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.
Now all references to ctdb->recovery_lock are encapsulated in the
cluster lock code.
Signed-off-by: Martin Schwenke <martin@meltin.net>
Reviewed-by: Amitay Isaacs <amitay@gmail.com>
It is no longer just a recovery lock but is always held by the cluster
leader.
Signed-off-by: Martin Schwenke <martin@meltin.net>
Reviewed-by: Amitay Isaacs <amitay@gmail.com>
The introduction of the leader broadcast timeout provides an
alternative to the current leader validation. Using the leader
broadcast may not be as fast but it is more correct.
When the leader node is stopped or banned, the only way of triggering
an election is currently to fetch the leader's node map to check
whether the it is still active. This is because the leader will no
longer push the node map to other nodes. However, having all nodes
fetch the node map from an inactive leader may be unreliable.
Most of the other cases are also handled more reliably by the leader
broadcast timeout.
Signed-off-by: Martin Schwenke <martin@meltin.net>
Reviewed-by: Amitay Isaacs <amitay@gmail.com>
This no longer occurs at startup due to the leader broadcast timeout.
Signed-off-by: Martin Schwenke <martin@meltin.net>
Reviewed-by: Amitay Isaacs <amitay@gmail.com>
If no leader broadcasts have been received from the leader for more
than 5s then trigger an election.
Apart from being sane behaviour, this avoids elected-before-connected
bugs at startup, where a node elects itself leader before it is
connected to other nodes.
When a node processes a leader broadcast timeout it sends an unknown
leader broadcast to all nodes. That causes cancellation of the leader
broadcast timeout across the cluster. This is particular important at
startup, since nodes may be started in a staggered fashion. Without
this cluster-wide cancellation, a node might notice the lack of
leader, win an election and complete a recovery before other nodes
notice the lack of leader. When the leader broadcast timeout finally
occurs on the other nodes then they'll put the cluster back into an
unnecessary recovery.
Signed-off-by: Martin Schwenke <martin@meltin.net>
Reviewed-by: Amitay Isaacs <amitay@gmail.com>
These are triggered on 1 second timer, but are only sent if the node
is the current leader and there is no election underway.
If this node can not be the leader then ensure it releases the
recovery lock.
Signed-off-by: Martin Schwenke <martin@meltin.net>
Reviewed-by: Amitay Isaacs <amitay@gmail.com>
An alternate election method will be added that doesn't use the
election timeout, so this provides a common way for recognising when
an election is in progress.
Signed-off-by: Martin Schwenke <martin@meltin.net>
Reviewed-by: Amitay Isaacs <amitay@gmail.com>
This makes the code self-documenting.
In ctdb_election_data() there is a slight behaviour change. An
inactive node will now try to lose an election. This case should not happen
because:
* An inactive node can't win an election round and then send a reply.
* Any inactive node should never start an election. There are
currently places where this happens and they will be fixed later.
There is an instance where this could be used in
validate_recovery_master() but this involves a more serious logic
change. Overhaul this function later.
Signed-off-by: Martin Schwenke <martin@meltin.net>
Reviewed-by: Amitay Isaacs <amitay@gmail.com>
There are some remaining instances in this file but they will be
removed in subsequent commits.
Modernise debug macros as appropriate.
Signed-off-by: Martin Schwenke <martin@meltin.net>
Reviewed-by: Amitay Isaacs <amitay@gmail.com>
Recovery master is being renamed to leader. This follows clustering
best practice (e.g. RAFT).
Signed-off-by: Martin Schwenke <martin@meltin.net>
Reviewed-by: Amitay Isaacs <amitay@gmail.com>
This is currently referenced in a number of inconsistent
ways, including:
* pnn
* rec->ctdb->pnn
* ctdb->pnn
* ctdb_get_pnn(ctdb)
* ctdb_get_pnn(rec->ctdb)
The first of these always requires some thought about the context - is
this the node PNN or some other PNN (e.g. argument to function)?
rec->pnn is now always used when referring to the recovery daemon's
PNN.
Doing this also reduces reliance on struct ctdb_context internals.
Signed-off-by: Martin Schwenke <martin@meltin.net>
Reviewed-by: Amitay Isaacs <amitay@gmail.com>
ban_time argument is always ctdb->tunable.recovery_ban_period, so
build this in and make the calling code more readable.
ctdb_ban_node() already logs how long a node is banned for, so don't
repeatedly log this.
Signed-off-by: Martin Schwenke <martin@meltin.net>
Reviewed-by: Amitay Isaacs <amitay@gmail.com>
All other arguments are available via rec, so simplify.
Signed-off-by: Martin Schwenke <martin@meltin.net>
Reviewed-by: Amitay Isaacs <amitay@gmail.com>
pnn and nodemap are both available via the rec context, so simplify.
vnnmap is unused.
Signed-off-by: Martin Schwenke <martin@meltin.net>
Reviewed-by: Amitay Isaacs <amitay@gmail.com>
The pnn and nodemap arguments to force_election() and
send_election_request() are always effectively rec->pnn and
rec->nodemap, so simplify.
Signed-off-by: Martin Schwenke <martin@meltin.net>
Reviewed-by: Amitay Isaacs <amitay@gmail.com>
This is currently referenced in a number of inconsistent
ways, including:
* pnn
* rec->ctdb->pnn
* ctdb->pnn
* ctdb_get_pnn(ctdb)
* ctdb_get_pnn(rec->ctdb)
The first of these always requires some thought about the context - is
this the node PNN or some other PNN (e.g. argument to function)?
The intention is to always use rec->pnn when referring to the recovery
daemon's PNN.
Doing this also reduces reliance on struct ctdb_context internals.
Signed-off-by: Martin Schwenke <martin@meltin.net>
Reviewed-by: Amitay Isaacs <amitay@gmail.com>
Make the code self-documenting.
This preempts an upcoming change to terminology but doing it now saves
a lot of churn.
Signed-off-by: Martin Schwenke <martin@meltin.net>
Reviewed-by: Amitay Isaacs <amitay@gmail.com>
The recovery and takeover helpers can run for a while and generate
non-trivial logs, so have them reopen their logs to support log
rotation.
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 Jan 17 04:36:30 UTC 2022 on sn-devel-184
Recovery and takeover helpers can run for a while and generate
non-trivial logs. They should support log reopening.
Signed-off-by: Martin Schwenke <martin@meltin.net>
Reviewed-by: Amitay Isaacs <amitay@gmail.com>
Pass on a SIGHUP to the recovery daemon, which will then reopen its
logs.
Signed-off-by: Martin Schwenke <martin@meltin.net>
Reviewed-by: Amitay Isaacs <amitay@gmail.com>
Remote nodes are already initialised as UNHEALTHY when the node list
is initialised at startup (ctdb_load_nodes_file() calls
convert_node_map_to_list()) and when disconnected (ctdb_node_dead()).
So, drop this code.
BUG: https://bugzilla.samba.org/show_bug.cgi?id=14784
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): Thu Sep 9 02:38:34 UTC 2021 on sn-devel-184
If this node is not connected to a node then we shouldn't know
anything about it. The state will be pushed later by the recovery
master.
BUG: https://bugzilla.samba.org/show_bug.cgi?id=14784
Signed-off-by: Martin Schwenke <martin@meltin.net>
Signed-off-by: Amitay Isaacs <amitay@gmail.com>
Now that there are separate disable/enable controls used by the ctdb
tool this control can ignore any flag updates for the current nodes.
These only come from the recovery master, which depends on being able
to fetch flags for all nodes.
BUG: https://bugzilla.samba.org/show_bug.cgi?id=14784
Signed-off-by: Martin Schwenke <martin@meltin.net>
Reviewed-by: Amitay Isaacs <amitay@gmail.com>
CTDB_SRVID_SET_NODE_FLAGS is no longer sent so drop monitor_handler()
and replace with srvid_not_implemented(). Mark the SRVID obsolete in
its comment.
BUG: https://bugzilla.samba.org/show_bug.cgi?id=14784
Signed-off-by: Martin Schwenke <martin@meltin.net>
Reviewed-by: Amitay Isaacs <amitay@gmail.com>
The code that handles this message is
ctdb_recoverd.c:monitor_handler(). Although it appears to do
something potentially useful, it only logs the flags changes. All
changes made are to local structures - there are no actual
side-effects.
It used to trigger a takeover run when the DISABLED flag changed.
This was dropped back in commit
662f06de9f.
BUG: https://bugzilla.samba.org/show_bug.cgi?id=14784
Signed-off-by: Martin Schwenke <martin@meltin.net>
Reviewed-by: Amitay Isaacs <amitay@gmail.com>
When flags change, promote the message to NOTICE level and switch the
message to the style that is currently generated by
ctdb-recoverd.c:monitor_handler(). This will allow monitor_handler()
to go away in future.
Drop logging when flags do not change. The recovery master now logs
when it pushes flags for a node, so the lack of a corresponding
"changed flags" message here indicates that no update was required.
BUG: https://bugzilla.samba.org/show_bug.cgi?id=14784
Signed-off-by: Martin Schwenke <martin@meltin.net>
Reviewed-by: Amitay Isaacs <amitay@gmail.com>
Don't trust the old flags from the recovery master.
Surrounding code will change in future comments, including the use of
old-style debug macros, so just make this change clear.
BUG: https://bugzilla.samba.org/show_bug.cgi?id=14784
Signed-off-by: Martin Schwenke <martin@meltin.net>
Reviewed-by: Amitay Isaacs <amitay@gmail.com>
DISABLED is UNHEALTHY | PERMANENTLY_DISABLED, which is not what is
intended here. Luckily, it doesn't do any harm because nodes are
marked unhealthy at startup anyway.
BUG: https://bugzilla.samba.org/show_bug.cgi?id=14784
Signed-off-by: Martin Schwenke <martin@meltin.net>
Reviewed-by: Amitay Isaacs <amitay@gmail.com>
This will usually happen if flags on the node in question change, so
keeping the code simple and pushing to all nodes won't hurt. When all
nodes come up there might be differences in connected nodes, causing
such "fix ups". Receiving nodes will ignore no-op pushes.
BUG: https://bugzilla.samba.org/show_bug.cgi?id=14784
Signed-off-by: Martin Schwenke <martin@meltin.net>
Reviewed-by: Amitay Isaacs <amitay@gmail.com>
The resulting code structure looks a little weird. However, there is
another condition that requires the flags to be pushed that will be
inserted before the continue statement in a subsequent commit..
BUG: https://bugzilla.samba.org/show_bug.cgi?id=14784
Signed-off-by: Martin Schwenke <martin@meltin.net>
Reviewed-by: Amitay Isaacs <amitay@gmail.com>
The socket is set close-on-exec but that doesn't help for processes
that do not exec(). This should be done for all child processes.
This has been seen in testing where "ctdb shutdown" waits for the
socket to close before succeeding. It appears that lingering
vacuuming processes have not closed the socket when becoming clients
so they cause "ctdb shutdown" to hang even though the main daemon
process has exited. The cause of the lingering vacuuming processes
has been previously examined but still isn't understood.
Signed-off-by: Martin Schwenke <martin@meltin.net>
Reviewed-by: Amitay Isaacs <amitay@gmail.com>
1. PID of lock helper waiting for lock
2. Scope of lock: "record" or "db"
3. Path to database that lock helper is trying to lock
4. Whether the database uses mutexes: "mutex" or "fcntl"
Signed-off-by: Amitay Isaacs <amitay@gmail.com>
Reviewed-by: Martin Schwenke <martin@meltin.net>
