There is a problem when an instance is marked "disabled". Its ports are
still bound but will not be unbound upon termination. This causes processes
to accumulate during soft restarts, and might even cause failures to restart
new ones due to the inability to bind to the same port.
The ideal solution would be to bind all ports at the end of the configuration
parsing. An acceptable workaround is to unbind all listeners of disabled
proxies. This is what the current patch does.
During a configuration reload, haproxy tried to pause all proxies.
Unfortunately, it also tried to pause backends, which would fail
and cause trouble to the new process since the port was still bound.
srv_dynamic_maxconn() is clearly documented as returning at least 1
possible connection under throttling. But the computation was wrong,
the minimum 1 was divided and got lost in case of very low maxconns.
Apply the MAX(1, max) before returning the result in order to ensure
that a newly appeared server will get some traffic.
(cherry picked from commit 819970098f134453c0934047b3bd3440b0996b55)
A bug was introduced during last queue management fix. If a server
connection fails, the allocated connection slot is released, but it
will be needed again after the turn-around. This also causes more
connections than expected to go to the server because it appears to
have less connections than real.
Many thanks to Rupert Fiasco, Mark Imbriaco, Cody Fauser, Brian
Gupta and Alexander Staubo for promptly providing configuration
and diagnosis elements to help reproduce this problem easily.
(cherry picked from commit 8262d8bd7fdb262c980bd70cb2931e51df07513f)
maintain_proxies could disable backends (p->maxconn == 0) which is
wrong (but apparently harmless). Add a check for p->maxconn == 0.
(cherry picked from commit d5382b4aaa099ce5ce2af5828bd4d6dc38e9e8ea)
(cherry picked from commit 2f9127b4b91de1ac685498e145f29342115bcb71)
If __fd_clo() was called on a file descriptor which was previously
disabled, it was not removed from the spec list. This apparently
could not happen on previous code because the TCP states prevented
this, but now it happens regularly. The effects are spec entries
stuck populated, leading to busy loops.
(cherry picked from commit 7a52a5c4680477272b2f34eaf5896b85746e6fd6)
(cherry picked from commit 116f4105d4fc6fbd8f2d0a139f691973332176de)
Due to a copy-paste typo, the client timeout was refreshed instead
of the server's when waiting for server response. This means that
the server's timeout remained eternity.
(cherry picked from commit 9f1f24bb7fb8ebd6b43b5fee1bda0afbdbcb768e)
(cherry picked from commit df82605d3e73573ae842a1ddaf418997bef33274)
The INTBITS macro was found to be already defined on some platforms,
and to equal 32 (while INTBITS was 5 here). Due to pure luck, there
was no declaration conflict, but it's nonetheless a problem to fix.
Looking at the code showed that this macro was only used for left
shifts and nothing else anymore. So the replacement is obvious. The
new macro, BITS_PER_INT is more obviously correct.
(cherry picked from commit 177e2b012723ef65c6c7f850df3e6e0cd2cca2b4)
(cherry picked from commit 0e3e59b11f7926a570cfc98d8967b61098c91602)
I got a report that Intel C++ complains about the size of the
base64tab in base64.c. Setting it to 65 chars to allow for the
trailing zero fixes the problem.
(cherry picked from commit 69e989ccbcc1d5cbb623493d6c9cca169fb36ff6)
(cherry picked from commit 66c9f287c2c2d016eb12cb3ab12cd80b5c225f5e)
The buffer read timeouts were not reset when stats were produced. This
caused unneeded wakeups.
(cherry picked from commit 284c7b319566a66d5b742c905072175aac6445e1)
(cherry picked from commit 80f35306e97e8ae762f81a178c8c225b5bbac91e)
A copy-paste typo made use_backend not correctly consider the "unless"
case, depending on the previous "block" rule.
(cherry picked from commit a8cfa34a9c011cecfaedfaf7d91de3e5f7f004a0)
The dequeuing logic was completely wrong. First, a task was assigned
to all servers to process the queue, but this task was never scheduled
and was only woken up on session free. Second, there was no reservation
of server entries when a task was assigned a server. This means that
as long as the task was not connected to the server, its presence was
not accounted for. This was causing trouble when detecting whether or
not a server had reached maxconn. Third, during a redispatch, a session
could lose its place at the server's and get blocked because another
session at the same moment would have stolen the entry. Fourth, the
redispatch option did not work when maxqueue was reached for a server,
and it was not possible to do so without indefinitely hanging a session.
The root cause of all those problems was the lack of pre-reservation of
connections at the server's, and the lack of tracking of servers during
a redispatch. Everything relied on combinations of flags which could
appear similarly in quite distinct situations.
This patch is a major rework but there was no other solution, as the
internal logic was deeply flawed. The resulting code is cleaner, more
understandable, uses less magics and is overall more robust.
As an added bonus, "option redispatch" now works when maxqueue has
been reached on a server.
The reported queue position in the logs was 0 for the first pending request
in the queue, which is wrong because it means that one request will have to
be completed before the queued one may execute. It caused the undesired side
effect that 0/0 was reported when either 0 or 1 request was pending in the
queue. Thus, we have to increment the queue size before reporting the value.
When a server terminates a connection, the next session in its
own queue was immediately processed. Because of this, if all
server queues are always filled, then no new anonymous request
will be processed. Consider oldest request between global and
server queues to choose from which to pick the request.
An improvement over this will consist in adding a configurable
offset when comparing expiration dates, so that cookie-less
requests can get either less or more priority.
Under some circumstances, a task may already lie in the run queue
(eg: inter-task wakeup). It is disastrous to wait for an event in
this case because some processing gets delayed.
If too many events are set for spec I/O, those ones can starve the
polled events. Experiments show that when polled events starve, they
quickly turn into spec I/O, making the situation even worse. While
we can reduce the number of polled events processed at once, we
cannot do this on speculative events because most of them are new
ones (avg 2/3 new - 1/3 old from experiments).
The solution against this problem relies on those two factors :
1) one FD registered as a spec event cannot be polled at the same time
2) even during very high loads, we will almost never be interested in
simultaneous read and write streaming on the same FD.
The first point implies that during starvation, we will not have more than
half of our FDs in the poll list, otherwise it means there is less than that
in the spec list, implying there is no starvation.
The second point implies that we're statically only interested in half of
the maximum number of file descriptors at once, because we will unlikely
have simultaneous read and writes for a same buffer during long periods.
So, if we make it possible to drain maxsock/2/2 during peak loads, then we
can ensure that there will be no starvation effect. This means that we must
always allocate maxsock/4 events for the poller.
Last, sepoll uses an optimization consisting in reducing the number of calls
to epoll_wait() to once every too polls. However, when dealing with many
spec events, we can wait very long and skipping epoll_wait() every second
time increases latency. For this reason, we try to detect if we are beyond
a reasonable limit and stop doing so at this stage.
I noticed it was possible to get truncated http/csv stats. Sometimes.
Usually the problem disappeared as fast as it appeared, but once it
happend that my http-stats page was truncated for about one hour.
It was quite weird as it happened independently for csv and http
output and it took me some time to track & fix this bug.
Both buffer_write & buffer_write_chunk used to return 0 in two
situations: is case of success or where there was exactly 0 bytes
left. The first one is intentional but I believe the second one
is not as it was not possible to distinguish between successful
write and unsuccessful one, which means that if the buffer was 100%
filled, it was never flushed and it was not possible to write
more data.
This patch fixes this problem.
Due to the way the stats socket work, it was not possible to
maintain the information related to the command entered, so
after filling a whole buffer, the request was lost and it was
considered that there was nothing to write anymore.
The major reason was that some flags were passed directly
during the first call to stats_dump_raw() instead of being
stored persistently in the session.
To definitely fix this problem, flags were added to the stats
member of the session structure.
A second problem appeared. When the stats were produced, a first
call to client_retnclose() was performed, then one or multiple
subsequent calls to buffer_write_chunks() were done. But once the
stats buffer was full and a reschedule operated, the buffer was
flushed, the write flag cleared from the buffer and nothing was
done to re-arm it.
For this reason, a check was added in the proto_uxst_stats()
function in order to re-call the client FSM when data were added
by stats_dump_raw(). Finally, the whole unix stats dump FSM was
rewritten to avoid all the magics it depended on. It is now
simpler and looks more like the HTTP one.
Commit 3168223a7b33a1d5aad1e11b8f2ad917645d7f27 broke option
"allbackups" in roundrobin mode due to an erroneous structure
member replacement in backend.c. The PR_O_USE_ALL_BK flag was
not tested in the right member anymore.
This bug uncoverred another one, by which all backup servers would
be used whatever the option's value, if all of them had been seen
as simultaneously failed at one moment.
This patch fixes the two stupid errors. Correctness has been tested
using the test-fwrr.cfg config example.
(cherry picked from commit f4cca45b5e6c6ed88a0062cf92ae57e01405ab12)
Matt Farnsworth reported a memory leak in str2sun() in case a too large
socket path is passed. The bug is very minor because it only happens
once during config parsing, but has to be fixed nevertheless. The patch
Matt provided could even be improved by completely removing the useless
strdup() in this function.
(cherry picked from commit caf720d3ff7758273278aecab26bb7624ec2f555)
Commit 98937b875798e10fac671d109355cde29d2a411a while fixing
one bug introduced another one. With "retries 4" and
"option redispatch" haproxy tries to connect 4 times to
one server server and 1 time to a second one. However
logs showed 5 connections to the first server (the
last one was counted twice) and 2 to the second.
This patch also fixes srv->retries and be->retries increments.
Now I get: 3 retries and 1 error in a first server (4 cum_sess)
and 1 error in a second server (1 cum_sess) with:
retries 4
option redispatch
and: 4 retries and 1 error (5 cum_sess) with:
retries 4
So, the number of connections effectively served by a server is:
srv->cum_sess - srv->failed_conns - srv->retries
(cherry picked from commit 626a19b66f769a87e7c995267ccedf14149e03b3)
We've been trying to use the latest release (1.3.14.2) of haproxy to do
sticky sessions. Cookie insertion is not an option for us, although we
would much rather use it, as we are trying to work around a problem where
cookies are unreliable. The appsession functionality only partially worked
(it wouldn't read the session id out of a query string) until we made the
following code change to the get_srv_from_appsession function in
proto_http.c.
(cherry picked from commit 6d0b1fac23517f16b3972b529ea41718b3643c9f)
An unknown keyword was always reported in section "listen" for any
section type (defaults, listen, frontend, backend, ...).
(cherry picked from commit 6daf34352f325699efa8f731e5525275523786b9)
Commit 8b3977ffe36190e45fb974bf813bfbd935ac99a1 removed "t->logs.bytes_in = 0;"
but instead it should change it into "t->logs.bytes_out = 0;" as since
583bc966064e248771e75c253dddec7351afd8a2 counters are incremented not set.
It should be incremented in session_process_counters while sending data to a
client:
bytes = s->rep->total - s->logs.bytes_out;
s->logs.bytes_out = s->rep->total;
However, if we increment (set) s->logs.bytes_out while processing
"logasap", statistics get wrong values added for headers: 0 or even
negative if haproxy adds some headers itself.
To test it, please enable logasap and download one empty file and look at
stats. Without my fix information available on that page are invalid, for
example:
# pxname,svname,qcur,qmax,scur,smax,slim,stot,bin,bout,dreq,dresp,ereq,econ,eresp,wretr,wredis,status,weight,act,bck,chkfail,chkdown,lastchg,downtime,qlimit,pid,iid,sid,throttle,lbtot,
www,b,0,0,0,1,,1,24,-92,,0,,0,0,0,,UP,1,1,0,0,0,3121,0,,1,2,1,,1,
www,BACKEND,0,0,0,1,0,1,24,-92,0,0,,0,0,0,0,UP,1,1,0,,0,3121,0,,1,2,0,,1,
Due to the way Linux delivers EPOLLIN and EPOLLHUP, a closed connection
received after some server data sometimes results in truncated responses
if the client disconnects before server starts to respond. The reason
is that the EPOLLHUP flag is processed as an indication of end of
transfer while some data may remain in the system's socket buffers.
This problem could only be triggered with sepoll, although nothing should
prevent it from happening with normal epoll. In fact, the work factoring
performed by sepoll increases the risk that this bug appears.
The fix consists in making FD_POLL_HUP and FD_POLL_ERR sticky and that
they are only checked if FD_POLL_IN is not set, meaning that we have
read all pending data.
That way, the problem is definitely fixed and sepoll still remains about
17% faster than epoll since it can take into account all information
returned by the kernel.
a copy-paste typo was present in the reconnection code responsible
for respatching. The client's FSM would not be re-evaluated if an
error occurred. It looks harmless but better fix it.
It was abnormal to see more connect errors than connect attempts.
This was caused by the fact that the server's connection count was
not incremented for failed connect() attempts.
Now the per-server connections are correctly incremented for each
connect() attempt. This includes the retries too. The number of
connections effectively served by a server will then be :
srv->cum_sess - srv->errors - srv->warnings
In order to offer DoS protection, it may be required to lower the maximum
accepted time to receive a complete HTTP request without affecting the client
timeout. This helps protecting against established connections on which
nothing is sent. The client timeout cannot offer a good protection against
this abuse because it is an inactivity timeout, which means that if the
attacker sends one character every now and then, the timeout will not
trigger. With the HTTP request timeout, no matter what speed the client
types, the request will be aborted if it does not complete in time.
This new parameter makes it possible to override the default
number of consecutive incoming connections which can be
accepted on a socket. By default it is not limited on single
process mode, and limited to 8 in multi-process mode.
Add the "backlog" parameter to frontends, to give hints to
the system about the approximate listen backlog desired size.
In order to protect against SYN flood attacks, one solution is
to increase the system's SYN backlog size. Depending on the
system, sometimes it is just tunable via a system parameter,
sometimes it is not adjustable at all, and sometimes the system
relies on hints given by the application at the time of the
listen() syscall. By default, HAProxy passes the frontend's
maxconn value to the listen() syscall. On systems which can
make use of this value, it can sometimes be useful to be able
to specify a different value, hence this backlog parameter.
It is sometimes required to know some informations such as the
process uptime when consulting statistics. This patch adds the
"show info" command to query those informations on the UNIX
socket.
The build process was getting annoying under some conditions,
especially on platforms which are used to set CFLAGS, as well
as those which set a lot of complex defines. The new Makefile
takes care of this situation by not mixing TARGET, CPU and user
values, and by making privileging the pre-setting of common
variables with the ability to override them.
Now CFLAGS and LDFLAGS are set by default and may be overridden
without the risk of breaking useful defines. Options are better
dealt with, and as a bonus, it was possible to merge the FreeBSD
and OpenBSD targets into the common GNU Makefile.
The report of build options by "haproxy -vv" has been slightly
adapted to the new mode. Options implied by architecture are not
reported, only user-specified options are. It is also possible to
add options which will not be reported in order not to mangle the
output when specifying dirty informations such as URLs...
The Makefile was copiously documented and it should be easier to
build for any target now. Backwards compatibility with older
build processes was kept, and warnings are emitted for deprecated
build options.
This patch adds a possibility to invert most of available options by
introducing the "no" keyword, available as an additional prefix.
If it is found arguments are shifted left and an additional flag (inv)
is set.
It allows to use all options from a current defaults section, except
the selected ones, for example:
-- cut here --
defaults
contimeout 4200
clitimeout 50000
srvtimeout 40000
option contstats
listen stats 1.2.3.4:80
no option contstats
-- cut here --
Currenly inversion works only with the "option" keyword.
The patch also moves last_checks calculation at the end of the readcfgfile()
function and changes "PR_O_FORCE_CLO | PR_O_HTTP_CLOSE" into "PR_O_FORCE_CLO"
in cfg_opts so it is possible to invert forceclose without breaking httpclose
(and vice versa) and to invert tcpsplice in one proxy but to keep a proper
last_checks value when tcpsplice is used in another proxy. Now, the code
checks for PR_O_FORCE_CLO everywhere it checks for PR_O_HTTP_CLOSE.
I also decided to depreciate "redisp" and "redispatch" keywords as it is IMHO
better to use "option redispatch" which can be inverted.
Some useful documentation were added and at the same time I sorted
(alfabetically) all valid options both in the code and the documentation.
The error check in return of start_proxies checked for exact ERR_RETRYABLE
but did not consider the return as a bit field. The function returned both
ERR_RETRYABLE and ERR_ALERT, hence the problem.
Solaris, as well as many other unixes doesn't know about sun_len
for UNIX domain sockets. It does not honnor the __SOCKADDR_COMMON
macro either. After looking at MacOS-X man (which is the same as
BSD man), OpenBSD man, and examples on the net, it appears that
those which support sun_len do not actually use it, or at least
ignore it as long as it's zero. Since all the sockaddr structures
are zeroed prior to being filled, it causes no problem not to set
sun_len, and this fixes build on other platforms.
Another problem on Solaris was that the "sun" name is already
defined as a macro returning a number, so it was necessary to
rename it.
The code in haproxy-1.3.13.1 only supports syslogging to an internet
address. The attached patch:
- Adds support for syslogging to a UNIX domain socket (e.g., /dev/log).
If the address field begins with '/' (absolute file path), then
AF_UNIX is used to construct the socket. Otherwise, AF_INET is used.
- Achieves clean single-source build on both Mac OS X and Linux
(sockaddr_in.sin_len and sockaddr_un.sun_len field aren't always present).
For handling sendto() failures in send_log(), it appears that the existing
code is fine (no need to close/recreate socket) for both UDP and UNIX-domain
syslog server. So I left things alone (did not close/recreate socket).
Closing/recreating socket after each failure would also work, but would lead
to increased amount of unnecessary socket creation/destruction if syslog is
temporarily unavailable for some reason (especially for verbose loggers).
Please consider this patch for inclusion into the upstream haproxy codebase.
One user reported that an indicator was missing in the statistics:
the number of times each server was selected by load balancing. It
is in fact the total number of sessions assigned to a server by the
load balancing algorithm. It should directly reflect the weight for
"fair" algorithms such as round-robin, since it will not account for
persistant connections.
It should help a lot tuning each server's weight depending on the
load it receives.
Because of a divide, it was possible to have a null weight during
a slowstart, which is pretty annoying, especially with a single
server and a long slowstart.
Also, fix the way we report the values in the stats page to avoid
confusion.
A new "timeout" keyword replaces old "{con|cli|srv}timeout", and
provides the ability to independantly set the following timeouts :
- client
- tarpit
- queue
- connect
- server
- appsession
Additionally, the "clitimeout", "contimeout" and "srvtimeout" values
are supported but deprecated. No warning is emitted yet when they are
used since the option is very new.
Other timeouts should follow soon now.
