IF YOU WOULD LIKE TO GET AN ACCOUNT, please write an
email to Administrator. User accounts are meant only to access repo
and report issues and/or generate pull requests.
This is a purpose-specific Git hosting for
BaseALT
projects. Thank you for your understanding!
Только зарегистрированные пользователи имеют доступ к сервису!
Для получения аккаунта, обратитесь к администратору.
If the cluster filesystem is unavailable then I/O errors may occur.
This is no worse than contention, so don't ban. This avoids having
services unavailable for longer than necessary.
Update the associated test to simply confirm that this results in a
leaderless cluster, and leadership is restored when the lock can once
again be taken.
Signed-off-by: Martin Schwenke <martin@meltin.net>
Reviewed-by: Amitay Isaacs <amitay@gmail.com>
A stalled node probably continues to hold the cluster lock, so confirm
elections work in this case.
BUG: https://bugzilla.samba.org/show_bug.cgi?id=14958
Signed-off-by: Martin Schwenke <martin@meltin.net>
Reviewed-by: Amitay Isaacs <amitay@gmail.com>
Autobuild-User(master): Amitay Isaacs <amitay@samba.org>
Autobuild-Date(master): Mon Feb 14 02:46:01 UTC 2022 on sn-devel-184
Rename test, clean up node selection. Duplicate for for banning and
removing leader capability cases. Repeat all 3 tests without cluster
lock.
All of the standard election triggers are now tested, with and without
cluster lock. Due to test cluster configuration limitations, the
tests without cluster lock are skipped on a real cluster.
Signed-off-by: Martin Schwenke <martin@meltin.net>
Reviewed-by: Amitay Isaacs <amitay@gmail.com>
If the cluster is partitioned then nodes in one partition can not take
the lock anyway, so election is pointless. It just introduces
unnecessary corner cases.
Instead just race for the lock.
When a node notices a lack of leader and notifies other nodes of an
election via an unknown leader broadcast, the cluster lock election is
hooked into this broadcast.
The test needs to be updated because losing the cluster lock can now
result in a leadership change.
Signed-off-by: Martin Schwenke <martin@meltin.net>
Reviewed-by: Amitay Isaacs <amitay@gmail.com>
The following command names are changed:
recmaster -> leader
setrecmasterrole -> setleaderrole
Command output changed for the following commands:
status
getcapabilities
Documentation and tests are updated to reflect these changes.
Signed-off-by: Martin Schwenke <martin@meltin.net>
Reviewed-by: Amitay Isaacs <amitay@gmail.com>
ctdb_test_init() doesn't actually pass arguments to local_daemons.sh.
This needs to be done using ctdb_nodes_start_custom().
Signed-off-by: Martin Schwenke <martin@meltin.net>
Reviewed-by: Amitay Isaacs <amitay@gmail.com>
This test has been failing with:
Wait until record is migrated to lmaster node 0
<30|BAD: node 0 is not dmaster
dmaster: 1
rsn: 8
flags: 0x00010000 MIGRATED_WITH_DATA
data(6) = "value1"
*** TEST COMPLETED (RC=1) AT 2021-02-02 06:18:48, CLEANING UP...
This should never happen. If this really fails then the wait should
time out.
The problem is that wait_until() does:
"$@" || _rc=$?
and vacuum_test_key_dmaster() currently calls ctdb_test_fail() on
failure, which causes the shell to exit. Instead, pass a variant to
wait_until() that simply returns the correct status instead of
exiting.
An alternative would be to change the statement in wait_until() to do:
("$@") || _rc=$?
so it captures the exit. However, this is a global change and
requires more thought.
Signed-off-by: Martin Schwenke <martin@meltin.net>
Reviewed-by: Jeremy Allison <jra@samba.org>
Check that the desired state is set on all nodes instead of just the
test node. This ensures that node flags have correctly propagated
across the cluster.
BUG: https://bugzilla.samba.org/show_bug.cgi?id=14513
Signed-off-by: Martin Schwenke <martin@meltin.net>
Reviewed-by: Amitay Isaacs <amitay@gmail.com>
Autobuild-User(master): Amitay Isaacs <amitay@samba.org>
Autobuild-Date(master): Tue Oct 6 04:32:06 UTC 2020 on sn-devel-184
This isn't used anywhere and can easily be checked via "ctdb pnn" and
"ctdb recmaster" commands.
Signed-off-by: Martin Schwenke <martin@meltin.net>
Reviewed-by: Amitay Isaacs <amitay@gmail.com>
Avoid use of non-portable md5sum by constructing database names using
index. Improve indentation, use more modern commands, code
improvements (shellcheck).
Signed-off-by: Martin Schwenke <martin@meltin.net>
Reviewed-by: Amitay Isaacs <amitay@gmail.com>
Autobuild-User(master): Martin Schwenke <martins@samba.org>
Autobuild-Date(master): Wed Jul 22 09:14:35 UTC 2020 on sn-devel-184
Simplify code, use more modern commands, code improvements (shellcheck).
Signed-off-by: Martin Schwenke <martin@meltin.net>
Reviewed-by: Amitay Isaacs <amitay@gmail.com>
"wc -l" on some platforms (e.g. FreeBSD) contains leading spaces, so
strip them.
Signed-off-by: Martin Schwenke <martin@meltin.net>
Reviewed-by: Amitay Isaacs <amitay@gmail.com>
Select test node with IPs instead of using a fixed node. Remove
unnecessary code, use more modern commands, code
improvements (shellcheck).
Signed-off-by: Martin Schwenke <martin@meltin.net>
Reviewed-by: Amitay Isaacs <amitay@gmail.com>
"wc -l" on some platforms (e.g. FreeBSD) contains leading spaces and
stops "$num from being a number. Create a more portable solution and
put it in a function instead of repeating the logic.
Signed-off-by: Martin Schwenke <martin@meltin.net>
Reviewed-by: Amitay Isaacs <amitay@gmail.com>
It would be nice to get rid of "onnode any". There's no use making
tests nondeterministic. If covering different cases matters then they
should be explicitly handled.
In most places "any" is replaced by "$test_node". In some cases,
where $test_node is not set, a fixed node that is already used
elsewhere can be reused.
Signed-off-by: Martin Schwenke <martin@meltin.net>
Reviewed-by: Amitay Isaacs <amitay@gmail.com>
Separate cluster startup from test initialisation for tests that start
the cluster with customised configuration. In these cases the result
of the cluster startup is actually the point of the test.
Additionally, pubips.013.failover_noop.sh claims to have completed
test initialisation twice, which just seems wrong.
The result is:
* ctdb_test_init() takes one option (-n) to indicate when it should
not configure/start the cluster
* New function ctdb_nodes_start_custom() accepts options for special
cluster configuration, only operates on local daemons and triggers a
test failure rather than a test error on failure.
Signed-off-by: Martin Schwenke <martin@meltin.net>
Reviewed-by: Amitay Isaacs <amitay@gmail.com>
* Use "#!/usr/bin/env bash" for improved portability
* Drop test_info() definition and replace it with a comment
The use of test_info() is pointless.
* Drop call to cluster_is_healthy()
This is a holdover from when the previous test would restart daemons
to get things ready for a test. There was also a bug where going
into recovery during the restart would sometimes cause the cluster
to become unhealthy. If we really need something like this then we
can add it to ctdb_test_init().
* Make order of preamble consistent
Signed-off-by: Martin Schwenke <martin@meltin.net>
Reviewed-by: Amitay Isaacs <amitay@gmail.com>
The output in a test failure appears to contain no pstree output
because "00\.test\.script,.*" does not match. However, this is just a
guess because the output is not shown.
Showing the output makes it easier to understand test failures.
Signed-off-by: Martin Schwenke <martin@meltin.net>
Reviewed-by: Amitay Isaacs <amitay@gmail.com>
Avoid use of non-portable md5sum by constructing database names using
index. Improve indentation, use more modern commands, code
improvements (shellcheck).
Signed-off-by: Martin Schwenke <martin@meltin.net>
Select test node with IPs instead of using a fixed node. Remove
unnecessary code, use more modern commands, code
improvements (shellcheck).
Signed-off-by: Martin Schwenke <martin@meltin.net>
"wc -l" on some platforms (e.g. FreeBSD) contains leading spaces and
stops "$num from being a number. Create a more portable solution and
put it in a function instead of repeating the logic.
Signed-off-by: Martin Schwenke <martin@meltin.net>
It would be nice to get rid of "onnode any". There's no use making
tests nondeterministic. If covering different cases matters then they
should be explicitly handled.
In most places "any" is replaced by "$test_node". In some cases,
where $test_node is not set, a fixed node that is already used
elsewhere can be reused.
Signed-off-by: Martin Schwenke <martin@meltin.net>
Separate cluster startup from test initialisation for tests that start
the cluster with customised configuration. In these cases the result
of the cluster startup is actually the point of the test.
Additionally, pubips.013.failover_noop.sh claims to have completed
test initialisation twice, which just seems wrong.
The result is:
* ctdb_test_init() takes one option (-n) to indicate when it should
not configure/start the cluster
* New function ctdb_nodes_start_custom() accepts options for special
cluster configuration, only operates on local daemons and triggers a
test failure rather than a test error on failure.
Signed-off-by: Martin Schwenke <martin@meltin.net>
* Use "#!/usr/bin/env bash" for improved portability
* Drop test_info() definition and replace it with a comment
The use of test_info() is pointless.
* Drop call to cluster_is_healthy()
This is a holdover from when the previous test would restart daemons
to get things ready for a test. There was also a bug where going
into recovery during the restart would sometimes cause the cluster
to become unhealthy. If we really need something like this then we
can add it to ctdb_test_init().
* Make order of preamble consistent
Signed-off-by: Martin Schwenke <martin@meltin.net>
The output in a test failure appears to contain no pstree output
because "00\.test\.script,.*" does not match. However, this is just a
guess because the output is not shown.
Showing the output makes it easier to understand test failures.
Signed-off-by: Martin Schwenke <martin@meltin.net>
When "ctdb shutdown" is run with -n <N> it does not wait for the node
<N>'s ctdbd to go down but exits immediately. This means that the
local_daemons.sh shutdown command can find the PID file still present
and then attempt the shutdown, but the daemon can have exited between
the check and the shutdown. Although the test waits until the node is
disconnected, the transport is taken down just before the exit, so
this does not guarantee the daemon has exited.
A local shutdown command (no -n <N>) waits until the socket
disconnects and this happens *after* the PID file is gone, so this is
safe to use with the local_daemons.sh shutdown command.
Signed-off-by: Martin Schwenke <martin@meltin.net>
Reviewed-by: Amitay Isaacs <amitay@gmail.com>
Autobuild-User(master): Amitay Isaacs <amitay@samba.org>
Autobuild-Date(master): Mon Mar 2 10:39:28 UTC 2020 on sn-devel-184
Separate functions are not needed for stopping/starting/restarting
individual nodes. The stop and start functions essentially just use
onnode, though for local daemons this is embedded in local_daemons.sh.
So, just provide one stop and one start function that takes an
optional nodespec, defaulting to all nodes.
Restarting becomes common.
Signed-off-by: Martin Schwenke <martin@meltin.net>
Reviewed-by: Amitay Isaacs <amitay@gmail.com>
* Typo in variable name db_map_pattern
* Variable num_db_init used before set
* dbmap_pattern does not cover database flags
Signed-off-by: Martin Schwenke <martin@meltin.net>
Reviewed-by: Amitay Isaacs <amitay@gmail.com>
Autobuild-User(master): Amitay Isaacs <amitay@samba.org>
Autobuild-Date(master): Wed Feb 12 04:38:47 UTC 2020 on sn-devel-184
select_test_node_and_ips() is not required in these cases.
Signed-off-by: Martin Schwenke <martin@meltin.net>
Reviewed-by: Amitay Isaacs <amitay@gmail.com>
