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