Skip to content

V
varnish-cache
Commit fa681397 authored by Dridi Boukelmoune's avatar Dridi Boukelmoune
Browse files
Options

param: Introduce the notion of thread pool parameters 

Unlike simple parameters, the field names in struct params don't match
the parameters names. To deal with that a new field is added to PARAM()
macros to allow simple and thread parameters to set themselves apart
in that regard without getting any more intrusive.

Refs #3250
parent 75b25921
Expand all Hide whitespace changes
Inline Side-by-side
Showing with 309 additions and 452 deletions
bin/varnishd/common/common_param.h
View file @ fa681397
......@@ -73,23 +73,27 @@ struct poolparam {
struct params {
#define ptyp_bool unsigned
#define ptyp_bytes ssize_t
#define ptyp_bytes_u unsigned
#define ptyp_double double
#define ptyp_poolparam struct poolparam
#define ptyp_timeout double
#define ptyp_uint unsigned
#define ptyp_vsl_buffer unsigned
#define ptyp_vsl_reclen unsigned
#define PARAM(ty, nm, ...) \
ptyp_##ty nm;
#define ptyp_bool unsigned
#define ptyp_bytes ssize_t
#define ptyp_bytes_u unsigned
#define ptyp_double double
#define ptyp_poolparam struct poolparam
#define ptyp_thread_pool_max unsigned
#define ptyp_thread_pool_min unsigned
#define ptyp_timeout double
#define ptyp_uint unsigned
#define ptyp_vsl_buffer unsigned
#define ptyp_vsl_reclen unsigned
#define PARAM(typ, fld, nm, ...) \
ptyp_##typ fld;
#include <tbl/params.h>
#undef ptyp_bool
#undef ptyp_bytes
#undef ptyp_bytes_u
#undef ptyp_double
#undef ptyp_poolparam
#undef ptyp_thread_pool_max
#undef ptyp_thread_pool_min
#undef ptyp_timeout
#undef ptyp_uint
#undef ptyp_vsl_buffer
......@@ -99,20 +103,6 @@ struct params {
uid_t uid;
gid_t gid;
/* Worker threads and pool */
unsigned wthread_min;
unsigned wthread_max;
unsigned wthread_reserve;
double wthread_timeout;
unsigned wthread_pools;
double wthread_add_delay;
double wthread_fail_delay;
double wthread_destroy_delay;
double wthread_watchdog;
unsigned wthread_stats_rate;
ssize_t wthread_stacksize;
unsigned wthread_queue_limit;
struct vre_limits vre_limits;
uint8_t vsl_mask[256>>3];
......
bin/varnishd/mgt/mgt_param.c
View file @ fa681397
......@@ -658,7 +658,6 @@ MCF_InitParams(struct cli *cli)
ssize_t def, low;
MCF_AddParams(mgt_parspec);
MCF_AddParams(WRK_parspec);
MCF_AddParams(VSL_parspec);
MCF_TcpParams();
......
bin/varnishd/mgt/mgt_param.h
View file @ fa681397
......@@ -71,6 +71,8 @@ tweak_t tweak_bytes_u;
tweak_t tweak_double;
tweak_t tweak_poolparam;
tweak_t tweak_string;
tweak_t tweak_thread_pool_min;
tweak_t tweak_thread_pool_max;
tweak_t tweak_timeout;
tweak_t tweak_uint;
tweak_t tweak_vsl_buffer;
......
bin/varnishd/mgt/mgt_param_tbl.c
View file @ fa681397
......@@ -38,7 +38,7 @@
struct parspec mgt_parspec[] = {
#define PARAM_ALL
#define PARAM(ty, nm, ...) { #nm, __VA_ARGS__ },
#define PARAM(typ, fld, nm, ...) { #nm, __VA_ARGS__ },
#include "tbl/params.h"
{ NULL, NULL, NULL }
};
bin/varnishd/mgt/mgt_pool.c
View file @ fa681397
......@@ -55,7 +55,7 @@
* limit, so they don't end up crossing.
*/
static int
int
tweak_thread_pool_min(struct vsb *vsb, const struct parspec *par,
const char *arg)
{
......@@ -69,7 +69,7 @@ tweak_thread_pool_min(struct vsb *vsb, const struct parspec *par,
return (0);
}
static int
int
tweak_thread_pool_max(struct vsb *vsb, const struct parspec *par,
const char *arg)
{
......@@ -81,192 +81,3 @@ tweak_thread_pool_max(struct vsb *vsb, const struct parspec *par,
"%u", mgt_param.wthread_max);
return (0);
}
/*--------------------------------------------------------------------
* The thread pool parameter definitions used to generate the varnishd
* manual. Check the generated RST after updating.
*/
struct parspec WRK_parspec[] = {
{ "thread_pools", tweak_uint, &mgt_param.wthread_pools,
"1", NULL, /* maximum defined in mgt_param.c */ "2",
"pools",
"Number of worker thread pools.\n"
"\n"
"Increasing the number of worker pools decreases lock "
"contention. Each worker pool also has a thread accepting "
"new connections, so for very high rates of incoming new "
"connections on systems with many cores, increasing the "
"worker pools may be required.\n"
"\n"
"Too many pools waste CPU and RAM resources, and more than one "
"pool for each CPU is most likely detrimental to performance.\n"
"\n"
"Can be increased on the fly, but decreases require a "
"restart to take effect, unless the drop_pools experimental "
"debug flag is set.",
EXPERIMENTAL | DELAYED_EFFECT,
NULL, "defined when Varnish is built" },
{ "thread_pool_max", tweak_thread_pool_max, &mgt_param.wthread_max,
NULL, NULL, "5000",
"threads",
"The maximum number of worker threads in each pool.\n"
"\n"
"Do not set this higher than you have to, since excess "
"worker threads soak up RAM and CPU and generally just get "
"in the way of getting work done.",
DELAYED_EFFECT,
"thread_pool_min" },
{ "thread_pool_min", tweak_thread_pool_min, &mgt_param.wthread_min,
"5" /* TASK_QUEUE__END */, NULL, "100",
"threads",
"The minimum number of worker threads in each pool.\n"
"\n"
"Increasing this may help ramp up faster from low load "
"situations or when threads have expired.\n"
"\n"
"Technical minimum is 5 threads, " // TASK_QUEUE__END
"but this parameter is strongly recommended to be "
"at least 10", // 2 * TASK_QUEUE__END
DELAYED_EFFECT,
NULL, "thread_pool_max" },
{ "thread_pool_reserve", tweak_uint,
&mgt_param.wthread_reserve,
NULL, NULL, "0",
"threads",
"The number of worker threads reserved for vital tasks "
"in each pool.\n"
"\n"
"Tasks may require other tasks to complete (for example, "
"client requests may require backend requests, http2 sessions "
"require streams, which require requests). This reserve is to "
"ensure that lower priority tasks do not prevent higher "
"priority tasks from running even under high load.\n"
"\n"
"The effective value is at least 5 (the number of internal "
// ^ TASK_QUEUE__END
"priority classes), irrespective of this parameter.\n"
"Default is 0 to auto-tune (5% of thread_pool_min).\n"
"Minimum is 1 otherwise, maximum is 95% of thread_pool_min.",
DELAYED_EFFECT,
NULL, "95% of thread_pool_min" },
{ "thread_pool_timeout",
tweak_timeout, &mgt_param.wthread_timeout,
"10", NULL, "300",
"seconds",
"Thread idle threshold.\n"
"\n"
"Threads in excess of thread_pool_min, which have been idle "
"for at least this long, will be destroyed.",
EXPERIMENTAL | DELAYED_EFFECT },
{ "thread_pool_watchdog",
tweak_timeout, &mgt_param.wthread_watchdog,
"0.1", NULL, "60",
"seconds",
"Thread queue stuck watchdog.\n"
"\n"
"If no queued work have been released for this long,"
" the worker process panics itself.",
EXPERIMENTAL },
{ "thread_pool_destroy_delay",
tweak_timeout, &mgt_param.wthread_destroy_delay,
"0.01", NULL, "1",
"seconds",
"Wait this long after destroying a thread.\n"
"\n"
"This controls the decay of thread pools when idle(-ish).",
EXPERIMENTAL | DELAYED_EFFECT },
{ "thread_pool_add_delay",
tweak_timeout, &mgt_param.wthread_add_delay,
"0", NULL, "0",
"seconds",
"Wait at least this long after creating a thread.\n"
"\n"
"Some (buggy) systems may need a short (sub-second) "
"delay between creating threads.\n"
"Set this to a few milliseconds if you see the "
"'threads_failed' counter grow too much.\n"
"\n"
"Setting this too high results in insufficient worker threads.",
EXPERIMENTAL },
{ "thread_pool_fail_delay",
tweak_timeout, &mgt_param.wthread_fail_delay,
"10e-3", NULL, "0.2",
"seconds",
"Wait at least this long after a failed thread creation "
"before trying to create another thread.\n"
"\n"
"Failure to create a worker thread is often a sign that "
" the end is near, because the process is running out of "
"some resource. "
"This delay tries to not rush the end on needlessly.\n"
"\n"
"If thread creation failures are a problem, check that "
"thread_pool_max is not too high.\n"
"\n"
"It may also help to increase thread_pool_timeout and "
"thread_pool_min, to reduce the rate at which treads are "
"destroyed and later recreated.",
EXPERIMENTAL },
{ "thread_stats_rate",
tweak_uint, &mgt_param.wthread_stats_rate,
"0", NULL, "10",
"requests",
"Worker threads accumulate statistics, and dump these into "
"the global stats counters if the lock is free when they "
"finish a job (request/fetch etc.)\n"
"This parameters defines the maximum number of jobs "
"a worker thread may handle, before it is forced to dump "
"its accumulated stats into the global counters.",
EXPERIMENTAL },
{ "thread_queue_limit", tweak_uint, &mgt_param.wthread_queue_limit,
"0", NULL, "20",
NULL,
"Permitted request queue length per thread-pool.\n"
"\n"
"This sets the number of requests we will queue, waiting "
"for an available thread. Above this limit sessions will "
"be dropped instead of queued.",
EXPERIMENTAL },
{ "thread_pool_stack",
tweak_bytes, &mgt_param.wthread_stacksize,
NULL, NULL, NULL, // default set in mgt_param.c
"bytes",
"Worker thread stack size.\n"
"This will likely be rounded up to a multiple of 4k"
" (or whatever the page_size might be) by the kernel.\n"
"\n"
"The required stack size is primarily driven by the"
" depth of the call-tree. The most common relevant"
" determining factors in varnish core code are GZIP"
" (un)compression, ESI processing and regular"
" expression matches. VMODs may also require"
" significant amounts of additional stack. The"
" nesting depth of VCL subs is another factor,"
" although typically not predominant.\n"
"\n"
"The stack size is per thread, so the maximum total"
" memory required for worker thread stacks is in the"
" order of size = thread_pools x thread_pool_max x"
" thread_pool_stack.\n"
"\n"
"Thus, in particular for setups with many threads,"
" keeping the stack size at a minimum helps reduce"
" the amount of memory required by Varnish.\n"
"\n"
"On the other hand, thread_pool_stack must be large"
" enough under all circumstances, otherwise varnish"
" will crash due to a stack overflow. Usually, a"
" stack overflow manifests itself as a segmentation"
" fault (aka segfault / SIGSEGV) with the faulting"
" address being near the stack pointer (sp).\n"
"\n"
"Unless stack usage can be reduced,"
" thread_pool_stack must be increased when a stack"
" overflow occurs. Setting it in 150%-200%"
" increments is recommended until stack overflows"
" cease to occur.",
DELAYED_EFFECT,
NULL, NULL, "sysconf(_SC_THREAD_STACK_MIN)" },
{ NULL, NULL, NULL }
};
include/tbl/params.h
View file @ fa681397
This diff is collapsed.
Markdown is supported
0% or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment