[Tarantool-patches] [PATCH 13/13] txn: stop TXN_SIGNATURE_ABORT override
Serge Petrenko
sergepetrenko at tarantool.org
Tue Jun 15 16:44:07 MSK 2021
12.06.2021 00:56, Vladislav Shpilevoy пишет:
> When txn_commit/try_async() failed before going to WAL thread,
> they installed TXN_SIGNATURE_ABORT signature meaning that the
> caller and the rollback triggers must look at the global diag.
>
> But they called txn_rollback() before doing return and calling the
> triggers, which overrode the signature with TXN_SIGNATURE_ROLLBACK
> leading to the original error loss.
>
> The patch makes TXN_SIGNATURE_ROLLBACK installed only when a real
> rollback happens (via box_txn_rollback()).
>
> This makes the original commit errors like a conflict in the
> transaction manager and OOM not lost.
>
> Besides, ERRINJ_TXN_COMMIT_ASYNC does not need its own diag_log()
> anymore. Because since this commit the applier logs the correct
> error instead of ER_WAL_IO/ER_TXN_ROLLBACK.
>
> Closes #6027
> ---
> .../unreleased/gh-6027-applier-lost-error.md | 7 ++
> src/box/txn.c | 10 ++-
> .../gh-6027-applier-error-show.result | 82 +++++++++++++++++++
> .../gh-6027-applier-error-show.test.lua | 31 +++++++
> test/replication/suite.cfg | 1 +
> test/replication/suite.ini | 2 +-
> 6 files changed, 129 insertions(+), 4 deletions(-)
> create mode 100644 changelogs/unreleased/gh-6027-applier-lost-error.md
> create mode 100644 test/replication/gh-6027-applier-error-show.result
> create mode 100644 test/replication/gh-6027-applier-error-show.test.lua
>
> diff --git a/changelogs/unreleased/gh-6027-applier-lost-error.md b/changelogs/unreleased/gh-6027-applier-lost-error.md
> new file mode 100644
> index 000000000..9c765b8e2
> --- /dev/null
> +++ b/changelogs/unreleased/gh-6027-applier-lost-error.md
> @@ -0,0 +1,7 @@
> +## bugfix/replication
> +
> +* When an error happened during appliance of a transaction received from a
> + remote instance via replication, it was always reported as "Failed to write
> + to disk" regardless of what really happened. Now the correct error is shown.
> + For example, "Out of memory", or "Transaction has been aborted by conflict",
> + and so on (gh-6027).
> diff --git a/src/box/txn.c b/src/box/txn.c
> index 5cae7b41d..c2734c237 100644
> --- a/src/box/txn.c
> +++ b/src/box/txn.c
> @@ -805,7 +805,6 @@ txn_commit_try_async(struct txn *txn)
> * Log it for the testing sake: we grep
> * output to mark this event.
> */
The comment became rather misleading now.
We log the error in applier, not here.
I would drop it altogether, but up to you.
> - diag_log();
> goto rollback;
> });
>
> @@ -983,11 +982,11 @@ void
> txn_rollback(struct txn *txn)
> {
> assert(txn == in_txn());
> + assert(txn->signature != TXN_SIGNATURE_UNKNOWN);
> txn->status = TXN_ABORTED;
> trigger_clear(&txn->fiber_on_stop);
> if (!txn_has_flag(txn, TXN_CAN_YIELD))
> trigger_clear(&txn->fiber_on_yield);
> - txn->signature = TXN_SIGNATURE_ROLLBACK;
> txn_complete_fail(txn);
> fiber_set_txn(fiber(), NULL);
> }
> @@ -1086,6 +1085,8 @@ box_txn_rollback(void)
> diag_set(ClientError, ER_ROLLBACK_IN_SUB_STMT);
> return -1;
> }
> + assert(txn->signature == TXN_SIGNATURE_UNKNOWN);
> + txn->signature = TXN_SIGNATURE_ROLLBACK;
> txn_rollback(txn); /* doesn't throw */
> fiber_gc();
> return 0;
> @@ -1221,7 +1222,10 @@ txn_on_stop(struct trigger *trigger, void *event)
> {
> (void) trigger;
> (void) event;
> - txn_rollback(in_txn()); /* doesn't yield or fail */
> + struct txn *txn = in_txn();
> + assert(txn->signature == TXN_SIGNATURE_UNKNOWN);
> + txn->signature = TXN_SIGNATURE_ROLLBACK;
> + txn_rollback(txn);
> fiber_gc();
> return 0;
> }
> diff --git a/test/replication/gh-6027-applier-error-show.result b/test/replication/gh-6027-applier-error-show.result
> new file mode 100644
> index 000000000..c3a01ab50
> --- /dev/null
> +++ b/test/replication/gh-6027-applier-error-show.result
> @@ -0,0 +1,82 @@
> +-- test-run result file version 2
> +test_run = require('test_run').new()
> + | ---
> + | ...
> +
> +--
> +-- gh-6027: on attempt to a commit transaction its original error was lost.
> +--
> +
> +box.schema.user.grant('guest', 'super')
> + | ---
> + | ...
> +s = box.schema.create_space('test')
> + | ---
> + | ...
> +_ = s:create_index('pk')
> + | ---
> + | ...
> +
> +test_run:cmd('create server replica with rpl_master=default, '.. \
> + 'script="replication/replica.lua"')
> + | ---
> + | - true
> + | ...
> +test_run:cmd('start server replica')
> + | ---
> + | - true
> + | ...
> +
> +test_run:switch('replica')
> + | ---
> + | - true
> + | ...
> +box.error.injection.set('ERRINJ_TXN_COMMIT_ASYNC', true)
> + | ---
> + | - ok
> + | ...
> +
> +test_run:switch('default')
> + | ---
> + | - true
> + | ...
> +_ = s:replace{1}
> + | ---
> + | ...
> +
> +test_run:switch('replica')
> + | ---
> + | - true
> + | ...
> +test_run:wait_upstream(1, {status = 'stopped'})
> + | ---
> + | - true
> + | ...
> +-- Should be something about error injection.
> +box.info.replication[1].upstream.message
> + | ---
> + | - Error injection 'txn commit async injection'
> + | ...
> +
> +test_run:switch('default')
> + | ---
> + | - true
> + | ...
> +test_run:cmd('stop server replica')
> + | ---
> + | - true
> + | ...
> +test_run:cmd('delete server replica')
> + | ---
> + | - true
> + | ...
> +box.error.injection.set('ERRINJ_TXN_COMMIT_ASYNC', false)
> + | ---
> + | - ok
> + | ...
> +s:drop()
> + | ---
> + | ...
> +box.schema.user.revoke('guest', 'super')
> + | ---
> + | ...
> diff --git a/test/replication/gh-6027-applier-error-show.test.lua b/test/replication/gh-6027-applier-error-show.test.lua
> new file mode 100644
> index 000000000..8e17cdfa9
> --- /dev/null
> +++ b/test/replication/gh-6027-applier-error-show.test.lua
> @@ -0,0 +1,31 @@
> +test_run = require('test_run').new()
> +
> +--
> +-- gh-6027: on attempt to a commit transaction its original error was lost.
> +--
> +
> +box.schema.user.grant('guest', 'super')
> +s = box.schema.create_space('test')
> +_ = s:create_index('pk')
> +
> +test_run:cmd('create server replica with rpl_master=default, '.. \
> + 'script="replication/replica.lua"')
> +test_run:cmd('start server replica')
> +
> +test_run:switch('replica')
> +box.error.injection.set('ERRINJ_TXN_COMMIT_ASYNC', true)
> +
> +test_run:switch('default')
> +_ = s:replace{1}
> +
> +test_run:switch('replica')
> +test_run:wait_upstream(1, {status = 'stopped'})
> +-- Should be something about error injection.
> +box.info.replication[1].upstream.message
> +
> +test_run:switch('default')
> +test_run:cmd('stop server replica')
> +test_run:cmd('delete server replica')
> +box.error.injection.set('ERRINJ_TXN_COMMIT_ASYNC', false)
> +s:drop()
> +box.schema.user.revoke('guest', 'super')
> diff --git a/test/replication/suite.cfg b/test/replication/suite.cfg
> index 9eccf9e19..3a0a8649f 100644
> --- a/test/replication/suite.cfg
> +++ b/test/replication/suite.cfg
> @@ -45,6 +45,7 @@
> "gh-5536-wal-limit.test.lua": {},
> "gh-5566-final-join-synchro.test.lua": {},
> "gh-5613-bootstrap-prefer-booted.test.lua": {},
> + "gh-6027-applier-error-show.test.lua": {},
> "gh-6032-promote-wal-write.test.lua": {},
> "gh-6057-qsync-confirm-async-no-wal.test.lua": {},
> "gh-6094-rs-uuid-mismatch.test.lua": {},
> diff --git a/test/replication/suite.ini b/test/replication/suite.ini
> index 80e968d56..ccf3559df 100644
> --- a/test/replication/suite.ini
> +++ b/test/replication/suite.ini
> @@ -3,7 +3,7 @@ core = tarantool
> script = master.lua
> description = tarantool/box, replication
> disabled = consistent.test.lua
> -release_disabled = catch.test.lua errinj.test.lua gc.test.lua gc_no_space.test.lua before_replace.test.lua qsync_advanced.test.lua qsync_errinj.test.lua quorum.test.lua recover_missing_xlog.test.lua sync.test.lua long_row_timeout.test.lua gh-4739-vclock-assert.test.lua gh-4730-applier-rollback.test.lua gh-5140-qsync-casc-rollback.test.lua gh-5144-qsync-dup-confirm.test.lua gh-5167-qsync-rollback-snap.test.lua gh-5506-election-on-off.test.lua gh-5536-wal-limit.test.lua hang_on_synchro_fail.test.lua anon_register_gap.test.lua gh-5213-qsync-applier-order.test.lua gh-5213-qsync-applier-order-3.test.lua gh-6032-promote-wal-write.test.lua gh-6057-qsync-confirm-async-no-wal.test.lua
> +release_disabled = catch.test.lua errinj.test.lua gc.test.lua gc_no_space.test.lua before_replace.test.lua qsync_advanced.test.lua qsync_errinj.test.lua quorum.test.lua recover_missing_xlog.test.lua sync.test.lua long_row_timeout.test.lua gh-4739-vclock-assert.test.lua gh-4730-applier-rollback.test.lua gh-5140-qsync-casc-rollback.test.lua gh-5144-qsync-dup-confirm.test.lua gh-5167-qsync-rollback-snap.test.lua gh-5506-election-on-off.test.lua gh-5536-wal-limit.test.lua hang_on_synchro_fail.test.lua anon_register_gap.test.lua gh-5213-qsync-applier-order.test.lua gh-5213-qsync-applier-order-3.test.lua gh-6027-applier-error-show.test.lua gh-6032-promote-wal-write.test.lua gh-6057-qsync-confirm-async-no-wal.test.lua
> config = suite.cfg
> lua_libs = lua/fast_replica.lua lua/rlimit.lua
> use_unix_sockets = True
--
Serge Petrenko
More information about the Tarantool-patches
mailing list