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.
This is a real bug: tevent_req_set_endtime already calls tevent_req_nomem.
Autobuild-User: Volker Lendecke <vlendec@samba.org>
Autobuild-Date: Mon Mar 21 16:29:22 CET 2011 on sn-devel-104
This makes us scale better with many simultaneous winbind requests,
some of which might be slow.
This implementation breaks offline logons, as the cached credentials are
maintained in a child (this needs fixing). So, if the offline logons are
active, only allow one DC connection.
Probably the offline logon and the scalable file server cases are
separate enough so that this patch is useful even with the restriction.
pass this in as the &now parameter. Push this call inside of
event_add_to_select_args() to the correct point so it doesn't
get called unless needed.
Jeremy.
Autobuild-User: Jeremy Allison <jra@samba.org>
Autobuild-Date: Thu Dec 23 01:08:11 CET 2010 on sn-devel-104
If a child dies, the parent process right away closes the socket.
This is wrong, with tevent we still have events pending. This works
fine for epoll but does not for at least the FreeBSD select variant.
Tevent sticks a closed socket into the select masks. This then
returns an error EBADF. When this happens, the parent winbind dies
instead of forking a new child.
This moves the socket close from the SIGCHLD cleanup function to
the socket receiver. I could not reproduce the parent death anymore
and it did not create an obvious fd leak.
Autobuild-User: Jeremy Allison <jra@samba.org>
Autobuild-Date: Mon Dec 6 23:21:02 CET 2010 on sn-devel-104
This will reduce the noise from merges of the rest of the
libcli/security code, without this commit changing what code
is actually used.
This includes (along with other security headers) dom_sid.h and
security_token.h
Andrew Bartlett
Autobuild-User: Andrew Bartlett <abartlet@samba.org>
Autobuild-Date: Tue Oct 12 05:54:10 UTC 2010 on sn-devel-104
Previously, only one fd handler was being called per main message loop
in all smbd child processes.
In the case where multiple fds are available for reading the fd
corresponding to the event closest to the beginning of the event list
would be run. Obviously this is arbitrary and could cause unfairness.
Usually, the first event fd is the network socket, meaning heavy load
of client requests can starve out other fd events such as oplock
or notify upcalls from the kernel.
In this patch, I have changed the behavior of run_events() to unset
any fd that it has already called a handler function, as well
as decrement the number of fds that were returned from select().
This allows the caller of run_events() to iterate it, until all
available fds have been handled.
I then changed the main loop in smbd child processes to iterate
run_events(). This way, all available fds are handled on each wake
of select, while still checking for timed or signalled events between
each handler function call. I also added an explicit check for
EINTR from select(), which previously was masked by the fact that
run_events() would handle any signal event before the return code
was checked.
This required a signature change to run_events() but all other callers
should have no change in their behavior. I also fixed a bug in
run_events() where it could be called with a selrtn value of -1,
doing unecessary looping through the fd_event list when no fds were
available.
Also, remove the temporary echo handler hack, as all fds should be
treated fairly now.
This is supposed to improve the winbind reconnect time after an ip address
has been moved away from a box. Any kind of HA scenario will benefit from
this, because winbindd does not have to wait for the TCP timeout to kick in
when a local IP address has been dropped and DC replies are not received
anymore.
This shrinks include/includes.h.gch by the size of 7 MB and reduces build time
as follows:
ccache build w/o patch
real 4m21.529s
ccache build with patch
real 3m6.402s
pch build w/o patch
real 4m26.318s
pch build with patch
real 3m6.932s
Guenther
The main problem is that we call CatchChild() within the
parent winbindd, which overwrites the signal handler
that was registered by winbindd_setup_sig_chld_handler().
That means winbindd_sig_chld_handler() and winbind_child_died()
are never triggered when a winbindd domain child dies.
As a result will get "broken pipe" for all requests to that domain.
To reduce the risk of similar bugs in future we call
CatchChild() in winbindd_reinit_after_fork() now.
We also use a full winbindd_reinit_after_fork() in the
cache validation child now instead instead of just resetting
the SIGCHLD handler by hand. This will also fix possible
tdb problems on systems without pread/pwrite and disabled mmap
as we now correctly reopen the tdb handle for the child.
metze
The machine password handler has code to deal with every node in the cluster
trying to change the machine password at the same time. However, it is not very
nice to the DC if everyone tries this simultaneously. This adds a random 0-255
second offset to our timed event. When this fires a bit later than strictly
calculated, someone else might have stepped in and have already changed it. The
timed event handler will handle this gracefully, it won't even try to do it
again.
In itself, this is pretty pointless. But in the next steps I'll convert the
winbind internal communication to wb_reqtrans which allocates the request
properly. This minimizes the later diff.
Volker
When we run out of file descriptors for some reason, every new
connection forks a child that immediately panics causing smbd to
coredump. This seems unnecessarily harsh; with this code change we
now catch that error and merely log a message about it and exit
without the core dump.
Signed-off-by: Tim Prouty <tprouty@samba.org>
The first is "kerberos method" and replaces the "use kerberos keytab"
with an enum. Valid options are:
secrets only - use only the secrets for ticket verification (default)
system keytab - use only the system keytab for ticket verification
dedicated keytab - use a dedicated keytab for ticket verification.
secrets and keytab - use the secrets.tdb first, then the system keytab
For existing installs:
"use kerberos keytab = yes" corresponds to secrets and keytab
"use kerberos keytab = no" corresponds to secrets only
The major difference between "system keytab" and "dedicated keytab" is
that the latter method relies on kerberos to find the correct keytab
entry instead of filtering based on expected principals.
The second parameter is "dedicated keytab file", which is the keytab
to use when in "dedicated keytab" mode. This keytab is only used in
ads_verify_ticket.
From analysis by hargagan <shargagan@novell.com> :
"The winbindd_child_died() is also getting called from process_loop() in case of
SIGCHLD signal. In this case it doesn't make the timeout_handler to NULL for
the first request. It then initiate a new request using
schedule_async_request() which installs a new timeout handler for the same
request. In such a case, for a badly unresponsive system both the timeout
handler can be called. For the first call the "private_data" will be cleared
and for another call the timeout handler will be detecting the double free. So,
for such a case as well, the winbindd_child_died() should make the
timeout_handler to NULL."
Jeremy.
The scanner did not figure out that we always have a primary domain, so it
complained about us potentially passing a NULL pointer down to
set_domain_online_request() where it is dereferenced.
Make the code a bit clearer.
(This used to be commit e6e8d108f9)
When SIGCHLD handling is delayed for some reason, sending a request to a child
can fail early because the child has died already. In this case
async_main_request_sent() directly called the continuation function without
properly removing the malfunctioning child process and the requests in the
queue. The next request would then crash in the DLIST_ADD_END() in
async_request() because the request pending for the child had been
talloc_free()'ed and yet still was referenced in the list.
This one is *old*...
Volker
(cherry picked from commit 8691709626)
(This used to be commit c70e2b6476)
In reloading the smb.conf, if a "log file" is specified in smb.conf,
winbind children will overwrite the logfile name to be the same as the
parent.
Jeremy.
(This used to be commit 62d319cc1a)
smbcontrol winbindd debug level would only set the debug level of the
parent winbindd process and not the child processes. This patch adds
the functionality of broadcasting the debug message to all winbindd
children. Now the debug level message is propagated to all the winbindd
processes that includes parent and children.
(This used to be commit cfbcfc3ffe)
CatchChild();
*before* we fork the domain child. This call establishes a signal handler that
eats SIGCLD signals and doesn't call sys_select_signal() as the main daemon
SIGCLD handler should do. This causes the parent to ignore dead children and
time out, instead of calling winbind_child_died() on receipt of the signal. The
correct fix is to move the CatchChild call into the child code after the fork.
Jeremy.
(This used to be commit 8d701a142b)
Thanks to Glenn Curtis and Kyle Stemen @ Likewise. Their explanation is:
In winbindd_dual.c, there is a list of children processes that
is maintained using macros DTLIST_ADD and DTLIST_REMOVE. In the
case when a scheduled_async_request fails, the particular child
was located in the list, and its attributes were cleared out
and it was reused for a subsequent async request. The bug was that
the new request would queue the same node into the doubly-linked
list and would result in list->next pointing to the same node as
list itself. This would set up an infinite loop in the processing of
the for loop when the list of children was referenced.
Solution was to fully remove the child node from the list, such that
it could be inserted without risk of being inserted twice.
Note that the child is re-added to the list in fork_domain_child() again.
(This used to be commit b379b5b5d8)
We now open messages.tdb even before we do the become_daemon. become_daemon()
involves a fork and an immediate exit of the parent, thus the
parent_is_longlived argument must be set to false in this case. The parent is
not really long lived :-)
(This used to be commit 4f4781c6d1)
