| Age | Commit message (Collapse) | Author |
|---|
|
This patch re-introduces `-s queue' but now it displays generic
queue statistics.
# conntrackd -s queue
active queue objects: 0
queue txqueue:
current elements: 0
maximum elements: 2147483647
not enough space errors: 0
queue rsqueue:
current elements: 72
maximum elements: 128
not enough space errors: 0
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
|
|
This patch adds the name field to identify the queue by means of
a string. This patch is used by the next one that introduces
per-queue statistics.
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
|
|
This patch adds a dedicated commit handler since there is a possible
race condition that can happen if the child process ends before we
have received all the event messages that the commit request has
triggered.
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
|
|
In 6f5666a29cb7cbff08ce926ee1edb84a311ff6ee, I moved the flush
operation into a child process and to use a disposable handler
to perform flush requests. This patch adds a dedicated flush
handler since there is a possible race condition that can
happen if the child process ends before we have received all
the event messages that the flush request has triggered.
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
|
|
This patch allows to limit the number of simultaneous child processes.
This is required by the next patch that replaces disposable handlers
to commit and flush with permanent handlers.
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
|
|
This patch removes the cache write-through clause. This feature
remained undocumented although some has found it looking at the
source code. This feature has remained in the tree for quite
some time although it has several limitations. Moreover, it
is specifically broken and dangerous for Linux kernels >= 2.6.29
since it generates loops in the synchronization.
We do this removal first to prepare the introduction of a feature
to bypass the external cache.
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
|
|
With this patch, the flush operation is performed by a child process.
Thus, the parent process digests destroy events that ctnetlink reports
back and, thanks to the origin infrastructure, we skip the messy
implicit synchronization that are triggered by such events.
This patch requires a Linux kernel >= 2.6.29 to benefit from this
change, otherwise it has no effect.
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
|
|
Since Linux kernel 2.6.29, ctnetlink reports the changes that have
been done using ctnetlink. With this patch, conntrackd can recognize
who is the origin of the event messages. For example, this is
interesting to avoid a messy implicit bulk send during the commit
of entries.
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
|
|
This patch adds a simple infrastructure that allows to account
the child processes that have been forked. This also includes
a callback handler that can be registered that is called once
the child process finishes.
We can extended this later to include an alarm to limit the
maximum lifetime of a forked child process. This is good to
ensure that child processes behave timely.
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
|
|
This patch fixes a problem that was introduced while adding the
multichannel support.
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
|
|
This patch obsoletes `-s multicast' by `-s link' to display the
dedicated link statistics, as the current dedicated link protocol
use can be unicast UDP or multicast. The term "link" is more
generic.
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
|
|
This patch is a cleanup, it removes a couple of reminiscent
references to multicast (as now conntrackd is independent of
the protocol used to replicate state-changes, currently supports
unicast UDP and multicast).
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
|
|
This patch saves one tab in the code.
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
|
|
This patch fixes a misleading output that shows the number of lost
and malformed packets. Instead, those numbers show the number of
the number of lost and malformed messages.
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
|
|
This patch reworks conntrackd to make it independent of the protocol
used to propagate state-changes. This patch adds the channel layer
abstraction, this layer allows you to add support for different
protocols like unicast UDP or TIPC.
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
|
|
This patch renames the mcast_iface to interface since this nlif
handler is not related with multicast itself, but to monitor the
link interface used to propagate state-changes. This patch is a
cleanup.
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
|
|
This patch is a cleanup. It renames the function mcast_send_sync()
to sync_send() since the function itself is not related to multicast
anymore (it enqueues state-changes to the upper layer).
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
|
|
This patch adds include limits.h to mcast.c and sync-mode.c. Why Gentoo
maintainers did not report me the problem? :(
http://bugs.gentoo.org/show_bug.cgi?id=256497
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
|
|
This patch removes debug() and debug_ct(), I haven't use the
debugging information that these functions provide in years.
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
|
|
This patch makes conntrackd change the current dedicated link if
the message is correct, ie. neither malformed nor out-of-sync.
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
|
|
This patch allows flushing the internal and/or the external cache.
The `-f' with no extra parameters still works to flush both the
internal and the external cache.
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
|
|
This patch moves the polling logic into the resync handler. The
dump handler action depended on the daemon working mode (polling
or event-driven) resulting in an inconsistent behaviour.
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
|
|
Currently, the daemon sends a flush request to the kernel-space.
With lots of entries and NetlinkOverrunResync disabled, the daemon
remains in an inconsistent state due to an overrun produced by the
flush report to userspace. With this patch, the daemon also flush
its internal cache after the kernel flush request.
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
|
|
This patch improves the polling support included in 0.9.10. The
polling now consists of getting the state table, wait for PollSecs,
then purge obsolete entries, and so on.
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
|
|
With this patch, the `-t' option adds an alarm that will flush the
cache after CONFIG(purge_timeout) seconds specified in the config
file. This looks much cleaner and more performance that looping
on the entire conntrack table to set the new timeout of every single
entry.
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
|
|
This patch moves the file descriptor registration after the
initialization instead of having a specific hook for this.
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
|
|
This patch is a cleanup. The overrun handler is actually a way to
resynchronize against the conntrack kernel table. The name overrun
was used because it was initially its purpose. The new naming shows
its genericity.
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
|
|
This patch adds support for redundant dedicated links. You can add
a pool of dedicated links that can be used if the current active
fails.
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
|
|
With this patch, all the states updates are enqueued in the
tx_queue. Thus, there's a single output path. This patch adds a
simple refcounting mechanism to note when an object is sitting in
the txqueue. This patch also removes the alarm that is required by
the ftfw approach.
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
|
|
This patch adds a generic tx queue for all synchronization modes.
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
|
|
This patch unifies the tx_list and the tx_queue to have only one
transmission queue. Since the tx_list hold state objects and
tx_queue control messages, I have introduced a queue node type
that can be used to differenciate the kind of information that
the node stores: object or control message.
This patch also reworks the existing queue class to include a
file descriptor that can be used to know if there are new data
added to the queue (see QUEUE_F_EVFD flag).
In this change, I have also modified the current evfd to make the
file descriptor to make read operations non-blocking. Moreover,
it keeps a counter that is used to know how many messages are
inserted in the queue.
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
|
|
Currently, the caching system is implemented in a two layer
architecture: hashtable (inner layer) and cache (upper layer).
This patch reworks the hash-cache infrastructure to solve some
initial design problems to make it more flexible, the main strong
points of this patch are:
* Memory handling is done in the cache layer, not in the inner
hashtable layer. This removes one of the main dependencies between
the hashtable and the cache classes.
* Remove excessive encapsulation: the former cache used to hide a
lot of details of the inner hashtable implementation.
* Fix over-hashing of some operations: lookup-delete-add required
three hash calculations. Similarly, the update-or-add operation
required two hash calculations. Now, we calculate the hash once
and re-use the value how many times as we need.
This patch simplifies the caching system. As a result, we save ~130
lines of code. Small code means and less complexity means less
chance to have bugs.
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
|
|
This patch removes unnecessary split lines in several log messages.
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
|
|
This patch adds cache statistics that you can check via
`conntrackd -s cache'. This information is useful for
trouble-shooting.
This patch replaces several log messages that can be triggered in
runtime. The idea behind this patch is to avoid log message flooding
under errors.
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
|
|
This patch adds networks statistics that you can check via
`conntrackd -s network'. This information is useful for
trouble-shooting.
This patch replaces several log messages that can be triggered in
runtime. The idea behind this patch is to avoid log message flooding
under errors.
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
|
|
This patch adds a message-type checking before we parse the message.
Thus, we skip the parsing of messages with bad types.
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
|
|
This patch reduces the size of alive messages by removing the
"from" and "to" fields which are not of any help. This patch also
removes the IS_CTL() macro since it does not return true for
the control messages anymore but only for IS_ACK(), IS_NACK() and
IS_RESYNC().
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
|
|
This patch replaces the use of NFCT_Q_* in the message type by
specific network message type NET_T_*. The query types are reserved
for libnetfilter_conntrack operations.
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
|
|
This patch simplifies the message format of the replication
messages. As a result, we save four bytes. The netpld header
was introduced in the early protocol design. Today, it does
not have any reason to exist.
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
|
|
This patch simplifies mcast_buffered_send_netmsg() by removing the
length parameter. Instead, we use the length field in the nethdr
to know the message size to be sent.
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
|
|
This patch changes the prototype of mcast_buffered_init() to receive
as argument the MTU size instead of the multicast configuration. This
decouples the initialization of the tx buffer from the multicast
configuration. This patch is needed by the multi-dedicated link
support.
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
|
|
This patch is a cleanup. It moves the callbacks from netlink.c to
run.c where they are actually invoked. This is better for code
readability as I usually have to switch from run.c to netlink.c
to remember what the callbacks actually do.
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
|
|
This patch adds the version field (8-bits long) to the nethdr
structure. This fields can be used to indicate the protocol version
in case that we detect an incompatibility between two conntrackd
daemons working with different protocol versions.
Unfortunately, this patch breaks backward compatibility, ie.
conntrackd <= 0.9.8 protocol is not compatible with the upcoming
conntrackd >= 0.9.9. Better do this now than later.
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
|
|
This patch avoids a double filtering in user-space and kernel-space if
the kernel support BSF. Since we do not use BSF for dumps and resyncs,
we add a new parameter to ignore_conntrack to indicate if we have to
perform the filtering in user-space or not.
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
|
|
This patch adds the clause PurgeTimeout that sets the new timer
when conntrackd -t is called. This command is particularly useful
when the sysadmin triggers hand-overs between several nodes without
rebooting as it reduces the timers of the remaining entries in
the kernel. Thus, avoiding clashes between new and old entries that
may trigger INVALID packets.
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
|
|
This patch adds the new option `-t' for conntrackd. This option shortens
the value of the timeout for the cached entries that lives in the
kernel. This option is particularly useful to remove the zombie
established entries that remain in kernel if the user tests the platform
by forcing the takeover from one to another node several times.
We currently use the value of CommitTimeout which is sane for it. Adding
a new option does not seem to add more flexibility IMO.
Once we get the patches to notify user changes via ctnetlink and the
netlink flag NLM_F_ECHO works, we may directly invoke a massive purge of
the entries, however, such solution would still need evaluation.
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
|
|
This patch fixes the non-timer-based cache deletion. This bug affects
the alarm-based approach since the backup replicas did not get the
deletion event, thus, delaying the deletion.
This patch introduces cache_find() to look up for a conntrack object
and __cache_del_timer() to perform direct deletions by means of the
pointer obtained with cache_find().
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
|
|
This patch reworks the user-space filtering. Although we have
kernel-space filtering since Linux kernel >= 2.6.26, we keep userspace
filtering to ensure backward compatibility. Moreover, this patch
prepares the implementation of the kernel-space filtering via
libnetfilter_conntrack's high-level berkeley socket filter API.
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
|
|
|
|
|
