2006-06-15 19:48:13 +00:00
|
|
|
/*
|
REORG/MAJOR: session: rename the "session" entity to "stream"
With HTTP/2, we'll have to support multiplexed streams. A stream is in
fact the largest part of what we currently call a session, it has buffers,
logs, etc.
In order to catch any error, this commit removes any reference to the
struct session and tries to rename most "session" occurrences in function
names to "stream" and "sess" to "strm" when that's related to a session.
The files stream.{c,h} were added and session.{c,h} removed.
The session will be reintroduced later and a few parts of the stream
will progressively be moved overthere. It will more or less contain
only what we need in an embryonic session.
Sample fetch functions and converters will have to change a bit so
that they'll use an L5 (session) instead of what's currently called
"L4" which is in fact L6 for now.
Once all changes are completed, we should see approximately this :
L7 - http_txn
L6 - stream
L5 - session
L4 - connection | applet
There will be at most one http_txn per stream, and a same session will
possibly be referenced by multiple streams. A connection will point to
a session and to a stream. The session will hold all the information
we need to keep even when we don't yet have a stream.
Some more cleanup is needed because some code was already far from
being clean. The server queue management still refers to sessions at
many places while comments talk about connections. This will have to
be cleaned up once we have a server-side connection pool manager.
Stream flags "SN_*" still need to be renamed, it doesn't seem like
any of them will need to move to the session.
2015-04-02 22:22:06 +00:00
|
|
|
* include/proto/stream.h
|
|
|
|
|
* This file defines everything related to streams.
|
2010-06-01 15:45:26 +00:00
|
|
|
*
|
|
|
|
|
* Copyright (C) 2000-2010 Willy Tarreau - w@1wt.eu
|
|
|
|
|
*
|
|
|
|
|
* This library is free software; you can redistribute it and/or
|
|
|
|
|
* modify it under the terms of the GNU Lesser General Public
|
|
|
|
|
* License as published by the Free Software Foundation, version 2.1
|
|
|
|
|
* exclusively.
|
|
|
|
|
*
|
|
|
|
|
* This library is distributed in the hope that it will be useful,
|
|
|
|
|
* but WITHOUT ANY WARRANTY; without even the implied warranty of
|
|
|
|
|
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
|
|
|
|
|
* Lesser General Public License for more details.
|
|
|
|
|
*
|
|
|
|
|
* You should have received a copy of the GNU Lesser General Public
|
|
|
|
|
* License along with this library; if not, write to the Free Software
|
|
|
|
|
* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
|
|
|
|
|
*/
|
2006-06-15 19:48:13 +00:00
|
|
|
|
REORG/MAJOR: session: rename the "session" entity to "stream"
With HTTP/2, we'll have to support multiplexed streams. A stream is in
fact the largest part of what we currently call a session, it has buffers,
logs, etc.
In order to catch any error, this commit removes any reference to the
struct session and tries to rename most "session" occurrences in function
names to "stream" and "sess" to "strm" when that's related to a session.
The files stream.{c,h} were added and session.{c,h} removed.
The session will be reintroduced later and a few parts of the stream
will progressively be moved overthere. It will more or less contain
only what we need in an embryonic session.
Sample fetch functions and converters will have to change a bit so
that they'll use an L5 (session) instead of what's currently called
"L4" which is in fact L6 for now.
Once all changes are completed, we should see approximately this :
L7 - http_txn
L6 - stream
L5 - session
L4 - connection | applet
There will be at most one http_txn per stream, and a same session will
possibly be referenced by multiple streams. A connection will point to
a session and to a stream. The session will hold all the information
we need to keep even when we don't yet have a stream.
Some more cleanup is needed because some code was already far from
being clean. The server queue management still refers to sessions at
many places while comments talk about connections. This will have to
be cleaned up once we have a server-side connection pool manager.
Stream flags "SN_*" still need to be renamed, it doesn't seem like
any of them will need to move to the session.
2015-04-02 22:22:06 +00:00
|
|
|
#ifndef _PROTO_STREAM_H
|
|
|
|
|
#define _PROTO_STREAM_H
|
2006-06-26 00:48:02 +00:00
|
|
|
|
2006-06-29 16:54:54 +00:00
|
|
|
#include <common/config.h>
|
2007-05-13 17:43:47 +00:00
|
|
|
#include <common/memory.h>
|
REORG/MAJOR: session: rename the "session" entity to "stream"
With HTTP/2, we'll have to support multiplexed streams. A stream is in
fact the largest part of what we currently call a session, it has buffers,
logs, etc.
In order to catch any error, this commit removes any reference to the
struct session and tries to rename most "session" occurrences in function
names to "stream" and "sess" to "strm" when that's related to a session.
The files stream.{c,h} were added and session.{c,h} removed.
The session will be reintroduced later and a few parts of the stream
will progressively be moved overthere. It will more or less contain
only what we need in an embryonic session.
Sample fetch functions and converters will have to change a bit so
that they'll use an L5 (session) instead of what's currently called
"L4" which is in fact L6 for now.
Once all changes are completed, we should see approximately this :
L7 - http_txn
L6 - stream
L5 - session
L4 - connection | applet
There will be at most one http_txn per stream, and a same session will
possibly be referenced by multiple streams. A connection will point to
a session and to a stream. The session will hold all the information
we need to keep even when we don't yet have a stream.
Some more cleanup is needed because some code was already far from
being clean. The server queue management still refers to sessions at
many places while comments talk about connections. This will have to
be cleaned up once we have a server-side connection pool manager.
Stream flags "SN_*" still need to be renamed, it doesn't seem like
any of them will need to move to the session.
2015-04-02 22:22:06 +00:00
|
|
|
#include <types/stream.h>
|
MAJOR: session: only wake up as many sessions as available buffers permit
We've already experimented with three wake up algorithms when releasing
buffers : the first naive one used to wake up far too many sessions,
causing many of them not to get any buffer. The second approach which
was still in use prior to this patch consisted in waking up either 1
or 2 sessions depending on the number of FDs we had released. And this
was still inaccurate. The third one tried to cover the accuracy issues
of the second and took into consideration the number of FDs the sessions
would be willing to use, but most of the time we ended up waking up too
many of them for nothing, or deadlocking by lack of buffers.
This patch completely removes the need to allocate two buffers at once.
Instead it splits allocations into critical and non-critical ones and
implements a reserve in the pool for this. The deadlock situation happens
when all buffers are be allocated for requests pending in a maxconn-limited
server queue, because then there's no more way to allocate buffers for
responses, and these responses are critical to release the servers's
connection in order to release the pending requests. In fact maxconn on
a server creates a dependence between sessions and particularly between
oldest session's responses and latest session's requests. Thus, it is
mandatory to get a free buffer for a response in order to release a
server connection which will permit to release a request buffer.
Since we definitely have non-symmetrical buffers, we need to implement
this logic in the buffer allocation mechanism. What this commit does is
implement a reserve of buffers which can only be allocated for responses
and that will never be allocated for requests. This is made possible by
the requester indicating how much margin it wants to leave after the
allocation succeeds. Thus it is a cooperative allocation mechanism : the
requester (process_session() in general) prefers not to get a buffer in
order to respect other's need for response buffers. The session management
code always knows if a buffer will be used for requests or responses, so
that is not difficult :
- either there's an applet on the initiator side and we really need
the request buffer (since currently the applet is called in the
context of the session)
- or we have a connection and we really need the response buffer (in
order to support building and sending an error message back)
This reserve ensures that we don't take all allocatable buffers for
requests waiting in a queue. The downside is that all the extra buffers
are really allocated to ensure they can be allocated. But with small
values it is not an issue.
With this change, we don't observe any more deadlocks even when running
with maxconn 1 on a server under severely constrained memory conditions.
The code becomes a bit tricky, it relies on the scheduler's run queue to
estimate how many sessions are already expected to run so that it doesn't
wake up everyone with too few resources. A better solution would probably
consist in having two queues, one for urgent requests and one for normal
requests. A failed allocation for a session dealing with an error, a
connection event, or the need for a response (or request when there's an
applet on the left) would go to the urgent request queue, while other
requests would go to the other queue. Urgent requests would be served
from 1 entry in the pool, while the regular ones would be served only
according to the reserve. Despite not yet having this, it works
remarkably well.
This mechanism is quite efficient, we don't perform too many wake up calls
anymore. For 1 million sessions elapsed during massive memory contention,
we observe about 4.5M calls to process_session() compared to 4.0M without
memory constraints. Previously we used to observe up to 16M calls, which
rougly means 12M failures.
During a test run under high memory constraints (limit enforced to 27 MB
instead of the 58 MB normally needed), performance used to drop by 53% prior
to this patch. Now with this patch instead it *increases* by about 1.5%.
The best effect of this change is that by limiting the memory usage to about
2/3 to 3/4 of what is needed by default, it's possible to increase performance
by up to about 18% mainly due to the fact that pools are reused more often
and remain hot in the CPU cache (observed on regular HTTP traffic with 20k
objects, buffers.limit = maxconn/10, buffers.reserve = limit/2).
Below is an example of scenario which used to cause a deadlock previously :
- connection is received
- two buffers are allocated in process_session() then released
- one is allocated when receiving an HTTP request
- the second buffer is allocated then released in process_session()
for request parsing then connection establishment.
- poll() says we can send, so the request buffer is sent and released
- process session gets notified that the connection is now established
and allocates two buffers then releases them
- all other sessions do the same till one cannot get the request buffer
without hitting the margin
- and now the server responds. stream_interface allocates the response
buffer and manages to get it since it's higher priority being for a
response.
- but process_session() cannot allocate the request buffer anymore
=> We could end up with all buffers used by responses so that none may
be allocated for a request in process_session().
When the applet processing leaves the session context, the test will have
to be changed so that we always allocate a response buffer regardless of
the left side (eg: H2->H1 gateway). A final improvement would consists in
being able to only retry the failed I/O operation without waking up a
task, but to date all experiments to achieve this have proven not to be
reliable enough.
2014-11-27 00:11:56 +00:00
|
|
|
#include <proto/fd.h>
|
2010-06-20 09:19:22 +00:00
|
|
|
#include <proto/freq_ctr.h>
|
2010-06-14 19:04:55 +00:00
|
|
|
#include <proto/stick_table.h>
|
MAJOR: session: only wake up as many sessions as available buffers permit
We've already experimented with three wake up algorithms when releasing
buffers : the first naive one used to wake up far too many sessions,
causing many of them not to get any buffer. The second approach which
was still in use prior to this patch consisted in waking up either 1
or 2 sessions depending on the number of FDs we had released. And this
was still inaccurate. The third one tried to cover the accuracy issues
of the second and took into consideration the number of FDs the sessions
would be willing to use, but most of the time we ended up waking up too
many of them for nothing, or deadlocking by lack of buffers.
This patch completely removes the need to allocate two buffers at once.
Instead it splits allocations into critical and non-critical ones and
implements a reserve in the pool for this. The deadlock situation happens
when all buffers are be allocated for requests pending in a maxconn-limited
server queue, because then there's no more way to allocate buffers for
responses, and these responses are critical to release the servers's
connection in order to release the pending requests. In fact maxconn on
a server creates a dependence between sessions and particularly between
oldest session's responses and latest session's requests. Thus, it is
mandatory to get a free buffer for a response in order to release a
server connection which will permit to release a request buffer.
Since we definitely have non-symmetrical buffers, we need to implement
this logic in the buffer allocation mechanism. What this commit does is
implement a reserve of buffers which can only be allocated for responses
and that will never be allocated for requests. This is made possible by
the requester indicating how much margin it wants to leave after the
allocation succeeds. Thus it is a cooperative allocation mechanism : the
requester (process_session() in general) prefers not to get a buffer in
order to respect other's need for response buffers. The session management
code always knows if a buffer will be used for requests or responses, so
that is not difficult :
- either there's an applet on the initiator side and we really need
the request buffer (since currently the applet is called in the
context of the session)
- or we have a connection and we really need the response buffer (in
order to support building and sending an error message back)
This reserve ensures that we don't take all allocatable buffers for
requests waiting in a queue. The downside is that all the extra buffers
are really allocated to ensure they can be allocated. But with small
values it is not an issue.
With this change, we don't observe any more deadlocks even when running
with maxconn 1 on a server under severely constrained memory conditions.
The code becomes a bit tricky, it relies on the scheduler's run queue to
estimate how many sessions are already expected to run so that it doesn't
wake up everyone with too few resources. A better solution would probably
consist in having two queues, one for urgent requests and one for normal
requests. A failed allocation for a session dealing with an error, a
connection event, or the need for a response (or request when there's an
applet on the left) would go to the urgent request queue, while other
requests would go to the other queue. Urgent requests would be served
from 1 entry in the pool, while the regular ones would be served only
according to the reserve. Despite not yet having this, it works
remarkably well.
This mechanism is quite efficient, we don't perform too many wake up calls
anymore. For 1 million sessions elapsed during massive memory contention,
we observe about 4.5M calls to process_session() compared to 4.0M without
memory constraints. Previously we used to observe up to 16M calls, which
rougly means 12M failures.
During a test run under high memory constraints (limit enforced to 27 MB
instead of the 58 MB normally needed), performance used to drop by 53% prior
to this patch. Now with this patch instead it *increases* by about 1.5%.
The best effect of this change is that by limiting the memory usage to about
2/3 to 3/4 of what is needed by default, it's possible to increase performance
by up to about 18% mainly due to the fact that pools are reused more often
and remain hot in the CPU cache (observed on regular HTTP traffic with 20k
objects, buffers.limit = maxconn/10, buffers.reserve = limit/2).
Below is an example of scenario which used to cause a deadlock previously :
- connection is received
- two buffers are allocated in process_session() then released
- one is allocated when receiving an HTTP request
- the second buffer is allocated then released in process_session()
for request parsing then connection establishment.
- poll() says we can send, so the request buffer is sent and released
- process session gets notified that the connection is now established
and allocates two buffers then releases them
- all other sessions do the same till one cannot get the request buffer
without hitting the margin
- and now the server responds. stream_interface allocates the response
buffer and manages to get it since it's higher priority being for a
response.
- but process_session() cannot allocate the request buffer anymore
=> We could end up with all buffers used by responses so that none may
be allocated for a request in process_session().
When the applet processing leaves the session context, the test will have
to be changed so that we always allocate a response buffer regardless of
the left side (eg: H2->H1 gateway). A final improvement would consists in
being able to only retry the failed I/O operation without waking up a
task, but to date all experiments to achieve this have proven not to be
reliable enough.
2014-11-27 00:11:56 +00:00
|
|
|
#include <proto/task.h>
|
2006-06-15 19:48:13 +00:00
|
|
|
|
2017-11-24 16:34:44 +00:00
|
|
|
extern struct pool_head *pool_head_stream;
|
REORG/MAJOR: session: rename the "session" entity to "stream"
With HTTP/2, we'll have to support multiplexed streams. A stream is in
fact the largest part of what we currently call a session, it has buffers,
logs, etc.
In order to catch any error, this commit removes any reference to the
struct session and tries to rename most "session" occurrences in function
names to "stream" and "sess" to "strm" when that's related to a session.
The files stream.{c,h} were added and session.{c,h} removed.
The session will be reintroduced later and a few parts of the stream
will progressively be moved overthere. It will more or less contain
only what we need in an embryonic session.
Sample fetch functions and converters will have to change a bit so
that they'll use an L5 (session) instead of what's currently called
"L4" which is in fact L6 for now.
Once all changes are completed, we should see approximately this :
L7 - http_txn
L6 - stream
L5 - session
L4 - connection | applet
There will be at most one http_txn per stream, and a same session will
possibly be referenced by multiple streams. A connection will point to
a session and to a stream. The session will hold all the information
we need to keep even when we don't yet have a stream.
Some more cleanup is needed because some code was already far from
being clean. The server queue management still refers to sessions at
many places while comments talk about connections. This will have to
be cleaned up once we have a server-side connection pool manager.
Stream flags "SN_*" still need to be renamed, it doesn't seem like
any of them will need to move to the session.
2015-04-02 22:22:06 +00:00
|
|
|
extern struct list streams;
|
2007-05-13 17:43:47 +00:00
|
|
|
|
2012-11-19 15:10:32 +00:00
|
|
|
extern struct data_cb sess_conn_cb;
|
|
|
|
|
|
2017-08-28 14:22:54 +00:00
|
|
|
struct stream *stream_new(struct session *sess, enum obj_type *origin);
|
2017-09-13 16:30:23 +00:00
|
|
|
int stream_create_from_cs(struct conn_stream *cs);
|
2006-06-15 19:48:13 +00:00
|
|
|
|
2007-05-13 17:43:47 +00:00
|
|
|
/* perform minimal intializations, report 0 in case of error, 1 if OK. */
|
REORG/MAJOR: session: rename the "session" entity to "stream"
With HTTP/2, we'll have to support multiplexed streams. A stream is in
fact the largest part of what we currently call a session, it has buffers,
logs, etc.
In order to catch any error, this commit removes any reference to the
struct session and tries to rename most "session" occurrences in function
names to "stream" and "sess" to "strm" when that's related to a session.
The files stream.{c,h} were added and session.{c,h} removed.
The session will be reintroduced later and a few parts of the stream
will progressively be moved overthere. It will more or less contain
only what we need in an embryonic session.
Sample fetch functions and converters will have to change a bit so
that they'll use an L5 (session) instead of what's currently called
"L4" which is in fact L6 for now.
Once all changes are completed, we should see approximately this :
L7 - http_txn
L6 - stream
L5 - session
L4 - connection | applet
There will be at most one http_txn per stream, and a same session will
possibly be referenced by multiple streams. A connection will point to
a session and to a stream. The session will hold all the information
we need to keep even when we don't yet have a stream.
Some more cleanup is needed because some code was already far from
being clean. The server queue management still refers to sessions at
many places while comments talk about connections. This will have to
be cleaned up once we have a server-side connection pool manager.
Stream flags "SN_*" still need to be renamed, it doesn't seem like
any of them will need to move to the session.
2015-04-02 22:22:06 +00:00
|
|
|
int init_stream();
|
2006-06-26 00:48:02 +00:00
|
|
|
|
2015-04-02 23:14:29 +00:00
|
|
|
/* kill a stream and set the termination flags to <why> (one of SF_ERR_*) */
|
REORG/MAJOR: session: rename the "session" entity to "stream"
With HTTP/2, we'll have to support multiplexed streams. A stream is in
fact the largest part of what we currently call a session, it has buffers,
logs, etc.
In order to catch any error, this commit removes any reference to the
struct session and tries to rename most "session" occurrences in function
names to "stream" and "sess" to "strm" when that's related to a session.
The files stream.{c,h} were added and session.{c,h} removed.
The session will be reintroduced later and a few parts of the stream
will progressively be moved overthere. It will more or less contain
only what we need in an embryonic session.
Sample fetch functions and converters will have to change a bit so
that they'll use an L5 (session) instead of what's currently called
"L4" which is in fact L6 for now.
Once all changes are completed, we should see approximately this :
L7 - http_txn
L6 - stream
L5 - session
L4 - connection | applet
There will be at most one http_txn per stream, and a same session will
possibly be referenced by multiple streams. A connection will point to
a session and to a stream. The session will hold all the information
we need to keep even when we don't yet have a stream.
Some more cleanup is needed because some code was already far from
being clean. The server queue management still refers to sessions at
many places while comments talk about connections. This will have to
be cleaned up once we have a server-side connection pool manager.
Stream flags "SN_*" still need to be renamed, it doesn't seem like
any of them will need to move to the session.
2015-04-02 22:22:06 +00:00
|
|
|
void stream_shutdown(struct stream *stream, int why);
|
2011-06-08 00:19:07 +00:00
|
|
|
|
REORG/MAJOR: session: rename the "session" entity to "stream"
With HTTP/2, we'll have to support multiplexed streams. A stream is in
fact the largest part of what we currently call a session, it has buffers,
logs, etc.
In order to catch any error, this commit removes any reference to the
struct session and tries to rename most "session" occurrences in function
names to "stream" and "sess" to "strm" when that's related to a session.
The files stream.{c,h} were added and session.{c,h} removed.
The session will be reintroduced later and a few parts of the stream
will progressively be moved overthere. It will more or less contain
only what we need in an embryonic session.
Sample fetch functions and converters will have to change a bit so
that they'll use an L5 (session) instead of what's currently called
"L4" which is in fact L6 for now.
Once all changes are completed, we should see approximately this :
L7 - http_txn
L6 - stream
L5 - session
L4 - connection | applet
There will be at most one http_txn per stream, and a same session will
possibly be referenced by multiple streams. A connection will point to
a session and to a stream. The session will hold all the information
we need to keep even when we don't yet have a stream.
Some more cleanup is needed because some code was already far from
being clean. The server queue management still refers to sessions at
many places while comments talk about connections. This will have to
be cleaned up once we have a server-side connection pool manager.
Stream flags "SN_*" still need to be renamed, it doesn't seem like
any of them will need to move to the session.
2015-04-02 22:22:06 +00:00
|
|
|
void stream_process_counters(struct stream *s);
|
|
|
|
|
void sess_change_server(struct stream *sess, struct server *newsrv);
|
2018-05-25 12:04:04 +00:00
|
|
|
struct task *process_stream(struct task *t, void *context, unsigned short state);
|
REORG/MAJOR: session: rename the "session" entity to "stream"
With HTTP/2, we'll have to support multiplexed streams. A stream is in
fact the largest part of what we currently call a session, it has buffers,
logs, etc.
In order to catch any error, this commit removes any reference to the
struct session and tries to rename most "session" occurrences in function
names to "stream" and "sess" to "strm" when that's related to a session.
The files stream.{c,h} were added and session.{c,h} removed.
The session will be reintroduced later and a few parts of the stream
will progressively be moved overthere. It will more or less contain
only what we need in an embryonic session.
Sample fetch functions and converters will have to change a bit so
that they'll use an L5 (session) instead of what's currently called
"L4" which is in fact L6 for now.
Once all changes are completed, we should see approximately this :
L7 - http_txn
L6 - stream
L5 - session
L4 - connection | applet
There will be at most one http_txn per stream, and a same session will
possibly be referenced by multiple streams. A connection will point to
a session and to a stream. The session will hold all the information
we need to keep even when we don't yet have a stream.
Some more cleanup is needed because some code was already far from
being clean. The server queue management still refers to sessions at
many places while comments talk about connections. This will have to
be cleaned up once we have a server-side connection pool manager.
Stream flags "SN_*" still need to be renamed, it doesn't seem like
any of them will need to move to the session.
2015-04-02 22:22:06 +00:00
|
|
|
void default_srv_error(struct stream *s, struct stream_interface *si);
|
2010-06-14 19:04:55 +00:00
|
|
|
int parse_track_counters(char **args, int *arg,
|
|
|
|
|
int section_type, struct proxy *curpx,
|
|
|
|
|
struct track_ctr_prm *prm,
|
2012-05-08 17:47:01 +00:00
|
|
|
struct proxy *defpx, char **err);
|
2010-06-14 19:04:55 +00:00
|
|
|
|
REORG/MAJOR: session: rename the "session" entity to "stream"
With HTTP/2, we'll have to support multiplexed streams. A stream is in
fact the largest part of what we currently call a session, it has buffers,
logs, etc.
In order to catch any error, this commit removes any reference to the
struct session and tries to rename most "session" occurrences in function
names to "stream" and "sess" to "strm" when that's related to a session.
The files stream.{c,h} were added and session.{c,h} removed.
The session will be reintroduced later and a few parts of the stream
will progressively be moved overthere. It will more or less contain
only what we need in an embryonic session.
Sample fetch functions and converters will have to change a bit so
that they'll use an L5 (session) instead of what's currently called
"L4" which is in fact L6 for now.
Once all changes are completed, we should see approximately this :
L7 - http_txn
L6 - stream
L5 - session
L4 - connection | applet
There will be at most one http_txn per stream, and a same session will
possibly be referenced by multiple streams. A connection will point to
a session and to a stream. The session will hold all the information
we need to keep even when we don't yet have a stream.
Some more cleanup is needed because some code was already far from
being clean. The server queue management still refers to sessions at
many places while comments talk about connections. This will have to
be cleaned up once we have a server-side connection pool manager.
Stream flags "SN_*" still need to be renamed, it doesn't seem like
any of them will need to move to the session.
2015-04-02 22:22:06 +00:00
|
|
|
/* Update the stream's backend and server time stats */
|
|
|
|
|
void stream_update_time_stats(struct stream *s);
|
|
|
|
|
void stream_release_buffers(struct stream *s);
|
2014-06-17 10:19:18 +00:00
|
|
|
|
2015-04-03 11:53:24 +00:00
|
|
|
/* returns the session this stream belongs to */
|
|
|
|
|
static inline struct session *strm_sess(const struct stream *strm)
|
|
|
|
|
{
|
|
|
|
|
return strm->sess;
|
|
|
|
|
}
|
|
|
|
|
|
2015-04-04 00:10:38 +00:00
|
|
|
/* returns the frontend this stream was initiated from */
|
|
|
|
|
static inline struct proxy *strm_fe(const struct stream *strm)
|
|
|
|
|
{
|
|
|
|
|
return strm->sess->fe;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/* returns the listener this stream was initiated from */
|
|
|
|
|
static inline struct listener *strm_li(const struct stream *strm)
|
|
|
|
|
{
|
|
|
|
|
return strm->sess->listener;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/* returns a pointer to the origin of the session which created this stream */
|
|
|
|
|
static inline enum obj_type *strm_orig(const struct stream *strm)
|
|
|
|
|
{
|
|
|
|
|
return strm->sess->origin;
|
|
|
|
|
}
|
|
|
|
|
|
REORG/MAJOR: session: rename the "session" entity to "stream"
With HTTP/2, we'll have to support multiplexed streams. A stream is in
fact the largest part of what we currently call a session, it has buffers,
logs, etc.
In order to catch any error, this commit removes any reference to the
struct session and tries to rename most "session" occurrences in function
names to "stream" and "sess" to "strm" when that's related to a session.
The files stream.{c,h} were added and session.{c,h} removed.
The session will be reintroduced later and a few parts of the stream
will progressively be moved overthere. It will more or less contain
only what we need in an embryonic session.
Sample fetch functions and converters will have to change a bit so
that they'll use an L5 (session) instead of what's currently called
"L4" which is in fact L6 for now.
Once all changes are completed, we should see approximately this :
L7 - http_txn
L6 - stream
L5 - session
L4 - connection | applet
There will be at most one http_txn per stream, and a same session will
possibly be referenced by multiple streams. A connection will point to
a session and to a stream. The session will hold all the information
we need to keep even when we don't yet have a stream.
Some more cleanup is needed because some code was already far from
being clean. The server queue management still refers to sessions at
many places while comments talk about connections. This will have to
be cleaned up once we have a server-side connection pool manager.
Stream flags "SN_*" still need to be renamed, it doesn't seem like
any of them will need to move to the session.
2015-04-02 22:22:06 +00:00
|
|
|
/* Remove the refcount from the stream to the tracked counters, and clear the
|
2010-06-14 19:04:55 +00:00
|
|
|
* pointer to ensure this is only performed once. The caller is responsible for
|
2015-09-21 15:48:24 +00:00
|
|
|
* ensuring that the pointer is valid first. We must be extremely careful not
|
|
|
|
|
* to touch the entries we inherited from the session.
|
2010-06-14 19:04:55 +00:00
|
|
|
*/
|
REORG/MAJOR: session: rename the "session" entity to "stream"
With HTTP/2, we'll have to support multiplexed streams. A stream is in
fact the largest part of what we currently call a session, it has buffers,
logs, etc.
In order to catch any error, this commit removes any reference to the
struct session and tries to rename most "session" occurrences in function
names to "stream" and "sess" to "strm" when that's related to a session.
The files stream.{c,h} were added and session.{c,h} removed.
The session will be reintroduced later and a few parts of the stream
will progressively be moved overthere. It will more or less contain
only what we need in an embryonic session.
Sample fetch functions and converters will have to change a bit so
that they'll use an L5 (session) instead of what's currently called
"L4" which is in fact L6 for now.
Once all changes are completed, we should see approximately this :
L7 - http_txn
L6 - stream
L5 - session
L4 - connection | applet
There will be at most one http_txn per stream, and a same session will
possibly be referenced by multiple streams. A connection will point to
a session and to a stream. The session will hold all the information
we need to keep even when we don't yet have a stream.
Some more cleanup is needed because some code was already far from
being clean. The server queue management still refers to sessions at
many places while comments talk about connections. This will have to
be cleaned up once we have a server-side connection pool manager.
Stream flags "SN_*" still need to be renamed, it doesn't seem like
any of them will need to move to the session.
2015-04-02 22:22:06 +00:00
|
|
|
static inline void stream_store_counters(struct stream *s)
|
2010-06-14 19:04:55 +00:00
|
|
|
{
|
2010-08-03 14:29:52 +00:00
|
|
|
void *ptr;
|
2012-12-09 14:55:40 +00:00
|
|
|
int i;
|
2017-06-13 17:37:32 +00:00
|
|
|
struct stksess *ts;
|
2010-08-03 14:29:52 +00:00
|
|
|
|
2013-07-23 17:15:30 +00:00
|
|
|
for (i = 0; i < MAX_SESS_STKCTR; i++) {
|
2017-06-13 17:37:32 +00:00
|
|
|
ts = stkctr_entry(&s->stkctr[i]);
|
|
|
|
|
if (!ts)
|
2012-12-09 14:55:40 +00:00
|
|
|
continue;
|
2015-09-21 15:48:24 +00:00
|
|
|
|
|
|
|
|
if (stkctr_entry(&s->sess->stkctr[i]))
|
|
|
|
|
continue;
|
|
|
|
|
|
2017-06-13 17:37:32 +00:00
|
|
|
ptr = stktable_data_ptr(s->stkctr[i].table, ts, STKTABLE_DT_CONN_CUR);
|
|
|
|
|
if (ptr) {
|
2017-11-07 09:42:54 +00:00
|
|
|
HA_RWLOCK_WRLOCK(STK_SESS_LOCK, &ts->lock);
|
2017-06-13 17:37:32 +00:00
|
|
|
|
2010-08-03 14:29:52 +00:00
|
|
|
stktable_data_cast(ptr, conn_cur)--;
|
2017-06-13 17:37:32 +00:00
|
|
|
|
2017-11-07 09:42:54 +00:00
|
|
|
HA_RWLOCK_WRUNLOCK(STK_SESS_LOCK, &ts->lock);
|
2017-11-29 15:15:07 +00:00
|
|
|
|
|
|
|
|
/* If data was modified, we need to touch to re-schedule sync */
|
|
|
|
|
stktable_touch_local(s->stkctr[i].table, ts, 0);
|
2017-06-13 17:37:32 +00:00
|
|
|
}
|
2014-01-28 22:18:23 +00:00
|
|
|
stkctr_set_entry(&s->stkctr[i], NULL);
|
2017-06-13 17:37:32 +00:00
|
|
|
stksess_kill_if_expired(s->stkctr[i].table, ts, 1);
|
2010-06-18 14:35:43 +00:00
|
|
|
}
|
2010-06-14 19:04:55 +00:00
|
|
|
}
|
|
|
|
|
|
REORG/MAJOR: session: rename the "session" entity to "stream"
With HTTP/2, we'll have to support multiplexed streams. A stream is in
fact the largest part of what we currently call a session, it has buffers,
logs, etc.
In order to catch any error, this commit removes any reference to the
struct session and tries to rename most "session" occurrences in function
names to "stream" and "sess" to "strm" when that's related to a session.
The files stream.{c,h} were added and session.{c,h} removed.
The session will be reintroduced later and a few parts of the stream
will progressively be moved overthere. It will more or less contain
only what we need in an embryonic session.
Sample fetch functions and converters will have to change a bit so
that they'll use an L5 (session) instead of what's currently called
"L4" which is in fact L6 for now.
Once all changes are completed, we should see approximately this :
L7 - http_txn
L6 - stream
L5 - session
L4 - connection | applet
There will be at most one http_txn per stream, and a same session will
possibly be referenced by multiple streams. A connection will point to
a session and to a stream. The session will hold all the information
we need to keep even when we don't yet have a stream.
Some more cleanup is needed because some code was already far from
being clean. The server queue management still refers to sessions at
many places while comments talk about connections. This will have to
be cleaned up once we have a server-side connection pool manager.
Stream flags "SN_*" still need to be renamed, it doesn't seem like
any of them will need to move to the session.
2015-04-02 22:22:06 +00:00
|
|
|
/* Remove the refcount from the stream counters tracked at the content level if
|
2010-08-03 14:29:52 +00:00
|
|
|
* any, and clear the pointer to ensure this is only performed once. The caller
|
2015-09-21 15:48:24 +00:00
|
|
|
* is responsible for ensuring that the pointer is valid first. We must be
|
|
|
|
|
* extremely careful not to touch the entries we inherited from the session.
|
2010-06-14 19:04:55 +00:00
|
|
|
*/
|
REORG/MAJOR: session: rename the "session" entity to "stream"
With HTTP/2, we'll have to support multiplexed streams. A stream is in
fact the largest part of what we currently call a session, it has buffers,
logs, etc.
In order to catch any error, this commit removes any reference to the
struct session and tries to rename most "session" occurrences in function
names to "stream" and "sess" to "strm" when that's related to a session.
The files stream.{c,h} were added and session.{c,h} removed.
The session will be reintroduced later and a few parts of the stream
will progressively be moved overthere. It will more or less contain
only what we need in an embryonic session.
Sample fetch functions and converters will have to change a bit so
that they'll use an L5 (session) instead of what's currently called
"L4" which is in fact L6 for now.
Once all changes are completed, we should see approximately this :
L7 - http_txn
L6 - stream
L5 - session
L4 - connection | applet
There will be at most one http_txn per stream, and a same session will
possibly be referenced by multiple streams. A connection will point to
a session and to a stream. The session will hold all the information
we need to keep even when we don't yet have a stream.
Some more cleanup is needed because some code was already far from
being clean. The server queue management still refers to sessions at
many places while comments talk about connections. This will have to
be cleaned up once we have a server-side connection pool manager.
Stream flags "SN_*" still need to be renamed, it doesn't seem like
any of them will need to move to the session.
2015-04-02 22:22:06 +00:00
|
|
|
static inline void stream_stop_content_counters(struct stream *s)
|
2010-06-14 19:04:55 +00:00
|
|
|
{
|
2017-06-13 17:37:32 +00:00
|
|
|
struct stksess *ts;
|
2010-08-03 14:29:52 +00:00
|
|
|
void *ptr;
|
2012-12-09 14:55:40 +00:00
|
|
|
int i;
|
2010-06-18 19:03:20 +00:00
|
|
|
|
2013-07-23 17:15:30 +00:00
|
|
|
for (i = 0; i < MAX_SESS_STKCTR; i++) {
|
2017-06-13 17:37:32 +00:00
|
|
|
ts = stkctr_entry(&s->stkctr[i]);
|
|
|
|
|
if (!ts)
|
2012-12-09 14:55:40 +00:00
|
|
|
continue;
|
|
|
|
|
|
2015-09-21 15:48:24 +00:00
|
|
|
if (stkctr_entry(&s->sess->stkctr[i]))
|
|
|
|
|
continue;
|
|
|
|
|
|
2014-01-28 22:18:23 +00:00
|
|
|
if (!(stkctr_flags(&s->stkctr[i]) & STKCTR_TRACK_CONTENT))
|
2012-12-09 14:55:40 +00:00
|
|
|
continue;
|
|
|
|
|
|
2017-06-13 17:37:32 +00:00
|
|
|
ptr = stktable_data_ptr(s->stkctr[i].table, ts, STKTABLE_DT_CONN_CUR);
|
|
|
|
|
if (ptr) {
|
2017-11-07 09:42:54 +00:00
|
|
|
HA_RWLOCK_WRLOCK(STK_SESS_LOCK, &ts->lock);
|
2017-06-13 17:37:32 +00:00
|
|
|
|
2010-08-06 18:11:05 +00:00
|
|
|
stktable_data_cast(ptr, conn_cur)--;
|
2017-06-13 17:37:32 +00:00
|
|
|
|
2017-11-07 09:42:54 +00:00
|
|
|
HA_RWLOCK_WRUNLOCK(STK_SESS_LOCK, &ts->lock);
|
2017-11-29 15:15:07 +00:00
|
|
|
|
|
|
|
|
/* If data was modified, we need to touch to re-schedule sync */
|
|
|
|
|
stktable_touch_local(s->stkctr[i].table, ts, 0);
|
2017-06-13 17:37:32 +00:00
|
|
|
}
|
2014-01-28 22:18:23 +00:00
|
|
|
stkctr_set_entry(&s->stkctr[i], NULL);
|
2017-06-13 17:37:32 +00:00
|
|
|
stksess_kill_if_expired(s->stkctr[i].table, ts, 1);
|
2010-08-06 18:11:05 +00:00
|
|
|
}
|
2010-08-03 14:29:52 +00:00
|
|
|
}
|
2010-06-18 19:03:20 +00:00
|
|
|
|
2010-08-03 14:29:52 +00:00
|
|
|
/* Increase total and concurrent connection count for stick entry <ts> of table
|
|
|
|
|
* <t>. The caller is responsible for ensuring that <t> and <ts> are valid
|
|
|
|
|
* pointers, and for calling this only once per connection.
|
|
|
|
|
*/
|
REORG/MAJOR: session: rename the "session" entity to "stream"
With HTTP/2, we'll have to support multiplexed streams. A stream is in
fact the largest part of what we currently call a session, it has buffers,
logs, etc.
In order to catch any error, this commit removes any reference to the
struct session and tries to rename most "session" occurrences in function
names to "stream" and "sess" to "strm" when that's related to a session.
The files stream.{c,h} were added and session.{c,h} removed.
The session will be reintroduced later and a few parts of the stream
will progressively be moved overthere. It will more or less contain
only what we need in an embryonic session.
Sample fetch functions and converters will have to change a bit so
that they'll use an L5 (session) instead of what's currently called
"L4" which is in fact L6 for now.
Once all changes are completed, we should see approximately this :
L7 - http_txn
L6 - stream
L5 - session
L4 - connection | applet
There will be at most one http_txn per stream, and a same session will
possibly be referenced by multiple streams. A connection will point to
a session and to a stream. The session will hold all the information
we need to keep even when we don't yet have a stream.
Some more cleanup is needed because some code was already far from
being clean. The server queue management still refers to sessions at
many places while comments talk about connections. This will have to
be cleaned up once we have a server-side connection pool manager.
Stream flags "SN_*" still need to be renamed, it doesn't seem like
any of them will need to move to the session.
2015-04-02 22:22:06 +00:00
|
|
|
static inline void stream_start_counters(struct stktable *t, struct stksess *ts)
|
2010-08-03 14:29:52 +00:00
|
|
|
{
|
|
|
|
|
void *ptr;
|
|
|
|
|
|
2017-11-07 09:42:54 +00:00
|
|
|
HA_RWLOCK_WRLOCK(STK_SESS_LOCK, &ts->lock);
|
2017-06-13 17:37:32 +00:00
|
|
|
|
2010-08-03 14:29:52 +00:00
|
|
|
ptr = stktable_data_ptr(t, ts, STKTABLE_DT_CONN_CUR);
|
|
|
|
|
if (ptr)
|
|
|
|
|
stktable_data_cast(ptr, conn_cur)++;
|
|
|
|
|
|
|
|
|
|
ptr = stktable_data_ptr(t, ts, STKTABLE_DT_CONN_CNT);
|
|
|
|
|
if (ptr)
|
|
|
|
|
stktable_data_cast(ptr, conn_cnt)++;
|
|
|
|
|
|
|
|
|
|
ptr = stktable_data_ptr(t, ts, STKTABLE_DT_CONN_RATE);
|
|
|
|
|
if (ptr)
|
|
|
|
|
update_freq_ctr_period(&stktable_data_cast(ptr, conn_rate),
|
|
|
|
|
t->data_arg[STKTABLE_DT_CONN_RATE].u, 1);
|
|
|
|
|
if (tick_isset(t->expire))
|
|
|
|
|
ts->expire = tick_add(now_ms, MS_TO_TICKS(t->expire));
|
2017-06-13 17:37:32 +00:00
|
|
|
|
2017-11-07 09:42:54 +00:00
|
|
|
HA_RWLOCK_WRUNLOCK(STK_SESS_LOCK, &ts->lock);
|
2017-11-29 15:15:07 +00:00
|
|
|
|
|
|
|
|
/* If data was modified, we need to touch to re-schedule sync */
|
|
|
|
|
stktable_touch_local(t, ts, 0);
|
2010-08-03 14:29:52 +00:00
|
|
|
}
|
2010-06-20 09:19:22 +00:00
|
|
|
|
REORG/MAJOR: session: rename the "session" entity to "stream"
With HTTP/2, we'll have to support multiplexed streams. A stream is in
fact the largest part of what we currently call a session, it has buffers,
logs, etc.
In order to catch any error, this commit removes any reference to the
struct session and tries to rename most "session" occurrences in function
names to "stream" and "sess" to "strm" when that's related to a session.
The files stream.{c,h} were added and session.{c,h} removed.
The session will be reintroduced later and a few parts of the stream
will progressively be moved overthere. It will more or less contain
only what we need in an embryonic session.
Sample fetch functions and converters will have to change a bit so
that they'll use an L5 (session) instead of what's currently called
"L4" which is in fact L6 for now.
Once all changes are completed, we should see approximately this :
L7 - http_txn
L6 - stream
L5 - session
L4 - connection | applet
There will be at most one http_txn per stream, and a same session will
possibly be referenced by multiple streams. A connection will point to
a session and to a stream. The session will hold all the information
we need to keep even when we don't yet have a stream.
Some more cleanup is needed because some code was already far from
being clean. The server queue management still refers to sessions at
many places while comments talk about connections. This will have to
be cleaned up once we have a server-side connection pool manager.
Stream flags "SN_*" still need to be renamed, it doesn't seem like
any of them will need to move to the session.
2015-04-02 22:22:06 +00:00
|
|
|
/* Enable tracking of stream counters as <stkctr> on stksess <ts>. The caller is
|
2010-08-03 14:29:52 +00:00
|
|
|
* responsible for ensuring that <t> and <ts> are valid pointers. Some controls
|
|
|
|
|
* are performed to ensure the state can still change.
|
|
|
|
|
*/
|
REORG/MAJOR: session: rename the "session" entity to "stream"
With HTTP/2, we'll have to support multiplexed streams. A stream is in
fact the largest part of what we currently call a session, it has buffers,
logs, etc.
In order to catch any error, this commit removes any reference to the
struct session and tries to rename most "session" occurrences in function
names to "stream" and "sess" to "strm" when that's related to a session.
The files stream.{c,h} were added and session.{c,h} removed.
The session will be reintroduced later and a few parts of the stream
will progressively be moved overthere. It will more or less contain
only what we need in an embryonic session.
Sample fetch functions and converters will have to change a bit so
that they'll use an L5 (session) instead of what's currently called
"L4" which is in fact L6 for now.
Once all changes are completed, we should see approximately this :
L7 - http_txn
L6 - stream
L5 - session
L4 - connection | applet
There will be at most one http_txn per stream, and a same session will
possibly be referenced by multiple streams. A connection will point to
a session and to a stream. The session will hold all the information
we need to keep even when we don't yet have a stream.
Some more cleanup is needed because some code was already far from
being clean. The server queue management still refers to sessions at
many places while comments talk about connections. This will have to
be cleaned up once we have a server-side connection pool manager.
Stream flags "SN_*" still need to be renamed, it doesn't seem like
any of them will need to move to the session.
2015-04-02 22:22:06 +00:00
|
|
|
static inline void stream_track_stkctr(struct stkctr *ctr, struct stktable *t, struct stksess *ts)
|
2010-08-03 14:29:52 +00:00
|
|
|
{
|
2017-06-13 17:37:32 +00:00
|
|
|
/* Why this test ???? */
|
2014-01-28 22:18:23 +00:00
|
|
|
if (stkctr_entry(ctr))
|
2010-08-03 14:29:52 +00:00
|
|
|
return;
|
|
|
|
|
|
2012-12-09 14:55:40 +00:00
|
|
|
ctr->table = t;
|
2014-01-28 22:18:23 +00:00
|
|
|
stkctr_set_entry(ctr, ts);
|
REORG/MAJOR: session: rename the "session" entity to "stream"
With HTTP/2, we'll have to support multiplexed streams. A stream is in
fact the largest part of what we currently call a session, it has buffers,
logs, etc.
In order to catch any error, this commit removes any reference to the
struct session and tries to rename most "session" occurrences in function
names to "stream" and "sess" to "strm" when that's related to a session.
The files stream.{c,h} were added and session.{c,h} removed.
The session will be reintroduced later and a few parts of the stream
will progressively be moved overthere. It will more or less contain
only what we need in an embryonic session.
Sample fetch functions and converters will have to change a bit so
that they'll use an L5 (session) instead of what's currently called
"L4" which is in fact L6 for now.
Once all changes are completed, we should see approximately this :
L7 - http_txn
L6 - stream
L5 - session
L4 - connection | applet
There will be at most one http_txn per stream, and a same session will
possibly be referenced by multiple streams. A connection will point to
a session and to a stream. The session will hold all the information
we need to keep even when we don't yet have a stream.
Some more cleanup is needed because some code was already far from
being clean. The server queue management still refers to sessions at
many places while comments talk about connections. This will have to
be cleaned up once we have a server-side connection pool manager.
Stream flags "SN_*" still need to be renamed, it doesn't seem like
any of them will need to move to the session.
2015-04-02 22:22:06 +00:00
|
|
|
stream_start_counters(t, ts);
|
2010-06-14 19:04:55 +00:00
|
|
|
}
|
2007-11-24 21:12:47 +00:00
|
|
|
|
2010-06-23 09:44:09 +00:00
|
|
|
/* Increase the number of cumulated HTTP requests in the tracked counters */
|
REORG/MAJOR: session: rename the "session" entity to "stream"
With HTTP/2, we'll have to support multiplexed streams. A stream is in
fact the largest part of what we currently call a session, it has buffers,
logs, etc.
In order to catch any error, this commit removes any reference to the
struct session and tries to rename most "session" occurrences in function
names to "stream" and "sess" to "strm" when that's related to a session.
The files stream.{c,h} were added and session.{c,h} removed.
The session will be reintroduced later and a few parts of the stream
will progressively be moved overthere. It will more or less contain
only what we need in an embryonic session.
Sample fetch functions and converters will have to change a bit so
that they'll use an L5 (session) instead of what's currently called
"L4" which is in fact L6 for now.
Once all changes are completed, we should see approximately this :
L7 - http_txn
L6 - stream
L5 - session
L4 - connection | applet
There will be at most one http_txn per stream, and a same session will
possibly be referenced by multiple streams. A connection will point to
a session and to a stream. The session will hold all the information
we need to keep even when we don't yet have a stream.
Some more cleanup is needed because some code was already far from
being clean. The server queue management still refers to sessions at
many places while comments talk about connections. This will have to
be cleaned up once we have a server-side connection pool manager.
Stream flags "SN_*" still need to be renamed, it doesn't seem like
any of them will need to move to the session.
2015-04-02 22:22:06 +00:00
|
|
|
static void inline stream_inc_http_req_ctr(struct stream *s)
|
2010-06-23 09:44:09 +00:00
|
|
|
{
|
2017-06-13 17:37:32 +00:00
|
|
|
struct stksess *ts;
|
2010-08-03 14:29:52 +00:00
|
|
|
void *ptr;
|
2012-12-09 14:55:40 +00:00
|
|
|
int i;
|
2010-06-23 09:44:09 +00:00
|
|
|
|
2013-07-23 17:15:30 +00:00
|
|
|
for (i = 0; i < MAX_SESS_STKCTR; i++) {
|
2015-04-04 14:29:12 +00:00
|
|
|
struct stkctr *stkctr = &s->stkctr[i];
|
|
|
|
|
|
2017-06-13 17:37:32 +00:00
|
|
|
ts = stkctr_entry(stkctr);
|
|
|
|
|
if (!ts) {
|
2015-04-04 14:29:12 +00:00
|
|
|
stkctr = &s->sess->stkctr[i];
|
2017-06-13 17:37:32 +00:00
|
|
|
ts = stkctr_entry(stkctr);
|
|
|
|
|
if (!ts)
|
2015-04-04 14:29:12 +00:00
|
|
|
continue;
|
|
|
|
|
}
|
2010-06-23 09:44:09 +00:00
|
|
|
|
2017-11-07 09:42:54 +00:00
|
|
|
HA_RWLOCK_WRLOCK(STK_SESS_LOCK, &ts->lock);
|
2017-06-13 17:37:32 +00:00
|
|
|
|
|
|
|
|
ptr = stktable_data_ptr(stkctr->table, ts, STKTABLE_DT_HTTP_REQ_CNT);
|
2010-08-03 14:29:52 +00:00
|
|
|
if (ptr)
|
|
|
|
|
stktable_data_cast(ptr, http_req_cnt)++;
|
2012-12-09 11:00:04 +00:00
|
|
|
|
2017-06-13 17:37:32 +00:00
|
|
|
ptr = stktable_data_ptr(stkctr->table, ts, STKTABLE_DT_HTTP_REQ_RATE);
|
2012-12-09 11:00:04 +00:00
|
|
|
if (ptr)
|
|
|
|
|
update_freq_ctr_period(&stktable_data_cast(ptr, http_req_rate),
|
2015-04-04 14:29:12 +00:00
|
|
|
stkctr->table->data_arg[STKTABLE_DT_HTTP_REQ_RATE].u, 1);
|
2017-06-13 17:37:32 +00:00
|
|
|
|
2017-11-07 09:42:54 +00:00
|
|
|
HA_RWLOCK_WRUNLOCK(STK_SESS_LOCK, &ts->lock);
|
2017-11-29 15:15:07 +00:00
|
|
|
|
|
|
|
|
/* If data was modified, we need to touch to re-schedule sync */
|
|
|
|
|
stktable_touch_local(stkctr->table, ts, 0);
|
2012-12-09 11:00:04 +00:00
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
2015-04-04 14:29:12 +00:00
|
|
|
/* Increase the number of cumulated HTTP requests in the backend's tracked
|
|
|
|
|
* counters. We don't look up the session since it cannot happen in the bakcend.
|
|
|
|
|
*/
|
REORG/MAJOR: session: rename the "session" entity to "stream"
With HTTP/2, we'll have to support multiplexed streams. A stream is in
fact the largest part of what we currently call a session, it has buffers,
logs, etc.
In order to catch any error, this commit removes any reference to the
struct session and tries to rename most "session" occurrences in function
names to "stream" and "sess" to "strm" when that's related to a session.
The files stream.{c,h} were added and session.{c,h} removed.
The session will be reintroduced later and a few parts of the stream
will progressively be moved overthere. It will more or less contain
only what we need in an embryonic session.
Sample fetch functions and converters will have to change a bit so
that they'll use an L5 (session) instead of what's currently called
"L4" which is in fact L6 for now.
Once all changes are completed, we should see approximately this :
L7 - http_txn
L6 - stream
L5 - session
L4 - connection | applet
There will be at most one http_txn per stream, and a same session will
possibly be referenced by multiple streams. A connection will point to
a session and to a stream. The session will hold all the information
we need to keep even when we don't yet have a stream.
Some more cleanup is needed because some code was already far from
being clean. The server queue management still refers to sessions at
many places while comments talk about connections. This will have to
be cleaned up once we have a server-side connection pool manager.
Stream flags "SN_*" still need to be renamed, it doesn't seem like
any of them will need to move to the session.
2015-04-02 22:22:06 +00:00
|
|
|
static void inline stream_inc_be_http_req_ctr(struct stream *s)
|
2012-12-09 11:00:04 +00:00
|
|
|
{
|
2017-06-13 17:37:32 +00:00
|
|
|
struct stksess *ts;
|
2012-12-09 11:00:04 +00:00
|
|
|
void *ptr;
|
2012-12-09 14:55:40 +00:00
|
|
|
int i;
|
2012-12-09 11:00:04 +00:00
|
|
|
|
2013-07-23 17:15:30 +00:00
|
|
|
for (i = 0; i < MAX_SESS_STKCTR; i++) {
|
2015-04-04 14:29:12 +00:00
|
|
|
struct stkctr *stkctr = &s->stkctr[i];
|
|
|
|
|
|
2017-06-13 17:37:32 +00:00
|
|
|
ts = stkctr_entry(stkctr);
|
|
|
|
|
if (!ts)
|
2012-12-09 14:55:40 +00:00
|
|
|
continue;
|
2012-12-09 11:00:04 +00:00
|
|
|
|
2014-01-28 22:18:23 +00:00
|
|
|
if (!(stkctr_flags(&s->stkctr[i]) & STKCTR_TRACK_BACKEND))
|
2012-12-09 14:55:40 +00:00
|
|
|
continue;
|
2012-12-09 11:00:04 +00:00
|
|
|
|
2017-11-07 09:42:54 +00:00
|
|
|
HA_RWLOCK_WRLOCK(STK_SESS_LOCK, &ts->lock);
|
2017-06-13 17:37:32 +00:00
|
|
|
|
|
|
|
|
ptr = stktable_data_ptr(stkctr->table, ts, STKTABLE_DT_HTTP_REQ_CNT);
|
2012-12-09 11:00:04 +00:00
|
|
|
if (ptr)
|
|
|
|
|
stktable_data_cast(ptr, http_req_cnt)++;
|
2010-08-03 14:29:52 +00:00
|
|
|
|
2017-06-13 17:37:32 +00:00
|
|
|
ptr = stktable_data_ptr(stkctr->table, ts, STKTABLE_DT_HTTP_REQ_RATE);
|
2010-08-03 14:29:52 +00:00
|
|
|
if (ptr)
|
|
|
|
|
update_freq_ctr_period(&stktable_data_cast(ptr, http_req_rate),
|
2015-04-04 14:29:12 +00:00
|
|
|
stkctr->table->data_arg[STKTABLE_DT_HTTP_REQ_RATE].u, 1);
|
2017-06-13 17:37:32 +00:00
|
|
|
|
2017-11-07 09:42:54 +00:00
|
|
|
HA_RWLOCK_WRUNLOCK(STK_SESS_LOCK, &ts->lock);
|
2017-11-29 15:15:07 +00:00
|
|
|
|
|
|
|
|
/* If data was modified, we need to touch to re-schedule sync */
|
|
|
|
|
stktable_touch_local(stkctr->table, ts, 0);
|
2010-06-23 09:44:09 +00:00
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/* Increase the number of cumulated failed HTTP requests in the tracked
|
|
|
|
|
* counters. Only 4xx requests should be counted here so that we can
|
|
|
|
|
* distinguish between errors caused by client behaviour and other ones.
|
|
|
|
|
* Note that even 404 are interesting because they're generally caused by
|
|
|
|
|
* vulnerability scans.
|
|
|
|
|
*/
|
REORG/MAJOR: session: rename the "session" entity to "stream"
With HTTP/2, we'll have to support multiplexed streams. A stream is in
fact the largest part of what we currently call a session, it has buffers,
logs, etc.
In order to catch any error, this commit removes any reference to the
struct session and tries to rename most "session" occurrences in function
names to "stream" and "sess" to "strm" when that's related to a session.
The files stream.{c,h} were added and session.{c,h} removed.
The session will be reintroduced later and a few parts of the stream
will progressively be moved overthere. It will more or less contain
only what we need in an embryonic session.
Sample fetch functions and converters will have to change a bit so
that they'll use an L5 (session) instead of what's currently called
"L4" which is in fact L6 for now.
Once all changes are completed, we should see approximately this :
L7 - http_txn
L6 - stream
L5 - session
L4 - connection | applet
There will be at most one http_txn per stream, and a same session will
possibly be referenced by multiple streams. A connection will point to
a session and to a stream. The session will hold all the information
we need to keep even when we don't yet have a stream.
Some more cleanup is needed because some code was already far from
being clean. The server queue management still refers to sessions at
many places while comments talk about connections. This will have to
be cleaned up once we have a server-side connection pool manager.
Stream flags "SN_*" still need to be renamed, it doesn't seem like
any of them will need to move to the session.
2015-04-02 22:22:06 +00:00
|
|
|
static void inline stream_inc_http_err_ctr(struct stream *s)
|
2010-06-23 09:44:09 +00:00
|
|
|
{
|
2017-06-13 17:37:32 +00:00
|
|
|
struct stksess *ts;
|
2010-08-03 14:29:52 +00:00
|
|
|
void *ptr;
|
2012-12-09 14:55:40 +00:00
|
|
|
int i;
|
2010-08-03 14:29:52 +00:00
|
|
|
|
2013-07-23 17:15:30 +00:00
|
|
|
for (i = 0; i < MAX_SESS_STKCTR; i++) {
|
2015-04-04 14:29:12 +00:00
|
|
|
struct stkctr *stkctr = &s->stkctr[i];
|
|
|
|
|
|
2017-06-13 17:37:32 +00:00
|
|
|
ts = stkctr_entry(stkctr);
|
|
|
|
|
if (!ts) {
|
2015-04-04 14:29:12 +00:00
|
|
|
stkctr = &s->sess->stkctr[i];
|
2017-06-13 17:37:32 +00:00
|
|
|
ts = stkctr_entry(stkctr);
|
|
|
|
|
if (!ts)
|
2015-04-04 14:29:12 +00:00
|
|
|
continue;
|
|
|
|
|
}
|
2010-06-23 09:44:09 +00:00
|
|
|
|
2017-11-07 09:42:54 +00:00
|
|
|
HA_RWLOCK_WRLOCK(STK_SESS_LOCK, &ts->lock);
|
2017-06-13 17:37:32 +00:00
|
|
|
|
|
|
|
|
ptr = stktable_data_ptr(stkctr->table, ts, STKTABLE_DT_HTTP_ERR_CNT);
|
2010-06-23 09:44:09 +00:00
|
|
|
if (ptr)
|
|
|
|
|
stktable_data_cast(ptr, http_err_cnt)++;
|
|
|
|
|
|
2017-06-13 17:37:32 +00:00
|
|
|
ptr = stktable_data_ptr(stkctr->table, ts, STKTABLE_DT_HTTP_ERR_RATE);
|
2010-06-23 09:44:09 +00:00
|
|
|
if (ptr)
|
|
|
|
|
update_freq_ctr_period(&stktable_data_cast(ptr, http_err_rate),
|
2015-04-04 14:29:12 +00:00
|
|
|
stkctr->table->data_arg[STKTABLE_DT_HTTP_ERR_RATE].u, 1);
|
2017-06-13 17:37:32 +00:00
|
|
|
|
2017-11-07 09:42:54 +00:00
|
|
|
HA_RWLOCK_WRUNLOCK(STK_SESS_LOCK, &ts->lock);
|
2017-11-29 15:15:07 +00:00
|
|
|
|
|
|
|
|
/* If data was modified, we need to touch to re-schedule sync */
|
|
|
|
|
stktable_touch_local(stkctr->table, ts, 0);
|
2010-06-23 09:44:09 +00:00
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
2017-11-26 17:48:14 +00:00
|
|
|
static void inline __stream_add_srv_conn(struct stream *sess, struct server *srv)
|
2011-06-21 05:34:57 +00:00
|
|
|
{
|
|
|
|
|
sess->srv_conn = srv;
|
|
|
|
|
LIST_ADD(&srv->actconns, &sess->by_srv);
|
2017-11-26 17:48:14 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
static void inline stream_add_srv_conn(struct stream *sess, struct server *srv)
|
|
|
|
|
{
|
|
|
|
|
HA_SPIN_LOCK(SERVER_LOCK, &srv->lock);
|
|
|
|
|
__stream_add_srv_conn(sess, srv);
|
2017-11-07 09:42:54 +00:00
|
|
|
HA_SPIN_UNLOCK(SERVER_LOCK, &srv->lock);
|
2011-06-21 05:34:57 +00:00
|
|
|
}
|
|
|
|
|
|
REORG/MAJOR: session: rename the "session" entity to "stream"
With HTTP/2, we'll have to support multiplexed streams. A stream is in
fact the largest part of what we currently call a session, it has buffers,
logs, etc.
In order to catch any error, this commit removes any reference to the
struct session and tries to rename most "session" occurrences in function
names to "stream" and "sess" to "strm" when that's related to a session.
The files stream.{c,h} were added and session.{c,h} removed.
The session will be reintroduced later and a few parts of the stream
will progressively be moved overthere. It will more or less contain
only what we need in an embryonic session.
Sample fetch functions and converters will have to change a bit so
that they'll use an L5 (session) instead of what's currently called
"L4" which is in fact L6 for now.
Once all changes are completed, we should see approximately this :
L7 - http_txn
L6 - stream
L5 - session
L4 - connection | applet
There will be at most one http_txn per stream, and a same session will
possibly be referenced by multiple streams. A connection will point to
a session and to a stream. The session will hold all the information
we need to keep even when we don't yet have a stream.
Some more cleanup is needed because some code was already far from
being clean. The server queue management still refers to sessions at
many places while comments talk about connections. This will have to
be cleaned up once we have a server-side connection pool manager.
Stream flags "SN_*" still need to be renamed, it doesn't seem like
any of them will need to move to the session.
2015-04-02 22:22:06 +00:00
|
|
|
static void inline stream_del_srv_conn(struct stream *sess)
|
2011-06-21 05:34:57 +00:00
|
|
|
{
|
2017-06-08 12:04:45 +00:00
|
|
|
struct server *srv = sess->srv_conn;
|
|
|
|
|
|
|
|
|
|
if (!srv)
|
2011-06-21 05:34:57 +00:00
|
|
|
return;
|
|
|
|
|
|
2017-11-07 09:42:54 +00:00
|
|
|
HA_SPIN_LOCK(SERVER_LOCK, &srv->lock);
|
2011-06-21 05:34:57 +00:00
|
|
|
sess->srv_conn = NULL;
|
|
|
|
|
LIST_DEL(&sess->by_srv);
|
2017-11-07 09:42:54 +00:00
|
|
|
HA_SPIN_UNLOCK(SERVER_LOCK, &srv->lock);
|
2011-06-21 05:34:57 +00:00
|
|
|
}
|
|
|
|
|
|
REORG/MAJOR: session: rename the "session" entity to "stream"
With HTTP/2, we'll have to support multiplexed streams. A stream is in
fact the largest part of what we currently call a session, it has buffers,
logs, etc.
In order to catch any error, this commit removes any reference to the
struct session and tries to rename most "session" occurrences in function
names to "stream" and "sess" to "strm" when that's related to a session.
The files stream.{c,h} were added and session.{c,h} removed.
The session will be reintroduced later and a few parts of the stream
will progressively be moved overthere. It will more or less contain
only what we need in an embryonic session.
Sample fetch functions and converters will have to change a bit so
that they'll use an L5 (session) instead of what's currently called
"L4" which is in fact L6 for now.
Once all changes are completed, we should see approximately this :
L7 - http_txn
L6 - stream
L5 - session
L4 - connection | applet
There will be at most one http_txn per stream, and a same session will
possibly be referenced by multiple streams. A connection will point to
a session and to a stream. The session will hold all the information
we need to keep even when we don't yet have a stream.
Some more cleanup is needed because some code was already far from
being clean. The server queue management still refers to sessions at
many places while comments talk about connections. This will have to
be cleaned up once we have a server-side connection pool manager.
Stream flags "SN_*" still need to be renamed, it doesn't seem like
any of them will need to move to the session.
2015-04-02 22:22:06 +00:00
|
|
|
static void inline stream_init_srv_conn(struct stream *sess)
|
2011-07-19 22:17:39 +00:00
|
|
|
{
|
|
|
|
|
sess->srv_conn = NULL;
|
|
|
|
|
LIST_INIT(&sess->by_srv);
|
|
|
|
|
}
|
|
|
|
|
|
BUG/MAJOR: Fix how the list of entities waiting for a buffer is handled
When an entity tries to get a buffer, if it cannot be allocted, for example
because the number of buffers which may be allocated per process is limited,
this entity is added in a list (called <buffer_wq>) and wait for an available
buffer.
Historically, the <buffer_wq> list was logically attached to streams because it
were the only entities likely to be added in it. Now, applets can also be
waiting for a free buffer. And with filters, we could imagine to have more other
entities waiting for a buffer. So it make sense to have a generic list.
Anyway, with the current design there is a bug. When an applet failed to get a
buffer, it will wait. But we add the stream attached to the applet in
<buffer_wq>, instead of the applet itself. So when a buffer is available, we
wake up the stream and not the waiting applet. So, it is possible to have
waiting applets and never awakened.
So, now, <buffer_wq> is independant from streams. And we really add the waiting
entity in <buffer_wq>. To be generic, the entity is responsible to define the
callback used to awaken it.
In addition, applets will still request an input buffer when they become
active. But they will not be sleeped anymore if no buffer are available. So this
is the responsibility to the applet I/O handler to check if this buffer is
allocated or not. This way, an applet can decide if this buffer is required or
not and can do additional processing if not.
[wt: backport to 1.7 and 1.6]
2016-12-09 16:30:18 +00:00
|
|
|
/* Callback used to wake up a stream when a buffer is available. The stream <s>
|
|
|
|
|
* is woken up is if it is not already running and if it is not already in the
|
|
|
|
|
* task run queue. This functions returns 1 is the stream is woken up, otherwise
|
|
|
|
|
* it returns 0. */
|
|
|
|
|
static int inline stream_res_wakeup(struct stream *s)
|
MAJOR: session: only wake up as many sessions as available buffers permit
We've already experimented with three wake up algorithms when releasing
buffers : the first naive one used to wake up far too many sessions,
causing many of them not to get any buffer. The second approach which
was still in use prior to this patch consisted in waking up either 1
or 2 sessions depending on the number of FDs we had released. And this
was still inaccurate. The third one tried to cover the accuracy issues
of the second and took into consideration the number of FDs the sessions
would be willing to use, but most of the time we ended up waking up too
many of them for nothing, or deadlocking by lack of buffers.
This patch completely removes the need to allocate two buffers at once.
Instead it splits allocations into critical and non-critical ones and
implements a reserve in the pool for this. The deadlock situation happens
when all buffers are be allocated for requests pending in a maxconn-limited
server queue, because then there's no more way to allocate buffers for
responses, and these responses are critical to release the servers's
connection in order to release the pending requests. In fact maxconn on
a server creates a dependence between sessions and particularly between
oldest session's responses and latest session's requests. Thus, it is
mandatory to get a free buffer for a response in order to release a
server connection which will permit to release a request buffer.
Since we definitely have non-symmetrical buffers, we need to implement
this logic in the buffer allocation mechanism. What this commit does is
implement a reserve of buffers which can only be allocated for responses
and that will never be allocated for requests. This is made possible by
the requester indicating how much margin it wants to leave after the
allocation succeeds. Thus it is a cooperative allocation mechanism : the
requester (process_session() in general) prefers not to get a buffer in
order to respect other's need for response buffers. The session management
code always knows if a buffer will be used for requests or responses, so
that is not difficult :
- either there's an applet on the initiator side and we really need
the request buffer (since currently the applet is called in the
context of the session)
- or we have a connection and we really need the response buffer (in
order to support building and sending an error message back)
This reserve ensures that we don't take all allocatable buffers for
requests waiting in a queue. The downside is that all the extra buffers
are really allocated to ensure they can be allocated. But with small
values it is not an issue.
With this change, we don't observe any more deadlocks even when running
with maxconn 1 on a server under severely constrained memory conditions.
The code becomes a bit tricky, it relies on the scheduler's run queue to
estimate how many sessions are already expected to run so that it doesn't
wake up everyone with too few resources. A better solution would probably
consist in having two queues, one for urgent requests and one for normal
requests. A failed allocation for a session dealing with an error, a
connection event, or the need for a response (or request when there's an
applet on the left) would go to the urgent request queue, while other
requests would go to the other queue. Urgent requests would be served
from 1 entry in the pool, while the regular ones would be served only
according to the reserve. Despite not yet having this, it works
remarkably well.
This mechanism is quite efficient, we don't perform too many wake up calls
anymore. For 1 million sessions elapsed during massive memory contention,
we observe about 4.5M calls to process_session() compared to 4.0M without
memory constraints. Previously we used to observe up to 16M calls, which
rougly means 12M failures.
During a test run under high memory constraints (limit enforced to 27 MB
instead of the 58 MB normally needed), performance used to drop by 53% prior
to this patch. Now with this patch instead it *increases* by about 1.5%.
The best effect of this change is that by limiting the memory usage to about
2/3 to 3/4 of what is needed by default, it's possible to increase performance
by up to about 18% mainly due to the fact that pools are reused more often
and remain hot in the CPU cache (observed on regular HTTP traffic with 20k
objects, buffers.limit = maxconn/10, buffers.reserve = limit/2).
Below is an example of scenario which used to cause a deadlock previously :
- connection is received
- two buffers are allocated in process_session() then released
- one is allocated when receiving an HTTP request
- the second buffer is allocated then released in process_session()
for request parsing then connection establishment.
- poll() says we can send, so the request buffer is sent and released
- process session gets notified that the connection is now established
and allocates two buffers then releases them
- all other sessions do the same till one cannot get the request buffer
without hitting the margin
- and now the server responds. stream_interface allocates the response
buffer and manages to get it since it's higher priority being for a
response.
- but process_session() cannot allocate the request buffer anymore
=> We could end up with all buffers used by responses so that none may
be allocated for a request in process_session().
When the applet processing leaves the session context, the test will have
to be changed so that we always allocate a response buffer regardless of
the left side (eg: H2->H1 gateway). A final improvement would consists in
being able to only retry the failed I/O operation without waking up a
task, but to date all experiments to achieve this have proven not to be
reliable enough.
2014-11-27 00:11:56 +00:00
|
|
|
{
|
2017-03-31 10:04:09 +00:00
|
|
|
if (s->task->state & TASK_RUNNING)
|
BUG/MAJOR: Fix how the list of entities waiting for a buffer is handled
When an entity tries to get a buffer, if it cannot be allocted, for example
because the number of buffers which may be allocated per process is limited,
this entity is added in a list (called <buffer_wq>) and wait for an available
buffer.
Historically, the <buffer_wq> list was logically attached to streams because it
were the only entities likely to be added in it. Now, applets can also be
waiting for a free buffer. And with filters, we could imagine to have more other
entities waiting for a buffer. So it make sense to have a generic list.
Anyway, with the current design there is a bug. When an applet failed to get a
buffer, it will wait. But we add the stream attached to the applet in
<buffer_wq>, instead of the applet itself. So when a buffer is available, we
wake up the stream and not the waiting applet. So, it is possible to have
waiting applets and never awakened.
So, now, <buffer_wq> is independant from streams. And we really add the waiting
entity in <buffer_wq>. To be generic, the entity is responsible to define the
callback used to awaken it.
In addition, applets will still request an input buffer when they become
active. But they will not be sleeped anymore if no buffer are available. So this
is the responsibility to the applet I/O handler to check if this buffer is
allocated or not. This way, an applet can decide if this buffer is required or
not and can do additional processing if not.
[wt: backport to 1.7 and 1.6]
2016-12-09 16:30:18 +00:00
|
|
|
return 0;
|
|
|
|
|
task_wakeup(s->task, TASK_WOKEN_RES);
|
|
|
|
|
return 1;
|
MAJOR: session: only wake up as many sessions as available buffers permit
We've already experimented with three wake up algorithms when releasing
buffers : the first naive one used to wake up far too many sessions,
causing many of them not to get any buffer. The second approach which
was still in use prior to this patch consisted in waking up either 1
or 2 sessions depending on the number of FDs we had released. And this
was still inaccurate. The third one tried to cover the accuracy issues
of the second and took into consideration the number of FDs the sessions
would be willing to use, but most of the time we ended up waking up too
many of them for nothing, or deadlocking by lack of buffers.
This patch completely removes the need to allocate two buffers at once.
Instead it splits allocations into critical and non-critical ones and
implements a reserve in the pool for this. The deadlock situation happens
when all buffers are be allocated for requests pending in a maxconn-limited
server queue, because then there's no more way to allocate buffers for
responses, and these responses are critical to release the servers's
connection in order to release the pending requests. In fact maxconn on
a server creates a dependence between sessions and particularly between
oldest session's responses and latest session's requests. Thus, it is
mandatory to get a free buffer for a response in order to release a
server connection which will permit to release a request buffer.
Since we definitely have non-symmetrical buffers, we need to implement
this logic in the buffer allocation mechanism. What this commit does is
implement a reserve of buffers which can only be allocated for responses
and that will never be allocated for requests. This is made possible by
the requester indicating how much margin it wants to leave after the
allocation succeeds. Thus it is a cooperative allocation mechanism : the
requester (process_session() in general) prefers not to get a buffer in
order to respect other's need for response buffers. The session management
code always knows if a buffer will be used for requests or responses, so
that is not difficult :
- either there's an applet on the initiator side and we really need
the request buffer (since currently the applet is called in the
context of the session)
- or we have a connection and we really need the response buffer (in
order to support building and sending an error message back)
This reserve ensures that we don't take all allocatable buffers for
requests waiting in a queue. The downside is that all the extra buffers
are really allocated to ensure they can be allocated. But with small
values it is not an issue.
With this change, we don't observe any more deadlocks even when running
with maxconn 1 on a server under severely constrained memory conditions.
The code becomes a bit tricky, it relies on the scheduler's run queue to
estimate how many sessions are already expected to run so that it doesn't
wake up everyone with too few resources. A better solution would probably
consist in having two queues, one for urgent requests and one for normal
requests. A failed allocation for a session dealing with an error, a
connection event, or the need for a response (or request when there's an
applet on the left) would go to the urgent request queue, while other
requests would go to the other queue. Urgent requests would be served
from 1 entry in the pool, while the regular ones would be served only
according to the reserve. Despite not yet having this, it works
remarkably well.
This mechanism is quite efficient, we don't perform too many wake up calls
anymore. For 1 million sessions elapsed during massive memory contention,
we observe about 4.5M calls to process_session() compared to 4.0M without
memory constraints. Previously we used to observe up to 16M calls, which
rougly means 12M failures.
During a test run under high memory constraints (limit enforced to 27 MB
instead of the 58 MB normally needed), performance used to drop by 53% prior
to this patch. Now with this patch instead it *increases* by about 1.5%.
The best effect of this change is that by limiting the memory usage to about
2/3 to 3/4 of what is needed by default, it's possible to increase performance
by up to about 18% mainly due to the fact that pools are reused more often
and remain hot in the CPU cache (observed on regular HTTP traffic with 20k
objects, buffers.limit = maxconn/10, buffers.reserve = limit/2).
Below is an example of scenario which used to cause a deadlock previously :
- connection is received
- two buffers are allocated in process_session() then released
- one is allocated when receiving an HTTP request
- the second buffer is allocated then released in process_session()
for request parsing then connection establishment.
- poll() says we can send, so the request buffer is sent and released
- process session gets notified that the connection is now established
and allocates two buffers then releases them
- all other sessions do the same till one cannot get the request buffer
without hitting the margin
- and now the server responds. stream_interface allocates the response
buffer and manages to get it since it's higher priority being for a
response.
- but process_session() cannot allocate the request buffer anymore
=> We could end up with all buffers used by responses so that none may
be allocated for a request in process_session().
When the applet processing leaves the session context, the test will have
to be changed so that we always allocate a response buffer regardless of
the left side (eg: H2->H1 gateway). A final improvement would consists in
being able to only retry the failed I/O operation without waking up a
task, but to date all experiments to achieve this have proven not to be
reliable enough.
2014-11-27 00:11:56 +00:00
|
|
|
}
|
|
|
|
|
|
2015-09-27 17:29:33 +00:00
|
|
|
void service_keywords_register(struct action_kw_list *kw_list);
|
|
|
|
|
|
REORG/MAJOR: session: rename the "session" entity to "stream"
With HTTP/2, we'll have to support multiplexed streams. A stream is in
fact the largest part of what we currently call a session, it has buffers,
logs, etc.
In order to catch any error, this commit removes any reference to the
struct session and tries to rename most "session" occurrences in function
names to "stream" and "sess" to "strm" when that's related to a session.
The files stream.{c,h} were added and session.{c,h} removed.
The session will be reintroduced later and a few parts of the stream
will progressively be moved overthere. It will more or less contain
only what we need in an embryonic session.
Sample fetch functions and converters will have to change a bit so
that they'll use an L5 (session) instead of what's currently called
"L4" which is in fact L6 for now.
Once all changes are completed, we should see approximately this :
L7 - http_txn
L6 - stream
L5 - session
L4 - connection | applet
There will be at most one http_txn per stream, and a same session will
possibly be referenced by multiple streams. A connection will point to
a session and to a stream. The session will hold all the information
we need to keep even when we don't yet have a stream.
Some more cleanup is needed because some code was already far from
being clean. The server queue management still refers to sessions at
many places while comments talk about connections. This will have to
be cleaned up once we have a server-side connection pool manager.
Stream flags "SN_*" still need to be renamed, it doesn't seem like
any of them will need to move to the session.
2015-04-02 22:22:06 +00:00
|
|
|
#endif /* _PROTO_STREAM_H */
|
2006-06-26 00:48:02 +00:00
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* Local variables:
|
|
|
|
|
* c-indent-level: 8
|
|
|
|
|
* c-basic-offset: 8
|
|
|
|
|
* End:
|
|
|
|
|
*/
|
