Home - Waterfall Grid T-Grid Console Builders Recent Builds Buildslaves Changesources - JSON API - About

Console View


Categories: connectors experimental galera main
Legend:   Passed Failed Warnings Failed Again Running Exception Offline No data

connectors experimental galera main
Marko Mäkelä
MDEV-20612: Replace lock_sys.mutex with lock_sys.latch

For now, we will acquire the lock_sys.latch only in exclusive mode,
that is, use it as a mutex. This is preparation for the next commit
where we will introduce srw_lock shards for page_id_t, table_id_t and
actually acquire lock_sys.latch.rd_lock() for individual shards.
Andrei Elkin
MDEV-21117 incremental commit fixes build with_embedded_server.
Andrei Elkin
MDEV-21117 Hot fixes, 1.

id_binlog was not computed correctly in case the last log is the binlog CP file one.
g1,G2,g3 -sequcence led to the 2nd round in which g3->G3 to commit (wrong!).

Fixed to preserve the UINT_MAX number for the only binlog in action.
Varun Gupta
taking selectivity for non-indexed equi-join predicates

Made minor changes to the function table_multi_eq_cond_selectivity.

First the function was checking if there is a keyuse only then we were
trying to use the function to get estimates for the equi-join predicated.
I think this can calculate the estimates even if we don't have a keyuse.

Second change is done when we try to get the avg frequency of a column,
then we need to make sure that statistics are available in the statistical
tables. This check was missing and so get_column_avg_frequency() was
returning zero for columns.

Also fixed a typo.
Oleksandr Byelkin
Fix ASAN issues in the tests
Andrei Elkin
MDEV-21117 Incremental commit fixes failover

The demoted into slave old master refused to accept own events by
existing rules. --replicate-same-server-id does not look to be a solution.

This is fixed with relaxing the same-server-id acceptence condition
for the semisync slave that connects to master in gtid mode.
This is safe to do when @@gtid_strict_mode = ON, which is recommended
therefore.
Rucha Deodhar
MDEV-20939: Race condition between mysqldump import and InnoDB persistent
statistics calculation

Analysis: When --replace or --insert-ignore is not given, dumping of
mysql.innodb_index_stats and mysql.innodb_table_stats will result into race
condition.
Fix: Check if these options are present with --system=stats (because dumping
under --system=stats is safe). Otherwise, dump only structure, ignoring data
because innodb will recalculate data anyway.
Marko Mäkelä
Cleanup: Remove lock_get_size()
Marko Mäkelä
WIP: Prepare to add lock_sys shards

dict_table_t::autoinc_trx: Use Atomic_relaxed
Sujatha
MDEV-21117: refine the server binlog-based recovery for semisync

Problem:
=======
When the semisync master is crashed and restarted as slave it could
recover transactions that former slaves may never have seen.
A known method existed to clear out all prepared transactions
with --tc-heuristic-recover=rollback does not care to adjust
binlog accordingly.

Fix:
===
The binlog-based recovery is made to concern of the slave semisync role of
post-crash restarted server.
No changes in behaviour is done to the "normal" binloggging server
and the semisync master.
When the restarted server is configured with
  --rpl-semi-sync-slave-enabled=1
the refined recovery attempts to rolled back prepared transactions
and truncate binlog accordingly.
In case of a partically committed (that is committed at least
in one of the engines) such transaction gets committed.
It's guaranteed no (partially as well) committed transactions
exist beyond the truncate position.
In case there exists a non-transactional replication event past the
computed truncate position the recovery ends with an error.
Marko Mäkelä
MDEV-20612 preparation: LockMutexGuard, LockGuard

Let us use the RAII wrapper LockMutexGuard for most operations where
lock_sys.mutex is acquired.

Let us use the RAII wrapper LockGuard for operations that are related
to a single buffer pool page.
Andrei Elkin