… | |
… | |
72 | use strict 'vars'; |
72 | use strict 'vars'; |
73 | |
73 | |
74 | use base 'Exporter'; |
74 | use base 'Exporter'; |
75 | |
75 | |
76 | BEGIN { |
76 | BEGIN { |
77 | our $VERSION = '1.1'; |
77 | our $VERSION = '1.2'; |
78 | |
78 | |
79 | our @BDB_REQ = qw( |
79 | our @BDB_REQ = qw( |
80 | db_env_open db_env_close db_env_txn_checkpoint db_env_lock_detect |
80 | db_env_open db_env_close db_env_txn_checkpoint db_env_lock_detect |
81 | db_env_memp_sync db_env_memp_trickle |
81 | db_env_memp_sync db_env_memp_trickle |
82 | db_open db_close db_compact db_sync db_put db_get db_pget db_del db_key_range |
82 | db_open db_close db_compact db_sync db_put db_get db_pget db_del db_key_range |
… | |
… | |
179 | db_sequence_remove (DB_SEQUENCE *seq, DB_TXN_ornull *txnid = 0, U32 flags = 0, SV *callback = &PL_sv_undef) |
179 | db_sequence_remove (DB_SEQUENCE *seq, DB_TXN_ornull *txnid = 0, U32 flags = 0, SV *callback = &PL_sv_undef) |
180 | flags: TXN_NOSYNC |
180 | flags: TXN_NOSYNC |
181 | |
181 | |
182 | =head4 db_txn_finish (DB_TXN *txn, U32 flags = 0, SV *callback = &PL_sv_undef) |
182 | =head4 db_txn_finish (DB_TXN *txn, U32 flags = 0, SV *callback = &PL_sv_undef) |
183 | |
183 | |
184 | This is not a Berkeley DB function but a BDB module extension. It is very |
184 | This is not actually a Berkeley DB function but a BDB module |
|
|
185 | extension. The background for this exytension is: It is very annoying to |
185 | annoying to have to check every single BDB function for error returns and |
186 | have to check every single BDB function for error returns and provide a |
186 | provide a codepath out of your transaction. While the BDB module still |
187 | codepath out of your transaction. While the BDB module still makes this |
187 | makes this possible, it contains the following extensions: |
188 | possible, it contains the following extensions: |
188 | |
189 | |
189 | When a transaction-protected function returns any operating system |
190 | When a transaction-protected function returns any operating system |
190 | error (errno > 0), BDB will set the C<TXN_DEADLOCK> flag on the |
191 | error (errno > 0), BDB will set the C<TXN_DEADLOCK> flag on the |
191 | transaction. This flag is also set by Berkeley DB functions externally |
192 | transaction. This flag is also set by Berkeley DB functions themselves |
192 | when an operation fails with LOCK_DEADLOCK, and it causes all further |
193 | when an operation fails with LOCK_DEADLOCK, and it causes all further |
193 | operations on that transaction (including C<db_txn_commit>) to fail. |
194 | operations on that transaction (including C<db_txn_commit>) to fail. |
194 | |
195 | |
195 | The C<db_txn_finish> request will look at this flag, and, if it is set, |
196 | The C<db_txn_finish> request will look at this flag, and, if it is set, |
196 | will automatically call C<db_txn_abort> (setting errno to C<LOCK_DEADLOCK> |
197 | will automatically call C<db_txn_abort> (setting errno to C<LOCK_DEADLOCK> |
197 | if it isn't set). If it isn't set, it will call C<db_txn_commit> and |
198 | if it isn't set to something else yet). If it isn't set, it will call |
198 | return the error normally. |
199 | C<db_txn_commit> and return the error normally. |
199 | |
200 | |
200 | How to use this? Easy: just write your transaction normally: |
201 | How to use this? Easy: just write your transaction normally: |
201 | |
202 | |
202 | my $txn = $db_env->txn_begin; |
203 | my $txn = $db_env->txn_begin; |
203 | db_get $db, $txn, "key", my $data; |
204 | db_get $db, $txn, "key", my $data; |
… | |
… | |
208 | That is, handle only the expected errors. If something unexpected happens |
209 | That is, handle only the expected errors. If something unexpected happens |
209 | (EIO, LOCK_NOTGRANTED or a deadlock in either db_get or db_put), then the remaining |
210 | (EIO, LOCK_NOTGRANTED or a deadlock in either db_get or db_put), then the remaining |
210 | requests (db_put in this case) will simply be skipped (they will fail with |
211 | requests (db_put in this case) will simply be skipped (they will fail with |
211 | LOCK_DEADLOCK) and the transaction will be aborted. |
212 | LOCK_DEADLOCK) and the transaction will be aborted. |
212 | |
213 | |
213 | You cna use the C<< $txn->failed >> method to check wether a transaction |
214 | You can use the C<< $txn->failed >> method to check wether a transaction |
214 | has failed in this way and abort further processing (excluding |
215 | has failed in this way and abort further processing (excluding |
215 | C<db_txn_finish>). |
216 | C<db_txn_finish>). |
216 | |
217 | |
217 | =head3 DB_ENV/database environment methods |
218 | =head3 DB_ENV/database environment methods |
218 | |
219 | |
… | |
… | |
434 | interactiveness when perl is not fast enough to process all requests in |
435 | interactiveness when perl is not fast enough to process all requests in |
435 | time. |
436 | time. |
436 | |
437 | |
437 | For interactive programs, values such as C<0.01> to C<0.1> should be fine. |
438 | For interactive programs, values such as C<0.01> to C<0.1> should be fine. |
438 | |
439 | |
439 | Example: Install an Event watcher that automatically calls |
440 | Example: Install an EV watcher that automatically calls |
440 | BDB::poll_cb with low priority, to ensure that other parts of the |
441 | BDB::poll_cb with low priority, to ensure that other parts of the |
441 | program get the CPU sometimes even under high AIO load. |
442 | program get the CPU sometimes even under high load. |
442 | |
443 | |
443 | # try not to spend much more than 0.1s in poll_cb |
444 | # try not to spend much more than 0.1s in poll_cb |
444 | BDB::max_poll_time 0.1; |
445 | BDB::max_poll_time 0.1; |
445 | |
446 | |
446 | # use a low priority so other tasks have priority |
447 | my $bdb_poll = EV::io BDB::poll_fileno, EV::READ, \&BDB::poll_cb); |
447 | Event->io (fd => BDB::poll_fileno, |
|
|
448 | poll => 'r', nice => 1, |
|
|
449 | cb => &BDB::poll_cb); |
|
|
450 | |
448 | |
451 | =item BDB::poll_wait |
449 | =item BDB::poll_wait |
452 | |
450 | |
453 | If there are any outstanding requests and none of them in the result |
451 | If there are any outstanding requests and none of them in the result |
454 | phase, wait till the result filehandle becomes ready for reading (simply |
452 | phase, wait till the result filehandle becomes ready for reading (simply |
… | |
… | |
466 | |
464 | |
467 | BDB::poll_wait, BDB::poll_cb |
465 | BDB::poll_wait, BDB::poll_cb |
468 | |
466 | |
469 | =item BDB::flush |
467 | =item BDB::flush |
470 | |
468 | |
471 | Wait till all outstanding AIO requests have been handled. |
469 | Wait till all outstanding BDB requests have been handled. |
472 | |
470 | |
473 | Strictly equivalent to: |
471 | Strictly equivalent to: |
474 | |
472 | |
475 | BDB::poll_wait, BDB::poll_cb |
473 | BDB::poll_wait, BDB::poll_cb |
476 | while BDB::nreqs; |
474 | while BDB::nreqs; |
… | |
… | |
481 | |
479 | |
482 | =over 4 |
480 | =over 4 |
483 | |
481 | |
484 | =item BDB::min_parallel $nthreads |
482 | =item BDB::min_parallel $nthreads |
485 | |
483 | |
486 | Set the minimum number of AIO threads to C<$nthreads>. The current |
484 | Set the minimum number of BDB threads to C<$nthreads>. The current |
487 | default is C<8>, which means eight asynchronous operations can execute |
485 | default is C<8>, which means eight asynchronous operations can execute |
488 | concurrently at any one time (the number of outstanding requests, |
486 | concurrently at any one time (the number of outstanding requests, |
489 | however, is unlimited). |
487 | however, is unlimited). |
490 | |
488 | |
491 | BDB starts threads only on demand, when an AIO request is queued and |
489 | BDB starts threads only on demand, when an BDB request is queued and |
492 | no free thread exists. Please note that queueing up a hundred requests can |
490 | no free thread exists. Please note that queueing up a hundred requests can |
493 | create demand for a hundred threads, even if it turns out that everything |
491 | create demand for a hundred threads, even if it turns out that everything |
494 | is in the cache and could have been processed faster by a single thread. |
492 | is in the cache and could have been processed faster by a single thread. |
495 | |
493 | |
496 | It is recommended to keep the number of threads relatively low, as some |
494 | It is recommended to keep the number of threads relatively low, as some |
… | |
… | |
501 | Under most circumstances you don't need to call this function, as the |
499 | Under most circumstances you don't need to call this function, as the |
502 | module selects a default that is suitable for low to moderate load. |
500 | module selects a default that is suitable for low to moderate load. |
503 | |
501 | |
504 | =item BDB::max_parallel $nthreads |
502 | =item BDB::max_parallel $nthreads |
505 | |
503 | |
506 | Sets the maximum number of AIO threads to C<$nthreads>. If more than the |
504 | Sets the maximum number of BDB threads to C<$nthreads>. If more than the |
507 | specified number of threads are currently running, this function kills |
505 | specified number of threads are currently running, this function kills |
508 | them. This function blocks until the limit is reached. |
506 | them. This function blocks until the limit is reached. |
509 | |
507 | |
510 | While C<$nthreads> are zero, aio requests get queued but not executed |
508 | While C<$nthreads> are zero, aio requests get queued but not executed |
511 | until the number of threads has been increased again. |
509 | until the number of threads has been increased again. |
… | |
… | |
614 | |
612 | |
615 | =head2 FORK BEHAVIOUR |
613 | =head2 FORK BEHAVIOUR |
616 | |
614 | |
617 | This module should do "the right thing" when the process using it forks: |
615 | This module should do "the right thing" when the process using it forks: |
618 | |
616 | |
619 | Before the fork, IO::AIO enters a quiescent state where no requests |
617 | Before the fork, BDB enters a quiescent state where no requests |
620 | can be added in other threads and no results will be processed. After |
618 | can be added in other threads and no results will be processed. After |
621 | the fork the parent simply leaves the quiescent state and continues |
619 | the fork the parent simply leaves the quiescent state and continues |
622 | request/result processing, while the child frees the request/result queue |
620 | request/result processing, while the child frees the request/result queue |
623 | (so that the requests started before the fork will only be handled in the |
621 | (so that the requests started before the fork will only be handled in the |
624 | parent). Threads will be started on demand until the limit set in the |
622 | parent). Threads will be started on demand until the limit set in the |
625 | parent process has been reached again. |
623 | parent process has been reached again. |
626 | |
624 | |
627 | In short: the parent will, after a short pause, continue as if fork had |
625 | In short: the parent will, after a short pause, continue as if fork had |
628 | not been called, while the child will act as if IO::AIO has not been used |
626 | not been called, while the child will act as if BDB has not been used |
629 | yet. |
627 | yet. |
630 | |
628 | |
631 | =head2 MEMORY USAGE |
629 | =head2 MEMORY USAGE |
632 | |
630 | |
633 | Per-request usage: |
631 | Per-request usage: |
… | |
… | |
656 | TXN_DEADLOCK flag will be set on the transaction. See C<db_txn_finish>, |
654 | TXN_DEADLOCK flag will be set on the transaction. See C<db_txn_finish>, |
657 | above. |
655 | above. |
658 | |
656 | |
659 | =head1 SEE ALSO |
657 | =head1 SEE ALSO |
660 | |
658 | |
661 | L<Coro::AIO>. |
659 | L<Coro::BDB>, L<IO::AIO>. |
662 | |
660 | |
663 | =head1 AUTHOR |
661 | =head1 AUTHOR |
664 | |
662 | |
665 | Marc Lehmann <schmorp@schmorp.de> |
663 | Marc Lehmann <schmorp@schmorp.de> |
666 | http://home.schmorp.de/ |
664 | http://home.schmorp.de/ |