| Commit message (Collapse) | Author | Age | Files | Lines |
| |
|
| |
|
| |
|
| |
|
| |
|
|
|
|
|
| |
Reset query cache after every test case and add wait after
load infile.
|
|
|
|
| |
Add wsrep_sync_wait as we want INSERT to fail.
|
|
|
|
|
|
| |
'Unknown database'
Wait in second node until tables with databases are created.
|
|
|
|
| |
Test changes only.
|
| |
|
| |
|
|
|
|
|
| |
Failure is due missing .rdiff files for debug build as tests have
@have_debug combination.
|
|
|
|
|
|
|
| |
with "Transport endpoint is not connected" or with a failure to start a node
Add correct suppressions and wait conditions for expected database
contents.
|
| |
|
|
|
|
|
|
| |
buildbot with "Last Applied Action message in non-primary configuration from member 0"
Add supression.
|
|
|
|
|
|
|
|
|
|
| |
and debug builds. Modified version for 10.1 from following commit:
commit 8e68876477eaec7944baa0b63ef26e551693c4f8
Author: Oleksandr Byelkin <sanja@mariadb.com>
Date: Thu Sep 13 15:06:44 2018 +0200
Fix of the test which has debug version
|
|
|
|
|
| |
Move MW-328[A,B,C] to big tests as there seem to be big variation
on their execution times and sometimes that could lead timeout.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The problem occurs in 10.2 and earlier releases of MariaDB Server because the
Partition Engine was not pushing the engine conditions to the underlying
storage engine of each partition. This caused Spider to return the first 5
rows in the table with the data provided by the customer. 2 of the 5 rows
did not qualify the WHERE clause, so they were removed from the result set by
the server.
To fix the problem, I have back-ported support for engine condition pushdown
in the Partition Engine from MariaDB Server 10.3.
Author:
Jacob Mathew.
Reviewer:
Kentoku Shiba.
Cherry-Picked:
Commit eb2ca3d on branch bb-10.2-MDEV-16912
|
| |
|
|\
| |
| | |
MDEV-17173: correct parsing of 12.13.14.15:4444/xtrabackup_sst leavin…
|
|/
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
LSN/SST_VER blank
Correcting commit e78e308e818
$ . scripts/wsrep_sst_common.sh --address 12.13.14.15:4444/xtrabackup_sst ; set | grep WSREP_SST
WSREP_SST_OPT_ADDR=12.13.14.15:4444/xtrabackup_sst
WSREP_SST_OPT_ADDR_PORT=4444
WSREP_SST_OPT_AUTH=
WSREP_SST_OPT_BINLOG=
WSREP_SST_OPT_BINLOG_INDEX=
WSREP_SST_OPT_BYPASS=0
WSREP_SST_OPT_CONF=' '
WSREP_SST_OPT_DATA=
WSREP_SST_OPT_DEFAULT=
WSREP_SST_OPT_EXTRA_DEFAULT=
WSREP_SST_OPT_HOST=12.13.14.15
WSREP_SST_OPT_HOST_UNESCAPED=12.13.14.15
WSREP_SST_OPT_LSN=
WSREP_SST_OPT_MODULE=xtrabackup_sst
WSREP_SST_OPT_PATH=xtrabackup_sst
WSREP_SST_OPT_PORT=4444
WSREP_SST_OPT_PSWD=
WSREP_SST_OPT_SST_VER=
WSREP_SST_OPT_SUFFIX_DEFAULT=
WSREP_SST_OPT_SUFFIX_VALUE=
WSREP_SST_OPT_USER=
. scripts/wsrep_sst_common.sh --address 12.13.14.15:4444/xtrabackup_sst/1234/5676 ; set | grep WSREP_SST
WSREP_SST_OPT_ADDR=12.13.14.15:4444/xtrabackup_sst/1234/5676
WSREP_SST_OPT_ADDR_PORT=4444
WSREP_SST_OPT_AUTH=
WSREP_SST_OPT_BINLOG=
WSREP_SST_OPT_BINLOG_INDEX=
WSREP_SST_OPT_BYPASS=0
WSREP_SST_OPT_CONF=' '
WSREP_SST_OPT_DATA=
WSREP_SST_OPT_DEFAULT=
WSREP_SST_OPT_EXTRA_DEFAULT=
WSREP_SST_OPT_HOST=12.13.14.15
WSREP_SST_OPT_HOST_UNESCAPED=12.13.14.15
WSREP_SST_OPT_LSN=1234
WSREP_SST_OPT_MODULE=xtrabackup_sst
WSREP_SST_OPT_PATH=xtrabackup_sst/1234/5676
WSREP_SST_OPT_PORT=4444
WSREP_SST_OPT_PSWD=
WSREP_SST_OPT_SST_VER=5676
WSREP_SST_OPT_SUFFIX_DEFAULT=
WSREP_SST_OPT_SUFFIX_VALUE=
WSREP_SST_OPT_USER=
|
|
|
|
| |
Test requires galera library version 25.3.24.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
of range
The bug was this scenario:
1. Join optimizer picks a range plan on index IDX1
(This index doesn't match the ORDER BY clause, so sorting will be needed)
2. Index Condition Pushdown pushes a part of WHERE down. The pushed
condition is removed from SQL_SELECT::cond
3. test_if_skip_sort_order() figures that it's better to use IDX2
(as it will match ORDER BY ... LIMIT and so will execute faster)
3.1 It sees that there was a possible range access on IDX2. It tries to
construct it by calling SQL_SELECT::test_quick_select(), but alas,
SQL_SELECT::cond doesn't have all parts of WHERE anymore.
So it uses full index scan which is slow.
(The execution works fine because there's code further in test_if_skip_sort_order()
which "Unpushes" the index condition and restores the original WHERE clause.
It was just the test_quick_select call that suffered).
|
| |
|
|
|
|
| |
Add suppression.
|
|
|
|
|
|
|
| |
should have failed with errno 1213
Replace sleep with debug sync point before insert commit to
make sure insert is not executed before truncate has started.
|
|
|
|
| |
While executing CTAS galera applier thread can cause CTAS to abort and rollback. Rollback can take time causing applier thread to shutdown node after serial unsuccessful retries to apply transaction. Don't set lock_wait_timeout to zero to wait for lock.
|
|
|
|
|
|
|
|
| |
On Unix, it is compiled-in datadir value.
On Windows, the directory is ..\data, relative to directory
mariabackup.exe
server uses the same logic to determine datadir.
|
|
|
|
|
| |
Add missing test case with proper wait conditions for expected
node contents.
|
|
|
|
| |
Add wait_conditions to wait expected node contents.
|
|
|
|
|
| |
Start general log OFF and then truncate mysql.general_log and
use proper wait timeouts to make sure it is really empty.
|
| |
|
| |
|
|\ |
|
| | |
|
| | |
|
| |\ |
|
| | | |
|
| | | |
|
| | | |
|
| | | |
|
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
If --innodb-undo-tablespaces is used, then InnoDB stores undo in a
separate file(s) which whould also be replicated.
This fixes
Issue#337 This filter will cause sst failed at applying undo...
https://github.com/codership/mysql-wsrep/issues/337
|
| | |
| | |
| | |
| | |
| | | |
This way it is more readable and easy to change, also if a new entry is
added or one removed, the diff will be easier to read.
|
| | |
| | |
| | |
| | | |
MTR test for galera#505
|
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
Check that cluster weight have proper values in galera_pc_weight
test. Removed sleeps from tests and added condition waits for
expected cluster sizes. Replaced galera suspend/resume with
gmcast.isolate in order to avoid breaking client connections
to server which is isolated from the cluster and to avoid
the need to reset wsrep_cluster_address.
Re-recorded galera_defaults.
|
| | | |
|
| | |\ |
|
| | | |
| | | |
| | | |
| | | | |
Fix misplaced `DBUG_RETURN` in `Alter_table_statement::execute`.
|
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
Adding a FK constraint to an existing table (ALTER TABLE ADD FOREIGN
KEY) causes the applier to fail, if a concurrent DML statement that
violate the new constraint (i.e. a DELETE or UPDATE of record in the
parent table).
For exmaple, the following scenario causes a crash in the applier:
1. ALTER successfully adds FK constraint in node_1
2. On node_2 is UPDATE is in pre_commit() and has certified successfully
3. ALTER is delivered in node_2 and BF aborts DML
4. Applying UPDATE event causes FK violation in node_1
To avoid this situation it is necessary for UPDATE to fail during
certification. And for the UPDATE to fail certfication it is necessary
that ALTER appends certification keys for both the child and the parent
table. Before this patch, ALTER TABLE ADD FK only appended keys for
child table which is ALTERed.
|
| | | |\ |
|