_ornull *txn) |
373 |
|
|
CODE: |
374 |
|
|
if (txn) |
375 |
|
|
txn->abort (txn); |
376 |
|
|
|
377 |
root |
1.15 |
$int = $txn->set_timeout (NV timeout_seconds, U32 flags = SET_TXN_TIMEOUT) |
378 |
root |
1.14 |
flags: SET_LOCK_TIMEOUT SET_TXN_TIMEOUT |
379 |
root |
1.10 |
|
380 |
root |
1.15 |
$bool = $txn->failed |
381 |
|
|
# see db_txn_finish documentation, above |
382 |
|
|
|
383 |
root |
1.10 |
|
384 |
|
|
=head3 DBC/cursor methods |
385 |
|
|
|
386 |
|
|
Methods available on DBC/$dbc handles: |
387 |
|
|
|
388 |
|
|
DESTROY (DBC_ornull *dbc) |
389 |
|
|
CODE: |
390 |
|
|
if (dbc) |
391 |
|
|
dbc->c_close (dbc); |
392 |
|
|
|
393 |
root |
1.12 |
=head4 Example: |
394 |
root |
1.10 |
|
395 |
|
|
my $c = $db->cursor; |
396 |
|
|
|
397 |
|
|
for (;;) { |
398 |
|
|
db_c_get $c, my $key, my $data, BDB::NEXT; |
399 |
|
|
warn "<$!,$key,$data>"; |
400 |
|
|
last if $!; |
401 |
|
|
} |
402 |
|
|
|
403 |
|
|
db_c_close $c; |
404 |
|
|
|
405 |
root |
1.12 |
|
406 |
root |
1.10 |
=head3 DB_SEQUENCE/sequence methods |
407 |
|
|
|
408 |
|
|
Methods available on DB_SEQUENCE/$seq handles: |
409 |
|
|
|
410 |
|
|
DESTROY (DB_SEQUENCE_ornull *seq) |
411 |
|
|
CODE: |
412 |
|
|
if (seq) |
413 |
|
|
seq->close (seq, 0); |
414 |
|
|
|
415 |
|
|
$int = $seq->initial_value (db_seq_t value) |
416 |
|
|
$int = $seq->set_cachesize (U32 size) |
417 |
|
|
$int = $seq->set_flags (U32 flags) |
418 |
root |
1.14 |
flags: SEQ_DEC SEQ_INC SEQ_WRAP |
419 |
root |
1.10 |
$int = $seq->set_range (db_seq_t min, db_seq_t max) |
420 |
|
|
|
421 |
root |
1.12 |
=head4 Example: |
422 |
root |
1.10 |
|
423 |
|
|
my $seq = $db->sequence; |
424 |
|
|
|
425 |
|
|
db_sequence_open $seq, undef, "seq", BDB::CREATE; |
426 |
|
|
db_sequence_get $seq, undef, 1, my $value; |
427 |
|
|
|
428 |
|
|
|
429 |
root |
1.1 |
=head2 SUPPORT FUNCTIONS |
430 |
|
|
|
431 |
|
|
=head3 EVENT PROCESSING AND EVENT LOOP INTEGRATION |
432 |
|
|
|
433 |
|
|
=over 4 |
434 |
|
|
|
435 |
root |
1.23 |
=item $msg = BDB::strerror [$errno] |
436 |
|
|
|
437 |
|
|
Returns the string corresponding to the given errno value. If no argument |
438 |
|
|
is given, use C<$!>. |
439 |
|
|
|
440 |
root |
1.2 |
=item $fileno = BDB::poll_fileno |
441 |
root |
1.1 |
|
442 |
|
|
Return the I<request result pipe file descriptor>. This filehandle must be |
443 |
|
|
polled for reading by some mechanism outside this module (e.g. Event or |
444 |
|
|
select, see below or the SYNOPSIS). If the pipe becomes readable you have |
445 |
|
|
to call C<poll_cb> to check the results. |
446 |
|
|
|
447 |
|
|
See C<poll_cb> for an example. |
448 |
|
|
|
449 |
root |
1.2 |
=item BDB::poll_cb |
450 |
root |
1.1 |
|
451 |
|
|
Process some outstanding events on the result pipe. You have to call this |
452 |
|
|
regularly. Returns the number of events processed. Returns immediately |
453 |
|
|
when no events are outstanding. The amount of events processed depends on |
454 |
root |
1.2 |
the settings of C<BDB::max_poll_req> and C<BDB::max_poll_time>. |
455 |
root |
1.1 |
|
456 |
|
|
If not all requests were processed for whatever reason, the filehandle |
457 |
|
|
will still be ready when C<poll_cb> returns. |
458 |
|
|
|
459 |
|
|
Example: Install an Event watcher that automatically calls |
460 |
root |
1.2 |
BDB::poll_cb with high priority: |
461 |
root |
1.1 |
|
462 |
root |
1.2 |
Event->io (fd => BDB::poll_fileno, |
463 |
root |
1.1 |
poll => 'r', async => 1, |
464 |
root |
1.2 |
cb => \&BDB::poll_cb); |
465 |
root |
1.1 |
|
466 |
root |
1.2 |
=item BDB::max_poll_reqs $nreqs |
467 |
root |
1.1 |
|
468 |
root |
1.2 |
=item BDB::max_poll_time $seconds |
469 |
root |
1.1 |
|
470 |
|
|
These set the maximum number of requests (default C<0>, meaning infinity) |
471 |
root |
1.2 |
that are being processed by C<BDB::poll_cb> in one call, respectively |
472 |
root |
1.1 |
the maximum amount of time (default C<0>, meaning infinity) spent in |
473 |
root |
1.2 |
C<BDB::poll_cb> to process requests (more correctly the mininum amount |
474 |
root |
1.1 |
of time C<poll_cb> is allowed to use). |
475 |
|
|
|
476 |
|
|
Setting C<max_poll_time> to a non-zero value creates an overhead of one |
477 |
|
|
syscall per request processed, which is not normally a problem unless your |
478 |
|
|
callbacks are really really fast or your OS is really really slow (I am |
479 |
|
|
not mentioning Solaris here). Using C<max_poll_reqs> incurs no overhead. |
480 |
|
|
|
481 |
|
|
Setting these is useful if you want to ensure some level of |
482 |
|
|
interactiveness when perl is not fast enough to process all requests in |
483 |
|
|
time. |
484 |
|
|
|
485 |
|
|
For interactive programs, values such as C<0.01> to C<0.1> should be fine. |
486 |
|
|
|
487 |
root |
1.18 |
Example: Install an EV watcher that automatically calls |
488 |
root |
1.2 |
BDB::poll_cb with low priority, to ensure that other parts of the |
489 |
root |
1.18 |
program get the CPU sometimes even under high load. |
490 |
root |
1.1 |
|
491 |
|
|
# try not to spend much more than 0.1s in poll_cb |
492 |
root |
1.2 |
BDB::max_poll_time 0.1; |
493 |
root |
1.1 |
|
494 |
root |
1.18 |
my $bdb_poll = EV::io BDB::poll_fileno, EV::READ, \&BDB::poll_cb); |
495 |
root |
1.1 |
|
496 |
root |
1.2 |
=item BDB::poll_wait |
497 |
root |
1.1 |
|
498 |
|
|
If there are any outstanding requests and none of them in the result |
499 |
|
|
phase, wait till the result filehandle becomes ready for reading (simply |
500 |
|
|
does a C<select> on the filehandle. This is useful if you want to |
501 |
|
|
synchronously wait for some requests to finish). |
502 |
|
|
|
503 |
|
|
See C<nreqs> for an example. |
504 |
|
|
|
505 |
root |
1.2 |
=item BDB::poll |
506 |
root |
1.1 |
|
507 |
|
|
Waits until some requests have been handled. |
508 |
|
|
|
509 |
|
|
Returns the number of requests processed, but is otherwise strictly |
510 |
|
|
equivalent to: |
511 |
|
|
|
512 |
root |
1.2 |
BDB::poll_wait, BDB::poll_cb |
513 |
root |
1.1 |
|
514 |
root |
1.2 |
=item BDB::flush |
515 |
root |
1.1 |
|
516 |
root |
1.18 |
Wait till all outstanding BDB requests have been handled. |
517 |
root |
1.1 |
|
518 |
|
|
Strictly equivalent to: |
519 |
|
|
|
520 |
root |
1.2 |
BDB::poll_wait, BDB::poll_cb |
521 |
|
|
while BDB::nreqs; |
522 |
root |
1.1 |
|
523 |
root |
1.8 |
=back |
524 |
|
|
|
525 |
root |
1.1 |
=head3 CONTROLLING THE NUMBER OF THREADS |
526 |
|
|
|
527 |
root |
1.8 |
=over 4 |
528 |
|
|
|
529 |
root |
1.2 |
=item BDB::min_parallel $nthreads |
530 |
root |
1.1 |
|
531 |
root |
1.18 |
Set the minimum number of BDB threads to C<$nthreads>. The current |
532 |
root |
1.1 |
default is C<8>, which means eight asynchronous operations can execute |
533 |
|
|
concurrently at any one time (the number of outstanding requests, |
534 |
|
|
however, is unlimited). |
535 |
|
|
|
536 |
root |
1.18 |
BDB starts threads only on demand, when an BDB request is queued and |
537 |
root |
1.1 |
no free thread exists. Please note that queueing up a hundred requests can |
538 |
|
|
create demand for a hundred threads, even if it turns out that everything |
539 |
|
|
is in the cache and could have been processed faster by a single thread. |
540 |
|
|
|
541 |
|
|
It is recommended to keep the number of threads relatively low, as some |
542 |
|
|
Linux kernel versions will scale negatively with the number of threads |
543 |
|
|
(higher parallelity => MUCH higher latency). With current Linux 2.6 |
544 |
|
|
versions, 4-32 threads should be fine. |
545 |
|
|
|
546 |
|
|
Under most circumstances you don't need to call this function, as the |
547 |
|
|
module selects a default that is suitable for low to moderate load. |
548 |
|
|
|
549 |
root |
1.2 |
=item BDB::max_parallel $nthreads |
550 |
root |
1.1 |
|
551 |
root |
1.18 |
Sets the maximum number of BDB threads to C<$nthreads>. If more than the |
552 |
root |
1.1 |
specified number of threads are currently running, this function kills |
553 |
|
|
them. This function blocks until the limit is reached. |
554 |
|
|
|
555 |
|
|
While C<$nthreads> are zero, aio requests get queued but not executed |
556 |
|
|
until the number of threads has been increased again. |
557 |
|
|
|
558 |
|
|
This module automatically runs C<max_parallel 0> at program end, to ensure |
559 |
|
|
that all threads are killed and that there are no outstanding requests. |
560 |
|
|
|
561 |
|
|
Under normal circumstances you don't need to call this function. |
562 |
|
|
|
563 |
root |
1.2 |
=item BDB::max_idle $nthreads |
564 |
root |
1.1 |
|
565 |
|
|
Limit the number of threads (default: 4) that are allowed to idle (i.e., |
566 |
|
|
threads that did not get a request to process within 10 seconds). That |
567 |
|
|
means if a thread becomes idle while C<$nthreads> other threads are also |
568 |
|
|
idle, it will free its resources and exit. |
569 |
|
|
|
570 |
|
|
This is useful when you allow a large number of threads (e.g. 100 or 1000) |
571 |
|
|
to allow for extremely high load situations, but want to free resources |
572 |
|
|
under normal circumstances (1000 threads can easily consume 30MB of RAM). |
573 |
|
|
|
574 |
|
|
The default is probably ok in most situations, especially if thread |
575 |
|
|
creation is fast. If thread creation is very slow on your system you might |
576 |
|
|
want to use larger values. |
577 |
|
|
|
578 |
root |
1.2 |
=item $oldmaxreqs = BDB::max_outstanding $maxreqs |
579 |
root |
1.1 |
|
580 |
|
|
This is a very bad function to use in interactive programs because it |
581 |
|
|
blocks, and a bad way to reduce concurrency because it is inexact: Better |
582 |
|
|
use an C<aio_group> together with a feed callback. |
583 |
|
|
|
584 |
|
|
Sets the maximum number of outstanding requests to C<$nreqs>. If you |
585 |
|
|
to queue up more than this number of requests, the next call to the |
586 |
|
|
C<poll_cb> (and C<poll_some> and other functions calling C<poll_cb>) |
587 |
|
|
function will block until the limit is no longer exceeded. |
588 |
|
|
|
589 |
|
|
The default value is very large, so there is no practical limit on the |
590 |
|
|
number of outstanding requests. |
591 |
|
|
|
592 |
|
|
You can still queue as many requests as you want. Therefore, |
593 |
|
|
C<max_oustsanding> is mainly useful in simple scripts (with low values) or |
594 |
|
|
as a stop gap to shield against fatal memory overflow (with large values). |
595 |
|
|
|
596 |
root |
1.3 |
=item BDB::set_sync_prepare $cb |
597 |
|
|
|
598 |
|
|
Sets a callback that is called whenever a request is created without an |
599 |
|
|
explicit callback. It has to return two code references. The first is used |
600 |
|
|
as the request callback, and the second is called to wait until the first |
601 |
|
|
callback has been called. The default implementation works like this: |
602 |
|
|
|
603 |
|
|
sub { |
604 |
|
|
my $status; |
605 |
|
|
( |
606 |
|
|
sub { $status = $! }, |
607 |
|
|
sub { BDB::poll while !defined $status; $! = $status }, |
608 |
|
|
) |
609 |
|
|
} |
610 |
|
|
|
611 |
|
|
=back |
612 |
|
|
|
613 |
root |
1.1 |
=head3 STATISTICAL INFORMATION |
614 |
|
|
|
615 |
root |
1.3 |
=over 4 |
616 |
|
|
|
617 |
root |
1.2 |
=item BDB::nreqs |
618 |
root |
1.1 |
|
619 |
|
|
Returns the number of requests currently in the ready, execute or pending |
620 |
|
|
states (i.e. for which their callback has not been invoked yet). |
621 |
|
|
|
622 |
|
|
Example: wait till there are no outstanding requests anymore: |
623 |
|
|
|
624 |
root |
1.2 |
BDB::poll_wait, BDB::poll_cb |
625 |
|
|
while BDB::nreqs; |
626 |
root |
1.1 |
|
627 |
root |
1.2 |
=item BDB::nready |
628 |
root |
1.1 |
|
629 |
|
|
Returns the number of requests currently in the ready state (not yet |
630 |
|
|
executed). |
631 |
|
|
|
632 |
root |
1.2 |
=item BDB::npending |
633 |
root |
1.1 |
|
634 |
|
|
Returns the number of requests currently in the pending state (executed, |
635 |
|
|
but not yet processed by poll_cb). |
636 |
|
|
|
637 |
|
|
=back |
638 |
|
|
|
639 |
|
|
=cut |
640 |
|
|
|
641 |
root |
1.3 |
set_sync_prepare { |
642 |
|
|
my $status; |
643 |
|
|
( |
644 |
|
|
sub { |
645 |
|
|
$status = $!; |
646 |
|
|
}, |
647 |
|
|
sub { |
648 |
|
|
BDB::poll while !defined $status; |
649 |
|
|
$! = $status; |
650 |
|
|
}, |
651 |
|
|
) |
652 |
|
|
}; |
653 |
|
|
|
654 |
root |
1.1 |
min_parallel 8; |
655 |
|
|
|
656 |
|
|
END { flush } |
657 |
|
|
|
658 |
|
|
1; |
659 |
|
|
|
660 |
|
|
=head2 FORK BEHAVIOUR |
661 |
|
|
|
662 |
|
|
This module should do "the right thing" when the process using it forks: |
663 |
|
|
|
664 |
root |
1.18 |
Before the fork, BDB enters a quiescent state where no requests |
665 |
root |
1.1 |
can be added in other threads and no results will be processed. After |
666 |
|
|
the fork the parent simply leaves the quiescent state and continues |
667 |
|
|
request/result processing, while the child frees the request/result queue |
668 |
|
|
(so that the requests started before the fork will only be handled in the |
669 |
|
|
parent). Threads will be started on demand until the limit set in the |
670 |
|
|
parent process has been reached again. |
671 |
|
|
|
672 |
|
|
In short: the parent will, after a short pause, continue as if fork had |
673 |
root |
1.18 |
not been called, while the child will act as if BDB has not been used |
674 |
root |
1.1 |
yet. |
675 |
|
|
|
676 |
root |
1.26 |
Win32 note: there is no fork on win32, and perls emulation of it is too |
677 |
|
|
broken to be supported, so do not use BDB in a windows pseudo-fork, better |
678 |
|
|
yet, switch to a more capable platform. |
679 |
|
|
|
680 |
root |
1.1 |
=head2 MEMORY USAGE |
681 |
|
|
|
682 |
|
|
Per-request usage: |
683 |
|
|
|
684 |
|
|
Each aio request uses - depending on your architecture - around 100-200 |
685 |
|
|
bytes of memory. In addition, stat requests need a stat buffer (possibly |
686 |
|
|
a few hundred bytes), readdir requires a result buffer and so on. Perl |
687 |
|
|
scalars and other data passed into aio requests will also be locked and |
688 |
|
|
will consume memory till the request has entered the done state. |
689 |
|
|
|
690 |
root |
1.13 |
This is not awfully much, so queuing lots of requests is not usually a |
691 |
root |
1.1 |
problem. |
692 |
|
|
|
693 |
|
|
Per-thread usage: |
694 |
|
|
|
695 |
|
|
In the execution phase, some aio requests require more memory for |
696 |
|
|
temporary buffers, and each thread requires a stack and other data |
697 |
|
|
structures (usually around 16k-128k, depending on the OS). |
698 |
|
|
|
699 |
|
|
=head1 KNOWN BUGS |
700 |
|
|
|
701 |
root |
1.15 |
Known bugs will be fixed in the next release, except: |
702 |
|
|
|
703 |
|
|
If you use a transaction in any request, and the request returns |
704 |
|
|
with an operating system error or DB_LOCK_NOTGRANTED, the internal |
705 |
|
|
TXN_DEADLOCK flag will be set on the transaction. See C<db_txn_finish>, |
706 |
|
|
above. |
707 |
root |
1.1 |
|
708 |
|
|
=head1 SEE ALSO |
709 |
|
|
|
710 |
root |
1.18 |
L<Coro::BDB>, L<IO::AIO>. |
711 |
root |
1.1 |
|
712 |
|
|
=head1 AUTHOR |
713 |
|
|
|
714 |
|
|
Marc Lehmann <schmorp@schmorp.de> |
715 |
|
|
http://home.schmorp.de/ |
716 |
|
|
|
717 |
|
|
=cut |
718 |
|
|
|