MEDIUM: lua: interrupt the Lua execution for running other process
This patch permits to interrupt the Lua execution each n# of
instructions. It is useful for controling the execution time
of an Lua code, and permits at the HAProxy schedule to process
some others tasks waiting execution.
This flag indicate that the current yield is returned by the Lua
execution task control. If this flag is set, the current task may
quit but will be set in the run queue to be re-executed immediatly.
This patch modify the "hlua_yieldk()" function, it adds an argument
that contain a field containing yield options.
MEDIUM: lua: each yielding function returns a wake up time.
This is used to ensure that the task doesn't become a zombie
when the Lua returns a yield. The yield wrapper ensure that an
timer used for waking up the task will be set.
The timer is reseted to TICK_ETERNITY if the Lua execution is
done.
MINOR: lua: set skeleton for Lua execution expiration
This first patch permits to cofigure the Lua execution exipiration.
This expiration is configured but it is not yet avalaible, it will
be add in a future patch.
MINOR: lua: use bitfield and macro in place of integer and enum
In the future, the lua execution must return scheduling informations.
We want more than one flag, so I convert an integer used with an
enum into an interer used as bitfield.
BUG/MAJOR: lua: send function fails and return bad bytes
In some cases the Lua "send" function fails. This is caused by the
return of "buffer_replace2()" is not tested. I checked avalaible space
in the buffer and I supposed than "buffer_replace2()" took all the
data. In some cases, "buffer_replace2()" cannot take the incoming
data, it returns the amount of data copied.
This patch check the amount of data really copied by "buffer_replace2()"
and advance the buffer with taking this value in account.
Cyril Bonté [Tue, 3 Mar 2015 22:26:14 +0000 (23:26 +0100)]
BUILD/CLEANUP: systemd: avoid a warning due to mixed code and declaration
Gcc complains because the systemd wrapper mixed code and declaration :
"warning: ISO C90 forbids mixed declarations and code
[-Wdeclaration-after-statement]".
Cyril Bonté [Tue, 3 Mar 2015 22:13:03 +0000 (23:13 +0100)]
BUILD: try to automatically detect the Lua library name
Depending on the distribution, the Lua library can have different names.
Some distributions will require -llua5.2, others -llua52, and other systems may
require -llua.
Now, the Makefile will try to guess the library name, in order of priority :
"lua5.2", "lua52", or "lua".
Cyril Bonté [Sun, 1 Mar 2015 23:08:41 +0000 (00:08 +0100)]
BUG/MEDIUM: lua: segfault when calling haproxy sample fetches from lua
When a Lua script calls an internal haproxy sample fetch, it may segfault in
some conditions :
- when a fetch has no argument,
- when there is no room left to store the special type ARGT_STOP in the argument
list (this one shouldn't happen currently as there isn't any sample fetch with
enough arguments to fill the allocated buffer).
Example of Lua code which reproduces a segfault :
core.register_fetches("segfault", function(txn, ...)
return txn.req_ver(txn)
end)
Cyril Bonté [Sun, 1 Mar 2015 23:08:40 +0000 (00:08 +0100)]
MINOR: lua: add a compilation error message when compiled with an incompatible version
haproxy Lua support begins with Lua 5.2. In order to ease the diagnostic on
compilation error, a preprocessor error is added when an incompatible version
is used.
The compatibility is determined by the presence of LUA_VERSION_NUM and its
magic value (502 for Lua 5.2.x).
Cyril Bonté [Sun, 1 Mar 2015 23:08:39 +0000 (00:08 +0100)]
MINOR: report the Lua version in -vv
As of the other libraries used by haproxy, it can be useful to display the Lua
version used at compilation time.
A new line is added to "haproxy -vv", which shows if Lua is supported by the
binary, and with which version it was compiled.
Willy Tarreau [Fri, 27 Feb 2015 15:38:20 +0000 (16:38 +0100)]
BUILD: lua: cleanup many mixed occurrences declarations & code
The code was a bit of a pain to follow because of this, especially
when some of it heavily relies on longjmp... This refreshing makes
it slightly easier to read.
Thierry FOURNIER [Tue, 17 Feb 2015 14:01:59 +0000 (15:01 +0100)]
MINOR: lua: core: can yield an execution stack
This patch provides a yield function. This function permits to
give back the hand at the HAProxy scheduler. It is used when the
lua processing consumes a lot of time.
The channel class permits manipulation of channels. A channel is
an FIFO buffer between the client and the server. This class provides
function to read, write, forward, destroy and alter data between
the input and the ouput of the buffer.
Thierry FOURNIER [Mon, 16 Feb 2015 18:27:16 +0000 (19:27 +0100)]
MEDIUM: lua: socket: add "socket" class for TCP I/O
This patch adds the TCP I/O functionnality. The class implemented
provides the same functions than the "lua socket" project. This
make network compatibility with another LUA project. The documentation
is located here:
Thierry FOURNIER [Mon, 16 Feb 2015 18:43:25 +0000 (19:43 +0100)]
MINOR: lua: core: add sleep functions
This version of sleep is based on a coroutine. A sleeping
task is started and a signal is registered. This sleep version
must disapear to be replaced by a version using the internal
timers.
Thierry FOURNIER [Mon, 16 Feb 2015 19:21:12 +0000 (20:21 +0100)]
MINOR: lua: register and execute converters in LUA
This patch permits to write LUA converters. Note that
all the converters declared trough LUA are automatically
prefixed by "lua.".
The Lua converters needs the current session to be executed,
but the writed and executed Lua code must be static and
contextless. The TXN object is not created for the converters.
Thierry FOURNIER [Mon, 16 Feb 2015 19:14:51 +0000 (20:14 +0100)]
MINOR: lua: txn: import existing sample-fetches in the class TXN
This patch adds the browsing of all the HAProxy fetches and
create associated LUA functions. The HAProxy internal fetches
can be used in LUA trough the class "TXN".
Note that the symbols "-", "+" and "." in the name of current
sample fetch are rewrited as "_" in LUA because ".", "-" and "+"
are operators.
Thierry FOURNIER [Wed, 25 Feb 2015 10:43:21 +0000 (11:43 +0100)]
MINOR: lua: add shared context in the lua stack
The shared context is a global reference dedicated to one Lua stack.
Even if the TXN is not required, the shared context it is associated
with a TXN object. In fact, the shared has no sense outside a TXN
context
Thierry FOURNIER [Fri, 23 Jan 2015 11:08:30 +0000 (12:08 +0100)]
MINOR: lua: post initialisation bindings
This system permits to execute some lua function after than HAProxy
complete his initialisation. These functions are executed between
the end of the configuration parsing and check and the begin of the
scheduler.
Thierry FOURNIER [Fri, 23 Jan 2015 10:08:20 +0000 (11:08 +0100)]
MEDIUM: lua: "com" signals
This system permits to send signals between lua tasks. A main lua stack can
register the signal in a coprocess. When the coprocess finish his job, it
send a signal, and the associated task is wakes. If the main lua execution
stack stop (with or without errors), the list or pending signals is purged.
Thierry FOURNIER [Fri, 23 Jan 2015 13:06:13 +0000 (14:06 +0100)]
MEDIUM: lua: lua integration in the build and init system.
This is the first step of the lua integration. We add the useful
files in the HAProxy project. These files contains the main
includes, the Makefile options and empty initialisation function.
Is is the LUA skeleton.
Thierry FOURNIER [Mon, 16 Feb 2015 18:26:48 +0000 (19:26 +0100)]
MINOR: channel: functions to get data from a buffer without copy
We now have functions to retrieve one block and one line from
either the input or the output part of a buffer. They return
up to two (pointer,length) values in case the buffer wraps.
Thierry FOURNIER [Wed, 25 Feb 2015 12:36:14 +0000 (13:36 +0100)]
MEDIUM: http/tcp: permit to resume http and tcp custom actions
Later, the processing of some actions needs to be interrupted and resumed
later. This patch permit to resume the actions. The actions that needs
to run with the resume mode are not yet avalaible. It will be soon with
Lua patches. So the code added by this patch is untestable for the moment.
The list of "tcp_exec_req_rules" cannot resme because is called by the
unresumable function "accept_session".
Thierry FOURNIER [Wed, 25 Feb 2015 12:51:19 +0000 (13:51 +0100)]
MEDIUM: http: change the code returned by the response processing rule functions
Actually, this function returns a pointer on the rule that stop
the evaluation of the rule list. Later we integrate the possibility
of interrupt and resue the processsing of some actions. The current
response mode is not sufficient to returns the "interrupt" information.
The pointer returned is never used, so I change the return type of
this function by an enum. With this enum, the function is ready to
return the "interupt" value.
Thierry FOURNIER [Fri, 12 Dec 2014 18:41:33 +0000 (19:41 +0100)]
MEDIUM: tcp: add register keyword system.
This patch introduces an action keyword registration system for TCP
rulesets similar to what is available for HTTP rulesets. This sytem
will be useful with lua.
Thierry FOURNIER [Mon, 23 Feb 2015 14:11:11 +0000 (15:11 +0100)]
MINOR: converters: give the session pointer as converter argument
Some usages of the converters need to know the attached session. The Lua
needs the session for retrieving his running context. This patch adds
the "session" as an argument of the converters prototype.
Thierry FOURNIER [Thu, 26 Feb 2015 09:40:09 +0000 (10:40 +0100)]
MEDIUM: channel: wake up any request analyzer on response activity
This behavior is already existing for the "WAIT_HTTP" analyzer,
this patch just extends the system to any analyzer that would
be waked up on response activity.
Thierry FOURNIER [Wed, 14 Jan 2015 10:31:49 +0000 (11:31 +0100)]
MEDIUM: protocol: automatically pick the proto associated to the connection.
When the destination IP is dynamically set, we can't use the "target"
to define the proto. This patch ensures that we always use the protocol
associated with the address family. The proto field was removed from
the server and check structs.
Thierry FOURNIER [Mon, 15 Dec 2014 12:26:01 +0000 (13:26 +0100)]
MAJOR: poll: only rely on wake_expired_tasks() to compute the wait delay
Actually, HAProxy uses the function "process_runnable_tasks" and
"wake_expired_tasks" to get the next task which can expires.
If a task is added with "task_schedule" or other method during
the execution of an other task, the expiration of this new task
is not taken into account, and the execution of this task can be
too late.
Actualy, HAProxy seems to be no sensitive to this bug.
This fix moves the call to process_runnable_tasks() before the timeout
calculation and ensures that all wakeups are processed together. Only
wake_expired_tasks() needs to return a timeout now.
Nenad Merdanovic [Fri, 27 Feb 2015 18:56:49 +0000 (19:56 +0100)]
MEDIUM: Add support for configurable TLS ticket keys
Until now, the TLS ticket keys couldn't have been configured and
shared between multiple instances or multiple servers running HAproxy.
The result was that if a request got a TLS ticket from one instance/server
and it hits another one afterwards, it will have to go through the full
SSL handshake and negotation.
This patch enables adding a ticket file to the bind line, which will be
used for all SSL contexts created from that bind line. We can use the
same file on all instances or servers to mitigate this issue and have
consistent TLS tickets assigned. Clients will no longer have to negotiate
every time they change the handling process.
Simon Horman [Thu, 26 Feb 2015 02:26:17 +0000 (11:26 +0900)]
BUG/MEDIUM: Do not consider an agent check as failed on L7 error
As failure to connect to the agent check is not sufficient to mark it as
failed it stands to reason that an L7 error shouldn't either.
Without this fix if an L7 error occurs, for example of connectivity to the
agent is lost immediately after establishing a connection to it, then the
agent check will be considered to have failed and thus may end up with zero
health. Once this has occurred if the primary health check also reaches
zero health, which is likely if connectivity to the server is lost, then
the server will be marked as down and not be marked as up again until a
successful agent check occurs regardless of the success of any primary
health checks.
This behaviour is not correct as a failed agent check should never cause a
server to be marked as down or by extension continue to be marked as down.
Willy Tarreau [Mon, 23 Feb 2015 15:07:01 +0000 (16:07 +0100)]
MEDIUM: task: always ensure that the run queue is consistent
As found by Thierry Fournier, if a task manages to kill another one and
if this other task is the next one in the run queue, we can do whatever
including crashing, because the scheduler restarts from the saved next
task. For now, there is no such concept of a task killing another one,
but with Lua it will come.
A solution consists in always performing the lookup of the first task in
the scheduler's loop, but it's expensive and costs around 2% of the
performance.
Another solution consists in keeping a global next run queue node and
ensuring that when this task gets removed, it updates this pointer to
the next one. This allows to simplify the code a bit and in the end to
slightly increase the performance (0.3-0.5%). The mechanism might still
be usable if we later migrate to a multi-threaded scheduler.
Thierry FOURNIER [Tue, 17 Feb 2015 15:09:00 +0000 (16:09 +0100)]
BUG/MINOR: ARG6 and ARG7 don't fit in a 32 bits word
The patch "MEDIUM: args: increase arg type to 5 bits and limit arg count
to 5" (dbc79d0a) increased the number of types supported, but forgot to
remove the ARG6/ARG7 macros.
Willy Tarreau [Fri, 20 Feb 2015 12:55:29 +0000 (13:55 +0100)]
MINOR: http: add the new sample fetches req.hdr_names and res.hdr_names
These new sample fetches retrieve the list of header names as they appear
in the request or response. This can be used for debugging, for statistics
as well as an aid to better detect the presence of proxies or plugins on
some browsers, which alter the request compared to a regular browser by
adding or reordering headers.
Willy Tarreau [Sat, 14 Feb 2015 13:14:57 +0000 (14:14 +0100)]
BUG/MAJOR: peers: initialize s->buffer_wait when creating the session
Commit bf883e0 ("MAJOR: session: implement a wait-queue for sessions who
need a buffer") introduced in 1.6 forgot to initialize the buffer_wait
list when the session is initiated by an applet for a peer, resulting in
a crash. Thanks to Chris Kopp for reporting the issue.
BUG/MEDIUM: pattern: some entries are not deleted with case insensitive match
ACL or map entries are not deleted with the command "del acl" or "del map"
if the case insentive flag is set.
This is because the the case insensitive string are stored in a list and the
default delete function associated with string looks in a tree. I add a check
of the case insensitive flag and execute the delete function for lists if it
is set.
Willy Tarreau [Thu, 5 Feb 2015 15:47:07 +0000 (16:47 +0100)]
MINOR: ssl/server: add the "no-ssl-reuse" server option
This option disables SSL session reuse when SSL is used to communicate with
the server. It will force the server to perform a full handshake for every
new connection. It's probably only useful for benchmarking, troubleshooting,
and for paranoid users.
Simon Horman [Fri, 6 Feb 2015 02:11:57 +0000 (11:11 +0900)]
MEDIUM: Allow suppression of email alerts by log level
This patch adds a new option which allows configuration of the maximum
log level of messages for which email alerts will be sent.
The default is alert which is more restrictive than
the current code which sends email alerts for all priorities.
That behaviour may be configured using the new configuration
option to set the maximum level to notice or greater.
Simon Horman [Fri, 6 Feb 2015 02:11:56 +0000 (11:11 +0900)]
MINOR: Remove trailing '.' from email alert messages
This removes the trailing '.' from both the header and the body of email
alerts.
The main motivation for this change is to make the format of email alerts
generated from srv_set_stopped() consistent with those generated from
set_server_check_status().
Willy Tarreau [Tue, 3 Feb 2015 23:45:58 +0000 (00:45 +0100)]
MEDIUM: tcp: implement tcp-ut bind option to set TCP_USER_TIMEOUT
On Linux since 2.6.37, it's possible to set the socket timeout for
pending outgoing data, with an accuracy of 1 millisecond. This is
pretty handy to deal with dead connections to clients and or servers.
For now we only implement it on the frontend side (bind line) so
that when a client disappears from the net, we're able to quickly
get rid of its connection and possibly release a server connection.
This can be useful with long-lived connections where an application
level timeout is not suited because long pauses are expected (remote
terminals, connection pools, etc).
Thanks to Thijs Houtenbos and John Eckersberg for the suggestion.
Simon Horman [Fri, 30 Jan 2015 02:22:56 +0000 (11:22 +0900)]
MEDIUM: Move proto and addr fields struct check
The motivation for this is to make checks more independent of each
other to allow further reuse of their infrastructure.
For nowserver->check and server->agent still always use the same values
for the addr and proto fields so this patch should not introduce any
behavioural changes.
Simon Horman [Fri, 30 Jan 2015 02:22:54 +0000 (11:22 +0900)]
MEDIUM: Refactor init_check and move to checks.c
Refactor init_check so that an error string is returned
rather than alerts being printed by it. Also
init_check to checks.c and provide a prototype to allow
it to be used from multiple C files.
Simon Horman [Fri, 30 Jan 2015 02:22:53 +0000 (11:22 +0900)]
MEDIUM: Remove connect_chk
Remove connect_chk and instead call connect_proc_chk()
and connect_conn_chk(). There no longer seems to be any
value in having a wrapper function here.
Willy Tarreau [Fri, 30 Jan 2015 19:58:58 +0000 (20:58 +0100)]
BUG/MINOR: http: abort request processing on filter failure
Commit c600204 ("BUG/MEDIUM: regex: fix risk of buffer overrun in
exp_replace()") added a control of failure on the response headers,
but forgot to check for the error during request processing. So if
the filters fail to apply, we could keep the request. It might
cause some headers to silently fail to be added for example. Note
that it's tagged MINOR because a standard configuration cannot make
this case happen.
The fix should be backported to 1.5 and 1.4 though.
Cyril Bonté [Thu, 29 Jan 2015 23:07:07 +0000 (00:07 +0100)]
BUG/MINOR: checks: prevent http keep-alive with http-check expect
Sébastien Rohaut reported that string negation in http-check expect didn't
work as expected.
The misbehaviour is caused by responses with HTTP keep-alive. When the
condition is not met, haproxy awaits more data until the buffer is full or the
connection is closed, resulting in a check timeout when "timeout check" is
lower than the keep-alive timeout on the server side.
In order to avoid the issue, when a "http-check expect" is used, haproxy will
ask the server to disable keep-alive by automatically appending a
"Connection: close" header to the request.
Willy Tarreau [Thu, 29 Jan 2015 13:01:34 +0000 (14:01 +0100)]
BUG/MINOR: http: fix incorrect header value offset in replace-hdr/replace-value
The two http-req/http-resp actions "replace-hdr" and "replace-value" were
expecting exactly one space after the colon, which is wrong. It was causing
the first char not to be seen/modified when no space was present, and empty
headers not to be modified either. Instead of using name->len+2, we must use
ctx->val which points to the first character of the value even if there is
no value.
Willy Tarreau [Wed, 28 Jan 2015 18:03:21 +0000 (19:03 +0100)]
MEDIUM: init: continue to enforce SYSTEM_MAXCONN with auto settings if set
Commit d025648 ("MAJOR: init: automatically set maxconn and/or maxsslconn
when possible") resulted in a case where if enough memory is available,
a maxconn value larger than SYSTEM_MAXCONN could be computed, resulting
in possibly overflowing other systems resources (eg: kernel socket buffers,
conntrack entries, etc). Let's bound any automatic maxconn to SYSTEM_MAXCONN
if it is defined. Note that the value is set to DEFAULT_MAXCONN since
SYSTEM_MAXCONN forces DEFAULT_MAXCONN, thus it is not an error.
Godbach [Wed, 28 Jan 2015 09:36:16 +0000 (17:36 +0800)]
BUG/MINOR: parse: check the validity of size string in a more strict way
If a stick table is defined as below:
stick-table type ip size 50ka expire 300s
HAProxy will stop parsing size after passing through "50k" and return the value
directly. But such format string of size should not be valid. The patch checks
the next character to report error if any.
Willy Tarreau [Tue, 27 Jan 2015 14:12:13 +0000 (15:12 +0100)]
MEDIUM: samples: provide basic arithmetic and bitwise operators
This commit introduces a new category of converters. They are bitwise and
arithmetic operators which support performing basic operations on integers.
Some bitwise operations are supported (and, or, xor, cpl) and some arithmetic
operations are supported (add, sub, mul, div, mod, neg). Some comparators
are provided (odd, even, not, bool) which make it possible to report a match
without having to write an ACL.
The detailed list of new operators as they appear in the doc is :
add(<value>)
Adds <value> to the input value of type unsigned integer, and returns the
result as an unsigned integer.
and(<value>)
Performs a bitwise "AND" between <value> and the input value of type unsigned
integer, and returns the result as an unsigned integer.
bool
Returns a boolean TRUE if the input value of type unsigned integer is
non-null, otherwise returns FALSE. Used in conjunction with and(), it can be
used to report true/false for bit testing on input values (eg: verify the
presence of a flag).
cpl
Takes the input value of type unsigned integer, applies a twos-complement
(flips all bits) and returns the result as an unsigned integer.
div(<value>)
Divides the input value of type unsigned integer by <value>, and returns the
result as an unsigned integer. If <value> is null, the largest unsigned
integer is returned (typically 2^32-1).
even
Returns a boolean TRUE if the input value of type unsigned integer is even
otherwise returns FALSE. It is functionally equivalent to "not,and(1),bool".
mod(<value>)
Divides the input value of type unsigned integer by <value>, and returns the
remainder as an unsigned integer. If <value> is null, then zero is returned.
mul(<value>)
Multiplies the input value of type unsigned integer by <value>, and returns
the product as an unsigned integer. In case of overflow, the higher bits are
lost, leading to seemingly strange values.
neg
Takes the input value of type unsigned integer, computes the opposite value,
and returns the remainder as an unsigned integer. 0 is identity. This
operator is provided for reversed subtracts : in order to subtract the input
from a constant, simply perform a "neg,add(value)".
not
Returns a boolean FALSE if the input value of type unsigned integer is
non-null, otherwise returns TRUE. Used in conjunction with and(), it can be
used to report true/false for bit testing on input values (eg: verify the
absence of a flag).
odd
Returns a boolean TRUE if the input value of type unsigned integer is odd
otherwise returns FALSE. It is functionally equivalent to "and(1),bool".
or(<value>)
Performs a bitwise "OR" between <value> and the input value of type unsigned
integer, and returns the result as an unsigned integer.
sub(<value>)
Subtracts <value> from the input value of type unsigned integer, and returns
the result as an unsigned integer. Note: in order to subtract the input from
a constant, simply perform a "neg,add(value)".
xor(<value>)
Performs a bitwise "XOR" (exclusive OR) between <value> and the input value
of type unsigned integer, and returns the result as an unsigned integer.
Cyril Bonté [Sat, 24 Jan 2015 23:16:08 +0000 (00:16 +0100)]
MINOR: ssl: load certificates in alphabetical order
As reported by Raphaël Enrici, certificates loaded from a directory are loaded
in a non predictive order. If no certificate was first loaded from a file, it
can result in different behaviours when haproxy is used in cluster.
We can also imagine other cases which weren't met yet.
Instead of using readdir(), we can use scandir() and sort files alphabetically.
This will ensure a predictive behaviour.