For an information on how to write a service that distributes dynamic configs see Writing your own configs server .
Dynamic config values could be obtained via the dynamic_config::Source client that could be retrieved from components::DynamicConfig :
#include <userver/dynamic_config/value.hpp>
namespace myservice::smth {
:
components ::LoggableComponentBase(config, context),
config_(
context.FindComponent<
components ::DynamicConfig>()
.GetSource()
) {
[[maybe_unused]]
auto url = config[
"some-url" ].
As <std::string>();
const auto fs_tp_name = config[
"fs-task-processor" ].
As <std::string>();
utils::Async (fs_task_processor,
"my-component/fs-work" , [] { }).Get();
}
}
To get a specific value you need a parser for it. For example, here's how you could parse and get the SAMPLE_INTEGER_FROM_RUNTIME_CONFIG option:
namespace myservice::smth {
namespace {
return docs_map.Get(
"SAMPLE_INTEGER_FROM_RUNTIME_CONFIG" ).
As <
int >();
}
}
int Component::DoSomething() const {
const auto runtime_config = config_.GetSnapshot();
return runtime_config[kMyConfig];
}
}
Token bucket throttling options for new connections (socket(3)).
*-limit - token bucket size, set to 0 to disable the limit*-per-second - token bucket size refill speed
schema:
type: object
properties:
http-limit:
type: integer
minimum: 0
http-per-second:
type: integer
minimum: 0
https-limit:
type: integer
minimum: 0
https-per-second:
type: integer
minimum: 0
per-host-limit:
type: integer
minimum: 0
per-host-per-second:
type: integer
minimum: 0
additionalProperties: false
required:
- http-limit
- https-limit
- per-host-limit
Example:
{
"http-limit" : 6000,
"http-per-second" : 1500,
"https-limit" : 100,
"https-per-second" : 25,
"per-host-limit" : 3000,
"per-host-per-second" : 500
}
Used by components::HttpClient , affects the behavior of clients::http::Client and all the clients that use it.
Open connections pool size for curl (CURLMOPT_MAXCONNECTS). -1 means "detect by multiplying easy handles count on 4".
Example:
Used by components::HttpClient , affects the behavior of clients::http::Client and all the clients that use it.
Dynamic config that controls default $maxTimeMS for mongo requests (0 - disables default timeout).
schema:
type: integer
minimum: 0
Example:
Used by components::Mongo , components::MultiMongo .
Dynamic config that controls whether task-inherited deadline is accounted for while executing mongodb queries.
Example:
Used by components::Mongo , components::MultiMongo .
Dynamic config that controls default network and statement timeouts. Overrides the built-in timeouts from components::Postgres::kDefaultCommandControl , but could be overridden by POSTGRES_HANDLERS_COMMAND_CONTROL , POSTGRES_QUERIES_COMMAND_CONTROL and storages::postgres::CommandControl .
type: object
additionalProperties: false
properties:
network_timeout_ms:
type: integer
minimum: 1
statement_timeout_ms:
type: integer
minimum: 1
Example:
{
"network_timeout_ms" : 750,
"statement_timeout_ms" : 500
}
Used by components::Postgres .
Dynamic config that controls per-handle statement and network timeouts. Overrides POSTGRES_DEFAULT_COMMAND_CONTROL and built-in timeouts from components::Postgres::kDefaultCommandControl , but may be overridden by POSTGRES_QUERIES_COMMAND_CONTROL and storages::postgres::CommandControl .
type: object
additionalProperties:
$ref: "#/definitions/CommandControlByMethodMap"
definitions:
CommandControlByMethodMap:
type: object
additionalProperties:
$ref: "#/definitions/CommandControl"
CommandControl:
type: object
additionalProperties: false
properties:
network_timeout_ms:
type: integer
minimum: 1
statement_timeout_ms:
type: integer
minimum: 1
Example:
{
"/v2/rules/create" : {
"POST" : {
"network_timeout_ms" : 500,
"statement_timeout_ms" : 250
}
},
"/v2/rules/select" : {
"POST" : {
"network_timeout_ms" : 70,
"statement_timeout_ms" : 30
}
}
}
Used by components::Postgres .
Dynamic config that controls per-query/per-transaction statement and network timeouts, if those were not explicitly set via storages::postgres::CommandControl . Overrides the POSTGRES_HANDLERS_COMMAND_CONTROL , POSTGRES_QUERIES_COMMAND_CONTROL , POSTGRES_DEFAULT_COMMAND_CONTROL , and built-in timeouts from components::Postgres::kDefaultCommandControl .
Transaction timeouts in POSTGRES_QUERIES_COMMAND_CONTROL override the per-query timeouts in POSTGRES_QUERIES_COMMAND_CONTROL, so the latter are ignored if transaction timeouts are set.
type: object
additionalProperties:
$ref: "#/definitions/CommandControl"
definitions:
CommandControl:
type: object
additionalProperties: false
properties:
network_timeout_ms:
type: integer
minimum: 1
statement_timeout_ms:
type: integer
minimum: 1
Example:
{
"cleanup_processed_data" : {
"network_timeout_ms" : 92000,
"statement_timeout_ms" : 90000
},
"select_recent_users" : {
"network_timeout_ms" : 70,
"statement_timeout_ms" : 30
}
}
Used by components::Postgres .
Dynamic config that controls connection pool settings of PostgreSQL driver.
Take note that it overrides the static configuration values of the service!
type: object
additionalProperties:
$ref: "#/definitions/PoolSettings"
definitions:
PoolSettings:
type: object
additionalProperties: false
properties:
min_pool_size:
type: integer
minimum: 0
max_pool_size:
type: integer
minimum: 1
max_queue_size:
type: integer
minimum: 1
connecting_limit:
type: integer
minimum: 0
required:
- min_pool_size
- max_pool_size
- max_queue_size
Example:
{
"__default__" : {
"min_pool_size" : 4,
"max_pool_size" : 15,
"max_queue_size" : 200,
"connecting_limit" : 10
},
"postgresql-orders" : {
"min_pool_size" : 8,
"max_pool_size" : 50,
"max_queue_size" : 200,
"connecting_limit" : 8
}
}
Used by components::Postgres .
Dynamic config that controls settings for newly created connections of PostgreSQL driver.
Take note that it overrides the static configuration values of the service!
type: object
additionalProperties: false
properties:
persistent-prepared-statements:
type: boolean
default: true
user-types-enabled:
type: boolean
default: true
max-prepared-cache-size:
type: integer
minimum: 1
default: 5000
recent-errors-threshold:
type: integer
minimum: 1
ignore-unused-query-params:
type: boolean
default: false
Example:
{
"__default__" : {
"persistent-prepared-statements" : true ,
"user-types-enabled" : true ,
"max-prepared-cache-size" : 5000,
"ignore-unused-query-params" : false ,
"recent-errors-threshold" : 2
}
}
Used by components::Postgres .
Dynamic config that enables connlimit_mode: auto for PostgreSQL connections. Auto mode ignores static and dynamic max_connections configs and verifies that the cluster services use max_connections equals to PostgreSQL server's max_connections divided by service instance count.
default: false
schema:
type: boolean
Dynamic config that controls statement metrics settings for specific service.
Dictionary keys can be either the service component names or __default__. The latter configuration will be applied for every PostgreSQL component of the service.
The value of max_statement_metrics controls the maximum size of LRU-cache for named statement metrics. When set to 0 (default) no metrics are being exported.
The exported data can be found as postgresql.statement_timings.
type: object
additionalProperties:
$ref: "#/definitions/StatementMetricsSettings"
definitions:
StatementMetricsSettings:
type: object
additionalProperties: false
properties:
max_statement_metrics:
type: integer
minimum: 0
{
"postgresql-database_name" : {
"max_statement_metrics" : 50
}
}
Used by components::Postgres .
Dynamic config that controls command buffering for specific service.
Command buffering is disabled by default.
type: object
additionalProperties: false
properties:
buffering_enabled:
type: boolean
commands_buffering_threshold:
type: integer
minimum: 1
watch_command_timer_interval_us:
type: integer
minimum: 0
required:
- buffering_enabled
- watch_command_timer_interval_us
{
"buffering_enabled" : true ,
"commands_buffering_threshold" : 10,
"watch_command_timer_interval_us" : 1000
}
Used by components::Redis .
Dynamic config that overrides the default timeouts, number of retries and server selection strategy for redis commands.
type: object
additionalProperties: false
properties:
best_dc_count:
type: integer
max_ping_latency_ms:
type: integer
max_retries:
type: integer
strategy:
enum:
- default
- every_dc
- local_dc_conductor
- nearest_server_ping
type: string
timeout_all_ms:
type: integer
timeout_single_ms:
type: integer
{
"best_dc_count" : 0,
"max_ping_latency_ms" : 0,
"max_retries" : 4,
"strategy" : "default" ,
"timeout_all_ms" : 2000,
"timeout_single_ms" : 500
}
Used by components::Redis .
Dynamic config that controls the metric settings for specific service.
type: object
additionalProperties:
$ref: "#/definitions/MetricsSettings"
definitions:
MetricsSettings:
type: object
additionalProperties: false
properties:
timings-enabled:
type: boolean
default: true
description: enable timings statistics
command-timings-enabled:
type: boolean
default: false
description: enable statistics for individual commands
request-sizes-enabled:
type: boolean
default: false
description: enable request sizes statistics
reply-sizes-enabled:
type: boolean
default: false
description: enable response sizes statistics
{
"redis-database_name" : {
"timings-enabled" : true ,
"command-timings-enabled" : false ,
"request-sizes-enabled" : false ,
"reply-sizes-enabled" : false
}
}
Used by components::Redis .
Dynamic config that controls the redis pubsub metric settings for specific service.
type: object
additionalProperties:
$ref: "#/definitions/PubsubMetricsSettings"
definitions:
PubsubMetricsSettings:
type: object
additionalProperties: false
properties:
per-shard-stats-enabled:
type: boolean
default: true
description: enable collecting statistics by shard
{
"redis-database_name" : {
"per-shard-stats-enabled" : true
}
}
Used by components::Redis .
The same as REDIS_DEFAULT_COMMAND_CONTROL but for subscription clients.
type: object
additionalProperties: false
properties:
timeout_single_ms:
type: integer
minimum: 1
timeout_all_ms:
type: integer
minimum: 1
best_dc_count:
type: integer
minimum: 1
max_ping_latency_ms:
type: integer
minimum: 1
strategy:
type: string
enum:
- default
- every_dc
- local_dc_conductor
- nearest_server_ping
{
"best_dc_count" : 0,
"max_ping_latency_ms" : 0,
"strategy" : "default" ,
"timeout_all_ms" : 2000,
"timeout_single_ms" : 500
}
Used by components::Redis .
Dynamic config that controls the minimal interval between redis subscription clients rebalancing.
minimum: 0
type: integer
default: 30
Used by components::Redis .
Dynamic config that controls if services will wait for connections with redis instances.
type: object
additionalProperties: false
properties:
mode:
type: string
enum:
- no_wait
- master
- slave
- master_or_slave
- master_and_slave
throw_on_fail:
type: boolean
timeout-ms:
type: integer
minimum: 1
x-taxi-cpp-type: std::chrono::milliseconds
required:
- mode
- throw_on_fail
- timeout-ms
{
"mode" : "master_or_slave" ,
"throw_on_fail" : false ,
"timeout-ms" : 11000
}
Used by components::Redis .
Cache update dynamic parameters.
schema:
type: object
additionalProperties:
type: object
additionalProperties: false
properties:
update-interval-ms:
type: integer
minimum: 1
update-jitter-ms:
type: integer
minimum: 0
full-update-interval-ms:
type: integer
minimum: 0
required:
- update-interval-ms
- update-jitter-ms
- full-update-interval-ms
Example:
{
"some-cache-name" : {
"full-update-interval-ms" : 86400000,
"update-interval-ms" : 30000,
"update-jitter-ms" : 1000
}
}
Used by all the caches derived from components::CachingComponentBase .
Controls whether authentication checks are performed in handlers.
Example:
Used by components::Server .
Controls whether the http request task should be cancelled when the deadline received from the client is reached.
Example:
Used by components::Server .
When false, disables deadline propagation in the service. This includes:
reading the task-inherited deadline from HTTP headers and gRPC metadata;
interrupting operations when deadline expires;
propagating the deadline to downstream services and databases.
Example:
Used by components::Server and ugrpc::server::ServerComponent .
Dynamic dump configuration. If the options are set for some dump then those options override the static configuration.
schema:
type: object
additionalProperties:
type: object
additionalProperties: false
properties:
dumps-enabled:
type: boolean
min-dump-interval:
description: dump interval in milliseconds
type: integer
minimum: 0
Example:
{
"some-cache-name" : {
"dumps-enabled" : true ,
"update-interval-ms" : 30000,
"min-dump-interval" : 100000
}
}
Used by dump::Dumper , especially by all the caches derived from components::CachingComponentBase .
Sets whether Stream API should be enabled for HTTP handlers.
Example:
Proxy for all the HTTP and HTTPS clients. Empty string disables proxy usage.
Proxy string may be prefixed with [scheme]:// to specify which kind of proxy is used. Schemes match the libcurl supported ones . A proxy host string can also embed user and password.
Example:
Used by components::HttpClient , affects the behavior of clients::http::Client and all the clients that use it.
Logging per line and file overrides.
default:
force-enabled: []
force-disabled: []
schema:
type: object
additionalProperties: false
required:
- force-enabled
- force-disabled
properties:
force-enabled:
type: array
description: logs to turn on
items:
type: string
force-disabled:
type: array
description: logs to turn off
items:
type: string
Used by components::LoggingConfigurator .
Controls HTTP requests and responses logging.
Example:
Used by components::Server .
Controls whether the logging of HTTP headers in handlers is performed.
Example:
Used by components::Server .
Dynamic config for controlling size and cache entry lifetime of the LRU based caches.
schema:
type: object
additionalProperties:
$ref: '#/definitions/CacheSettings'
definitions:
CacheSettings:
type: object
properties:
size:
type: integer
lifetime-ms:
type: integer
required:
- size
- lifetime-ms
additionalProperties: false
Example:
{
"some-cache-name" : {
"lifetime-ms" : 5000,
"size" : 100000
},
"some-other-cache-name" : {
"lifetime-ms" : 5000,
"size" : 400000
}
}
Used by all the caches derived from cache::LruCacheComponent .
Prefixes or full names of tracing::Span instances to not log.
schema:
type: object
additionalProperties: false
required:
- prefixes
- names
properties:
prefixes:
type: array
items:
type: string
names:
type: array
items:
type: string
Example:
{
"names" : [
"mongo_find"
],
"prefixes" : [
"http" ,
"handler"
]
}
Used by components::LoggingConfigurator and all the logging facilities.
Dynamic config for components::Server congestion control.
schema:
type: object
additionalProperties: false
properties:
min-limit:
type: integer
minimum: 1
description: |
Minimal RPS limit value.
up-rate-percent:
type: number
minimum: 0
exclusiveMinimum: true
description: |
On how many percents increase the RPS limit every second if not in overloaded state.
down-rate-percent:
type: number
minimum: 0
maximum: 100
exclusiveMinimum: true
description: |
On how many percents decrease the RPS limit every second if in overloaded state.
overload-on-seconds:
type: integer
minimum: 1
description: |
How many seconds overload should be higher than `up-level`,
to switch to overload state and start limiting RPS.
overload-off-seconds:
type: integer
minimum: 1
description: |
How many seconds overload should be higher than `down-level`,
to switch from overload state and the RPS limit start increasing.
up-level:
type: integer
minimum: 1
description: |
Overload level that should least for `overload-on-seconds` to switch to
overloaded state.
down-level:
type: integer
minimum: 1
description: |
Overload level that should least for `overload-on-seconds` to
increase RPS limits.
no-limit-seconds:
type: integer
minimum: 1
description: |
Seconds without overload to switch from overloaded state and
remove all the RPS limits.
load-limit-crit-percent:
type: integer
description: |
On reaching this load percent immediately switch to overloaded state.
Example:
{
"down-level" : 8,
"down-rate-percent" : 1,
"load-limit-crit-percent" : 70,
"min-limit" : 2,
"no-limit-seconds" : 120,
"overload-off-seconds" : 8,
"overload-on-seconds" : 8,
"up-level" : 2,
"up-rate-percent" : 1
}
Used by congestion_control::Component .
Controls whether congestion control limiting of RPS is performed (if main task processor is overloaded, then the server starts rejecting some requests).
Example:
Used by congestion_control::Component .
Dynamic config for profiling the coroutine based engine of userver. Dictionary key names are the names of engine::TaskProcessor.
schema:
type: object
properties: {}
additionalProperties:
$ref: '#/definitions/TaskProcessorSettings'
definitions:
TaskProcessorSettings:
type: object
required:
- enabled
- execution-slice-threshold-us
additionalProperties: false
properties:
enabled:
type: boolean
description: |
Set to `true` to turn on the profiling
execution-slice-threshold-us:
type: integer
description: |
Execution threshold of a coroutine on a thread without context switch.
If the threshold is reached then the coroutine is logged, otherwise
does nothing.
minimum: 1
Example:
{
"fs-task-processor" : {
"enabled" : false ,
"execution-slice-threshold-us" : 1000000
},
"main-task-processor" : {
"enabled" : false ,
"execution-slice-threshold-us" : 2000
}
}
Used by components::ManagerControllerComponent .
Controls engine::TaskProcessor quality of service dynamic config.
schema:
type: object
required:
- default-service
additionalProperties: false
properties:
default-service:
type: object
additionalProperties: false
required:
- default-task-processor
properties:
default-task-processor:
type: object
additionalProperties: false
required:
- wait_queue_overload
properties:
wait_queue_overload:
type: object
additionalProperties: false
required:
- time_limit_us
- length_limit
- action
properties:
time_limit_us:
type: integer
minimum: 0
description: |
Wait in queue time after which the `action is applied`.
length_limit:
type: integer
minimum: 0
description: |
Queue size after which the `action` is applied.
action
action:
type: string
enum:
- ignore
- cancel
description: |
Action to perform on tasks on queue overload.
`cancel` - cancels the tasks
sensor_time_limit_us:
type: integer
minimum: 0
description: |
Wait in queue time after which the overload events for
RPS congestion control are generated.
Example:
{
"default-service" : {
"default-task-processor" : {
"wait_queue_overload" : {
"action" : "cancel" ,
"length_limit" : 5000,
"sensor_time_limit_us" : 12000,
"time_limit_us" : 0
}
}
}
}
Used by components::ManagerControllerComponent .
Dynamic config for mapping extension files with http header content type
schema:
type: object
additionalProperties:
type: string
Example:
{
".css" : "text/css" ,
".gif" : "image/gif" ,
".htm" : "text/html" ,
".html" : "text/html" ,
".jpeg" : "image/jpeg" ,
".js" : "application/javascript" ,
".json" : "application/json" ,
".md" : "text/markdown" ,
".png" : "image/png" ,
".svg" : "image/svg+xml" ,
"__default__" : "text/plain"
}
Used by server::handlers::HttpHandlerStatic
