'build' and 'run_test' copies public ssh keys to all created VMs
Legacy code takes keys from different locations, e.g. ~/build-scripts/team_keys which
causes errors due to lack of such files.
Now public keys goes from file defined on or if not defined
from .ssh/id_rsa.pub of current host machine
mdbci call can heppen before maxsales object creation and segfault hapens
To avoid it call of ssh for 'maxscale --version-full' is moved to the end
of Testconnections constructor
Previously Maxscale version check was in the 'check_backend', but this was
removed from the tests execution process. Now 'maxscale --version-full'
is added to nodes creation function - to call_mdbci()
If case of MDBCI_VM_PATH variable does not have trailing slash full names of _network_config
and _lables files are defined in the wrong way: MDBCI_VM_PATH is interpreted as a part of file name
instead of direcoty name
In case when nodes/servers/replication/etc in backend are broken test can not
get proper version information and exits without doing anything.
To avoid it, test first checks backend, call fix_replication() if needed
and only after that tris to check if backend version is ok for this test
maxscale-system-test changed in order to control test environment by itself.
Every test checks which machines are running, compare with list of needed machines
and start new VMs is they are missing in the running machines list.
Tests are executiong MDBCI commands, MDBCI executable should be in the PATH
vagrant_lock was created in order to prevent parallel execution of two 'Vagrant up' operations in parallel. Previously it was necessary due to Vagrant internal Chef issue. Now several bugs in Vagrant are fixed and MDBCI uses own Chef to provision nodes for builds and tests. There is no need to wait for previous Vagrant run finished, removing all waiting from all scripts.
'rm vagrant_lock is still present to remove locks created by previous versions of build and test scripts-
MXS-2236 Add own long test and possibility to run tests under Valgrind
Long test executes INSERT queries, transactions, prepared statements in parallel to create weird load on Maxscale to catch crashes and leaks
Test is not included into ctest scope. Test should be executed manually. For BuildBot (and also for run_test.sh) 'test_set' should be set 'NAME# ./long_test'
Time to run test is defined by 'long_test_time' variable (in seconds)
Possibility to run Maxscale under Valgrind is also added. To run Maxscale under Vaslgrind 'use_valgrind=yes' variable have to be defined
By repeatedly doing reads instead of one read per second, it is more
likely that MXS-2311 is reproduced. This is still not a deterministic
process but in theory it can reproduce the problem.
The test now uses maxctrl to count how many connections there are. This
helps avoid creating new users on the database and works around the slave
syncing problems.
* add possibility to run named test
* remove sysbench_dir
* remove unsupported in 1.0 sysbench options
* exculed ReadConnRouter sysbench test
* fix sysbench table name
maxscale-system-test/set_env.sh script is slow because it checks if MariaDB start command is mysql or mysqld
Now all versions can be started with 'mysql' except MySQL 5.5 which is not supported
This should help prevent network disconnections and make the test more
stable. If the connection is lost, the automatic failover is disabled and
the test will fail.
The test doesn't work when ASAN is used as it increases the memory use of
the process. With the addition of more caches in 2.3, the test is also
more likely to fail. Due to the test being quite useless with ASAN, it is
better to remove it.
If the password field in mysql.user is empty, it is possible that the
actual password is stored in the authentication_string field. Most of the
time this happens due to MDEV-16774 which causes the password to be stored
in the authentication_string field.
Also added a test case that verifies the problem and that it is fixed by
this commit.
The intention was to send the lowest backend version string automatically
to the client instead of the default handshake version. This did not work
as the service version string was used instead of the server version.
The table creation was not detected as the function used to extract the
table name did not return the fully qualified names. Even if it did return
a fully qualified name, it wouldn't have been correctly processed.
The test did not use the wait_for_monitor function to sync with the
monitor. This function speeds up the testing greatly by removing
unnecessary sleeps from it.
Also reduced the amount of data inserted into the cluster. There's no real
need to test with large amounts of data as it is only a functional test.
