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.
Only look at "classic" messaging_register handlers in the main event
context loop
If we're sitting in a nested event context, only act upon the
messaging_filtered_read ones that are registered in the nested context.
Postpone everything else via an immediate to the main tevent context
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
We need to go through the event loop, which messaging_dgm_send does. We can
also use a tevent_immediate for the same purpose. Right now the main user is
messaging_ctdb: Here strace looks a bit weird when we receive a message.
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
This itself adds a lot of code, however it removes the unix_msg library.
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
Enable us to destroy a pthreadpool_tevent structure with active jobs
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
Restore EBUSY on pthreadpool_pipe_destroy.
We need to count jobs in pthreadpool_pipe so that pthreadpool can exit with
active jobs. Unfortunately this makes pthreadpool_pipe_add_job non-threadsafe.
We could add mutexes around "num_jobs", but this would mean another set of
pthread_atfork functions. As we don't use threaded pthreadpool_pipe_add_job
except in the tests, just remove the tests...
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
So far we used joinable threads. This prevented pthreadpool_destroy with
blocked threads. This patch converts pthreadpool to detached threads. Now
pthreadpool_destroy does not have to wait for the idle threads to finish, it
can immediately return. pthreadpool_destroy will tell all threads to exit, and
the last active thread will actually free(pthreadpool).
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
Just paranoia
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
Autobuild-User(master): Jeremy Allison <jra@samba.org>
Autobuild-Date(master): Wed Sep 28 03:58:22 CEST 2016 on sn-devel-144
The function should also return if it failed or not.
Signed-off-by: Andreas Schneider <asn@samba.org>
Reviewed-by: Stefan Metzmacher <metze@samba.org>
Autobuild-User(master): Stefan Metzmacher <metze@samba.org>
Autobuild-Date(master): Sun Sep 25 12:56:17 CEST 2016 on sn-devel-144
If there is no realm set we should not add it to the machine account.
Signed-off-by: Andreas Schneider <asn@samba.org>
Reviewed-by: Stefan Metzmacher <metze@samba.org>
Remove all tevent internal code.
Everything is now stock tevent.
Bug: https://bugzilla.samba.org/show_bug.cgi?id=12283
Signed-off-by: Jeremy Allison <jra@samba.org>
Reviewed-by: Volker Lendecke <vl@samba.org>
We should not free a DLIST element
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Ralph Böhme <slow@samba.org>
Autobuild-User(master): Volker Lendecke <vl@samba.org>
Autobuild-Date(master): Wed Sep 14 19:30:28 CEST 2016 on sn-devel-144
The two error callers of unix_dgram_sendq_schedule_free already run down the
sendq in the error case.
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Ralph Böhme <slow@samba.org>
Signed-off-by: Ralph Boehme <slow@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
Autobuild-User(master): Jeremy Allison <jra@samba.org>
Autobuild-Date(master): Tue Sep 13 04:11:30 CEST 2016 on sn-devel-144
Previously, we only created a send queue for a peer if the initial send
to the non-blocking non-connected socket reported EWOULDBOCK (because
the channel was full).
With this change, we now always create a send queue and use a connected,
non-blocking datagram socket from the beginning.
Initially, the socket of the send queue is set to non-blocking mode and
we attempt a direct send via sendmsg(). If that returns EWOULDBOCK, we
set the send queue to blocking mode and let the threadpool handle the
IO.
When a send queue becomes empty, we set the send queue socket back to
non-blocking.
Signed-off-by: Ralph Boehme <slow@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
This introduces caching of unix datagram send queues. Right now send
queues are only created for peers if the channel to the peer is full and
a send reported EWOULDBLOCK.
At this stage, performance will actually be slightly worse, because now
if there's a cached queue for a peer without queued messages, we don't
attempt direct send anymore until the send queue is removed from the
cache.
The next commit will modify unix_msg to always create a send queue with
the datagram socket in connected mode and again attempt an non-blocking
send on the connected socket first. Then only if that returns
EWOULDBLOCK, the send has to go through the threadpool.
Signed-off-by: Ralph Boehme <slow@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
This allows prepare_socket_nonblock() to be called to set a socket to
non-blocking (as before) as well as blocking. This will be used in a
subsequent commit.
Signed-off-by: Ralph Boehme <slow@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
In one of the next commits unix_dgram_send_queue_init() will be moved
into find_send_queue and that takes a struct sockaddr_un.
Signed-off-by: Ralph Boehme <slow@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
This essentially reverts 1c4284c739. We now call an alien function from
within pthreadpool, and we should not hold a mutex during that call. The alien
function could (and pthreadpool_tevent_job_signal actually does) lock a mutex.
We can't guarantee proper lock ordering here, so in theory we could deadlock. I
haven't seen it in the wild yet, but I could imagine that both _parent pieces
in pthreadpool and tevent could trigger such a deadlock.
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
Autobuild-User(master): Jeremy Allison <jra@samba.org>
Autobuild-Date(master): Tue Aug 30 04:06:20 CEST 2016 on sn-devel-144
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Ralph Boehme <slow@samba.org>
Autobuild-User(master): Ralph Böhme <slow@samba.org>
Autobuild-Date(master): Mon Aug 29 22:51:34 CEST 2016 on sn-devel-144
This is a replacement for fncall.[ch] without having to go through
a pipe for job completion signalling
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Stefan Metzmacher <metze@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
The pthreadpool_tevent wrapper will need this
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Stefan Metzmacher <metze@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
First step to separate the signalling mechanism from the core pthreadpool code.
A later patch will add a pthreadpool that directly indicates job completion via
tevent_threaded_schedule_immediate.
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
Logically this belongs into the previous patch, but tevent deserves isolated
patches :-)
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Stefan Metzmacher <metze@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
Autobuild-User(master): Jeremy Allison <jra@samba.org>
Autobuild-Date(master): Tue Aug 23 10:46:17 CEST 2016 on sn-devel-144
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Andreas Schneider <asn@samba.org>
Autobuild-User(master): Jeremy Allison <jra@samba.org>
Autobuild-Date(master): Sat Aug 20 09:41:33 CEST 2016 on sn-devel-144
Some systems required an explicit <signal.h>, which comes
via "system/wait.h"
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Reviewed-by: Volker Lendecke <vl@samba.org>
Autobuild-User(master): Stefan Metzmacher <metze@samba.org>
Autobuild-Date(master): Tue Aug 16 19:09:55 CEST 2016 on sn-devel-144
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
Autobuild-User(master): Jeremy Allison <jra@samba.org>
Autobuild-Date(master): Fri Aug 12 04:26:09 CEST 2016 on sn-devel-144
When fetching records from remote ctdb nodes via ctdbd_parse() or in
db_ctdb_traverse(), we already check for tombstone records and skip
them. This was originally also done for the ltdb checks.
See also bug: https://bugzilla.samba.org/show_bug.cgi?id=10008
(commit 1cae59ce11).
Commit 925625b528 reverted part of the
patch of bug 10008 due to a deadlock it introduced.
This patch re-introduces the consistent treatment of empty records in
the ltdb but avoids the deadlock by correctly signalling
NT_STATUS_NOT_FOUND if an empty record is found authoritatively in
the ltdb and not calling ctdb in this case.
BUG: https://bugzilla.samba.org/show_bug.cgi?id=12005
Pair-Programmed-With: Michael Adam <obnox@samba.org>
Signed-off-by: Ralph Boehme <slow@samba.org>
Signed-off-by: Michael Adam <obnox@samba.org>
Autobuild-User(master): Michael Adam <obnox@samba.org>
Autobuild-Date(master): Tue Aug 9 04:38:44 CEST 2016 on sn-devel-144
Signed-off-by: Michael Adam <obnox@samba.org>
Reviewed-by: Andreas Schneider <asn@samba.org>
Autobuild-User(master): Michael Adam <obnox@samba.org>
Autobuild-Date(master): Thu Aug 4 22:08:35 CEST 2016 on sn-devel-144
Trying to trim down messages.c a bit: Sending to all processes that are
registered in serverid.tdb and filtering to me is not really logic of general
messaging but more of the serverid code.
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Ralph Boehme <slow@samba.org>
This will be used between cleanupd and smbd for passing information
about exitted smbd childs.
Bug: https://bugzilla.samba.org/show_bug.cgi?id=12022
Signed-off-by: Ralph Boehme <slow@samba.org>
Reviewed-by: Volker Lendecke <vl@samba.org>
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
Autobuild-User(master): Jeremy Allison <jra@samba.org>
Autobuild-Date(master): Sat Jul 16 00:17:03 CEST 2016 on sn-devel-144
The existing one with a separate dbwrap_watchers.tdb turns out to
create a performance penalty in a clustered environment. Non-clustered,
dbwrap_parse_record on non-existent records is very cheap, but in a
cluster environment this is very noticable.
This implementation puts the watcher information into the records itself. For
large records, this might be another performance penalty, because we have to
assemble the final record together with talloc and memcpy, but this might be
fixed later with a tdb_storev call.
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
It's highly unlinkely that this will ever kick in, because our current tdb keys
are rather small, but offset calculations without overflow checks are bad.
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
Typicall, when we watch a record, we wait for a process to give up some
resource. Be it an oplock, a share mode or the g_lock. If everything goes well,
the blocker sends us a message. If the blocker dies hard, we want to also be
informed immediately.
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
This is a brute force variant, trying twice a second. We'll have better
variants with tmsgd in the future.
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
... something I've wanted to do for ages :-)
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Michael Adam <obnox@samba.org>
Autobuild-User(master): Michael Adam <obnox@samba.org>
Autobuild-Date(master): Thu Jul 14 20:50:15 CEST 2016 on sn-devel-144
This is the last step to fix a regression introduced by
3fe3226daa and
3fe3226daa8488e0fa787c40359c3401b6f05fc0^
where we pass the ctdb-messaging object conn to db_open() and add a
reference to it to the private db_ctdb_ctx for later use. Unfortunately
reinit_after_fork() destroys conn, leaving us with an invalid reference.
The previous patches added new lower level functions
messaging_ctdbd_reinit() and ctdbd_reinit_connection(), finally use them
them from messaging_reinit(). They preserve the conn object and simply
reinitialize the IPC fd.
Signed-off-by: Ralph Boehme <slow@samba.org>
Reviewed-by: Volker Lendecke <vl@samba.org>
Autobuild-User(master): Ralph Böhme <slow@samba.org>
Autobuild-Date(master): Mon Jul 11 23:45:20 CEST 2016 on sn-devel-144
Split out and internal function from messaging_ctdbd_init() that does
the connection setup. Keep the conn object allocation in
messaging_ctdbd_init().
This is in preperation of adding messaging_ctdbd_reinit() which will use
the new internal function as well.
messaging_ctdbd_init_internal() has a new reinit flag,
messaging_ctdbd_init() calls with reinit=false resulting in unmodified
behaviour.
Signed-off-by: Ralph Boehme <slow@samba.org>
Reviewed-by: Volker Lendecke <vl@samba.org>
Split ctdbd_init_connection() into an internal function that does the
connection setup and only keep the conn object allocation in
ctdbd_init_connection().
This is in preperation of adding ctdbd_reinit_connection() which will
use the new internal function as well.
Signed-off-by: Ralph Boehme <slow@samba.org>
Reviewed-by: Volker Lendecke <vl@samba.org>
Only expose a synchronous tldap_gensec_bind() call, as it can't
really be async anyway.
Signed-off-by: Jeremy Allison <jra@samba.org>
Reviewed-by: Ralph Boehme <slow@samba.org>
Autobuild-User(master): Ralph Böhme <slow@samba.org>
Autobuild-Date(master): Tue Jun 28 11:19:33 CEST 2016 on sn-devel-144
This should prevent nested event loops, and won't make the
code performance worse anyway.
Signed-off-by: Jeremy Allison <jra@samba.org>
Reviewed-by: Ralph Boehme <slow@samba.org>
CTDB can send CTDB_REPLY_ERROR in case it encounters an error condition.
This is treated as successful migration as "ret" is not set.
Signed-off-by: Amitay Isaacs <amitay@gmail.com>
Reviewed-by: Volker Lendecke <vl@samba.org>
Autobuild-User(master): Volker Lendecke <vl@samba.org>
Autobuild-Date(master): Fri Jun 24 22:39:23 CEST 2016 on sn-devel-144
If the node has flag NODE_FLAGS_PERMANENTLY_DISABLED, then it is
still working. Only avoid connections if node has any of the flags
in NODE_FLAGS_INACTIVE.
Signed-off-by: Amitay Isaacs <amitay@gmail.com>
Reviewed-by: Martin Schwenke <martin@meltin.net>
Reviewed-by: Volker Lendecke <vl@samba.org>
Autobuild-User(master): Volker Lendecke <vl@samba.org>
Autobuild-Date(master): Thu Jun 23 23:38:44 CEST 2016 on sn-devel-144
