[PATCH] CTDB test improvements
martin at meltin.net
Mon Nov 5 00:49:14 UTC 2018
2 mains changes:
* For integration testing ("simple" and "complex" test suites),
CTDB daemons are now started at the beginning of each test and
stopped at the end of each test
This is true when running with local daemons and against a real
cluster. A side-effect of this is obviously that after testing the
cluster will now be down. This should not cause a problem because
the tests already affect cluster availability.
Individual tests can now be run without depending on special tests
that start/stop the cluster. It is now also easier to loop a single
Daemons were previously not restarted at the end of each "simple"
test, so running that test suite now takes longer. The patch set
includes a small amount of rationalisation of tests to avoid
duplication and generally reduce the number of tests. More work is
needed... but a reasonable compromise has already been made.
* Starting and stopping of local daemons in the "simple" test suite has
been factored out into a separate script called
ctdb/tests/local_daemons.sh (installed as ctdb_local_daemons)
This script has setup/start/stop commands that are used by the test
suite, and can be used independently for hand testing/debugging. It
also has an onnode command, which sets any variables required to run
commands (e.g. ctdb status) against a cluster of local daemons for
debugging. There is also a print-socket command to print the ctdbd
socket being used by the designated nodes - this can be used by more
extensive clustered Samba testing to help configured (for example)
smbd instances against ctdbd instances.
Note that this patch set depends on the other 2 that I have just posted
(config improvements, various fixes).
Please review and maybe push...
peace & happiness,
-------------- next part --------------
A non-text attachment was scrubbed...
Size: 123647 bytes
Desc: not available
More information about the samba-technical