address [parameters]
Defines the address and other parameters\nof a server.\nThe address can be specified as a domain name or IP address,\nwith an optional port, or as a UNIX-domain socket path\nspecified after the “unix:” prefix.\nIf a port is not specified, the port 80 is used.\nA domain name that resolves to several IP addresses defines\nmultiple servers at once.
The following parameters can be defined:
\n\nweight=number
sets the weight of the server, by default, 1.
max_conns=number
limits the maximum number of simultaneous active\nconnections to the proxied server (1.11.5).\nDefault value is zero, meaning there is no limit.\nIf the server group does not reside in the shared memory,\nthe limitation works per each worker process.
\n\nIf idle keepalive connections,\nmultiple workers,\nand the shared memory are enabled,\nthe total number of active and idle connections to the proxied server\nmay exceed the
\n\nmax_connsvalue.Since version 1.5.9 and prior to version 1.11.5,\nthis parameter was available as part of our\ncommercial subscription.
\n
max_fails=number
sets the number of unsuccessful attempts to communicate with the server\nthat should happen in the duration set by the fail_timeout\nparameter to consider the server unavailable for a duration also set by the\nfail_timeout parameter.\nBy default, the number of unsuccessful attempts is set to 1.\nThe zero value disables the accounting of attempts.\nWhat is considered an unsuccessful attempt is defined by the\nproxy_next_upstream,\nfastcgi_next_upstream,\nuwsgi_next_upstream,\nscgi_next_upstream,\nmemcached_next_upstream, and\ngrpc_next_upstream\ndirectives.
fail_timeout=time
sets
\n\nBy default, the parameter is set to 10 seconds.
backup
marks the server as a backup server.\nIt will be passed requests when the primary servers are unavailable.
\n\n\n\nThe parameter cannot be used along with the\n
\nhash,ip_hash, andrandom\nload balancing methods.
down
marks the server as permanently unavailable.
Additionally,\nthe following parameters are available as part of our\ncommercial subscription:
\n\nresolve
monitors changes of the IP addresses\nthat correspond to a domain name of the server,\nand automatically modifies the upstream configuration\nwithout the need of restarting nginx (1.5.12).\nThe server group must reside in the shared memory.
\n\nIn order for this parameter to work,\nthe resolver directive\nmust be specified in the\nhttp block\nor in the corresponding upstream block.
route=string
sets the server route name.
service=name
enables resolving of DNS\nSRV\nrecords and sets the service name (1.9.13).\nIn order for this parameter to work, it is necessary to specify\nthe resolve parameter for the server\nand specify a hostname without a port number.
If the service name does not contain a dot (“.”), then\nthe RFC-compliant name\nis constructed\nand the TCP protocol is added to the service prefix.\nFor example, to look up the\n_http._tcp.backend.example.com SRV record,\nit is necessary to specify the directive:
server backend.example.com service=http resolve;\nIf the service name contains one or more dots, then the name is constructed\nby joining the service prefix and the server name.\nFor example, to look up the _http._tcp.backend.example.com\nand server1.backend.example.com SRV records,\nit is necessary to specify the directives:
server backend.example.com service=_http._tcp resolve;\nserver example.com service=server1.backend resolve;\nHighest-priority SRV records\n(records with the same lowest-number priority value)\nare resolved as primary servers,\nthe rest of SRV records are resolved as backup servers.\nIf the backup parameter is specified for the server,\nhigh-priority SRV records are resolved as backup servers,\nthe rest of SRV records are ignored.
slow_start=time
sets the time during which the server will recover its weight\nfrom zero to a nominal value, when unhealthy server becomes\nhealthy,\nor when the server becomes available after a period of time\nit was considered unavailable.\nDefault value is zero, i.e. slow start is disabled.
\n\nThe parameter cannot be used along with the\n
\nhash,ip_hash, andrandom\nload balancing methods.
drain
puts the server into the “draining” mode (1.13.6).\nIn this mode, only requests bound to the server\nwill be proxied to it.
\n\n\n\nPrior to version 1.13.6,\nthe parameter could be changed only with the\nAPI module.
\n
\n\n" + "description_md": "Defines the *`address`* and other *`parameters`*\nof a server.\nThe address can be specified as a domain name or IP address,\nwith an optional port, or as a UNIX-domain socket path\nspecified after the “`unix:`” prefix.\nIf a port is not specified, the port 80 is used.\nA domain name that resolves to several IP addresses defines\nmultiple servers at once.\n\nThe following parameters can be defined:\n- `weight`=*`number`*\n\n sets the weight of the server, by default, 1.\n- `max_conns`=*`number`*\n\n limits the maximum *`number`* of simultaneous active\n connections to the proxied server (1.11.5).\n Default value is zero, meaning there is no limit.\n If the server group does not reside in the [shared memory](https://nginx.org/en/docs/http/ngx_http_upstream_module.html#zone),\n the limitation works per each worker process.\n > If [idle keepalive](https://nginx.org/en/docs/http/ngx_http_upstream_module.html#keepalive) connections,\n > multiple [workers](https://nginx.org/en/docs/ngx_core_module.html#worker_processes),\n > and the [shared memory](https://nginx.org/en/docs/http/ngx_http_upstream_module.html#zone) are enabled,\n > the total number of active and idle connections to the proxied server\n > may exceed the `max_conns` value.\n \n > Since version 1.5.9 and prior to version 1.11.5,\n > this parameter was available as part of our\n > [commercial subscription](https://nginx.com/products/).\n- `max_fails`=*`number`*\n\n sets the number of unsuccessful attempts to communicate with the server\n that should happen in the duration set by the `fail_timeout`\n parameter to consider the server unavailable for a duration also set by the\n `fail_timeout` parameter.\n By default, the number of unsuccessful attempts is set to 1.\n The zero value disables the accounting of attempts.\n What is considered an unsuccessful attempt is defined by the\n [`proxy_next_upstream`](https://nginx.org/en/docs/http/ngx_http_proxy_module.html#proxy_next_upstream),\n [`fastcgi_next_upstream`](https://nginx.org/en/docs/http/ngx_http_fastcgi_module.html#fastcgi_next_upstream),\n [`uwsgi_next_upstream`](https://nginx.org/en/docs/http/ngx_http_uwsgi_module.html#uwsgi_next_upstream),\n [`scgi_next_upstream`](https://nginx.org/en/docs/http/ngx_http_scgi_module.html#scgi_next_upstream),\n [`memcached_next_upstream`](https://nginx.org/en/docs/http/ngx_http_memcached_module.html#memcached_next_upstream), and\n [`grpc_next_upstream`](https://nginx.org/en/docs/http/ngx_http_grpc_module.html#grpc_next_upstream)\n directives.\n- `fail_timeout`=*`time`*\n\n sets\n - the time during which the specified number of unsuccessful attempts to\n communicate with the server should happen to consider the server unavailable;\n - and the period of time the server will be considered unavailable.\n \n By default, the parameter is set to 10 seconds.\n- `backup`\n\n marks the server as a backup server.\n It will be passed requests when the primary servers are unavailable.\n > The parameter cannot be used along with the\n > [`hash`](https://nginx.org/en/docs/http/ngx_http_upstream_module.html#hash), [`ip_hash`](https://nginx.org/en/docs/http/ngx_http_upstream_module.html#ip_hash), and [`random`](https://nginx.org/en/docs/http/ngx_http_upstream_module.html#random)\n > load balancing methods.\n- `down`\n\n marks the server as permanently unavailable.\n- `resolve`\n\n monitors changes of the IP addresses\n that correspond to a domain name of the server,\n and automatically modifies the upstream configuration\n without the need of restarting nginx (1.5.12).\n The server group must reside in the [shared memory](https://nginx.org/en/docs/http/ngx_http_upstream_module.html#zone).\n \n In order for this parameter to work,\n the `resolver` directive\n must be specified in the\n [http](https://nginx.org/en/docs/http/ngx_http_core_module.html#resolver) block\n or in the corresponding [upstream](https://nginx.org/en/docs/http/ngx_http_upstream_module.html#resolver) block.\n \n \n \n > Prior to version 1.27.3, this parameter was available only as part of our\n > [commercial subscription](https://nginx.com/products/).\n- `service`=*`name`*\n\n enables resolving of DNS\n [SRV](https://datatracker.ietf.org/doc/html/rfc2782)\n records and sets the service *`name`* (1.9.13).\n In order for this parameter to work, it is necessary to specify\n the [`resolve`](https://nginx.org/en/docs/http/ngx_http_upstream_module.html#resolve) parameter for the server\n and specify a hostname without a port number.\n \n If the service name does not contain a dot (“`.`”), then\n the [RFC](https://datatracker.ietf.org/doc/html/rfc2782)-compliant name\n is constructed\n and the TCP protocol is added to the service prefix.\n For example, to look up the\n `_http._tcp.backend.example.com` SRV record,\n it is necessary to specify the directive:\n ```\n server backend.example.com service=http resolve;\n ```\n If the service name contains one or more dots, then the name is constructed\n by joining the service prefix and the server name.\n For example, to look up the `_http._tcp.backend.example.com`\n and `server1.backend.example.com` SRV records,\n it is necessary to specify the directives:\n ```\n server backend.example.com service=_http._tcp resolve;\n server example.com service=server1.backend resolve;\n ```\n \n \n \n Highest-priority SRV records\n (records with the same lowest-number priority value)\n are resolved as primary servers,\n the rest of SRV records are resolved as backup servers.\n If the [`backup`](https://nginx.org/en/docs/http/ngx_http_upstream_module.html#backup) parameter is specified for the server,\n high-priority SRV records are resolved as backup servers,\n the rest of SRV records are ignored.\n \n \n \n > Prior to version 1.27.3, this parameter was available only as part of our\n > [commercial subscription](https://nginx.com/products/).\n\nAdditionally,\nthe following parameters are available as part of our\n[commercial subscription](https://nginx.com/products/):\n- `route`=*`string`*\n\n sets the server route name.\n- `slow_start`=*`time`*\n\n sets the *`time`* during which the server will recover its weight\n from zero to a nominal value, when unhealthy server becomes\n [healthy](https://nginx.org/en/docs/http/ngx_http_upstream_hc_module.html#health_check),\n or when the server becomes available after a period of time\n it was considered [unavailable](https://nginx.org/en/docs/http/ngx_http_upstream_module.html#fail_timeout).\n Default value is zero, i.e. slow start is disabled.\n > The parameter cannot be used along with the\n > [`hash`](https://nginx.org/en/docs/http/ngx_http_upstream_module.html#hash), [`ip_hash`](https://nginx.org/en/docs/http/ngx_http_upstream_module.html#ip_hash), and [`random`](https://nginx.org/en/docs/http/ngx_http_upstream_module.html#random)\n > load balancing methods.\n- `drain`\n\n puts the server into the “draining” mode (1.13.6).\n In this mode, only requests [bound](https://nginx.org/en/docs/http/ngx_http_upstream_module.html#sticky) to the server\n will be proxied to it.\n > Prior to version 1.13.6,\n > the parameter could be changed only with the\n > [API](https://nginx.org/en/docs/http/ngx_http_api_module.html) module.\n\n> If there is only a single server in a group, `max_fails`,\n> `fail_timeout` and `slow_start` parameters\n> are ignored, and such a server will never be considered unavailable.", + "description_html": "If there is only a single server in a group,
\nmax_fails,\nfail_timeoutandslow_startparameters\nare ignored, and such a server will never be considered unavailable.
Defines the address and other parameters\nof a server.\nThe address can be specified as a domain name or IP address,\nwith an optional port, or as a UNIX-domain socket path\nspecified after the “unix:” prefix.\nIf a port is not specified, the port 80 is used.\nA domain name that resolves to several IP addresses defines\nmultiple servers at once.
The following parameters can be defined:
\n\nweight=number
sets the weight of the server, by default, 1.
max_conns=number
limits the maximum number of simultaneous active\nconnections to the proxied server (1.11.5).\nDefault value is zero, meaning there is no limit.\nIf the server group does not reside in the shared memory,\nthe limitation works per each worker process.
\n\nIf idle keepalive connections,\nmultiple workers,\nand the shared memory are enabled,\nthe total number of active and idle connections to the proxied server\nmay exceed the
\n\nmax_connsvalue.Since version 1.5.9 and prior to version 1.11.5,\nthis parameter was available as part of our\ncommercial subscription.
\n
max_fails=number
sets the number of unsuccessful attempts to communicate with the server\nthat should happen in the duration set by the fail_timeout\nparameter to consider the server unavailable for a duration also set by the\nfail_timeout parameter.\nBy default, the number of unsuccessful attempts is set to 1.\nThe zero value disables the accounting of attempts.\nWhat is considered an unsuccessful attempt is defined by the\nproxy_next_upstream,\nfastcgi_next_upstream,\nuwsgi_next_upstream,\nscgi_next_upstream,\nmemcached_next_upstream, and\ngrpc_next_upstream\ndirectives.
fail_timeout=time
sets
\n\nBy default, the parameter is set to 10 seconds.
backup
marks the server as a backup server.\nIt will be passed requests when the primary servers are unavailable.
\n\n\n\nThe parameter cannot be used along with the\n
\nhash,ip_hash, andrandom\nload balancing methods.
down
marks the server as permanently unavailable.
resolve
monitors changes of the IP addresses\nthat correspond to a domain name of the server,\nand automatically modifies the upstream configuration\nwithout the need of restarting nginx (1.5.12).\nThe server group must reside in the shared memory.
\n\nIn order for this parameter to work,\nthe resolver directive\nmust be specified in the\nhttp block\nor in the corresponding upstream block.
\n\nPrior to version 1.27.3, this parameter was available only as part of our\ncommercial subscription.
\n
service=name
enables resolving of DNS\nSRV\nrecords and sets the service name (1.9.13).\nIn order for this parameter to work, it is necessary to specify\nthe resolve parameter for the server\nand specify a hostname without a port number.
If the service name does not contain a dot (“.”), then\nthe RFC-compliant name\nis constructed\nand the TCP protocol is added to the service prefix.\nFor example, to look up the\n_http._tcp.backend.example.com SRV record,\nit is necessary to specify the directive:
server backend.example.com service=http resolve;\nIf the service name contains one or more dots, then the name is constructed\nby joining the service prefix and the server name.\nFor example, to look up the _http._tcp.backend.example.com\nand server1.backend.example.com SRV records,\nit is necessary to specify the directives:
server backend.example.com service=_http._tcp resolve;\nserver example.com service=server1.backend resolve;\nHighest-priority SRV records\n(records with the same lowest-number priority value)\nare resolved as primary servers,\nthe rest of SRV records are resolved as backup servers.\nIf the backup parameter is specified for the server,\nhigh-priority SRV records are resolved as backup servers,\nthe rest of SRV records are ignored.
\n\nPrior to version 1.27.3, this parameter was available only as part of our\ncommercial subscription.
\n
Additionally,\nthe following parameters are available as part of our\ncommercial subscription:
\n\nroute=string
sets the server route name.
slow_start=time
sets the time during which the server will recover its weight\nfrom zero to a nominal value, when unhealthy server becomes\nhealthy,\nor when the server becomes available after a period of time\nit was considered unavailable.\nDefault value is zero, i.e. slow start is disabled.
\n\nThe parameter cannot be used along with the\n
\nhash,ip_hash, andrandom\nload balancing methods.
drain
puts the server into the “draining” mode (1.13.6).\nIn this mode, only requests bound to the server\nwill be proxied to it.
\n\n\n\nPrior to version 1.13.6,\nthe parameter could be changed only with the\nAPI module.
\n
\n\n" }, { "name": "zone", @@ -9728,8 +9728,8 @@ "If there is only a single server in a group,
\nmax_fails,\nfail_timeoutandslow_startparameters\nare ignored, and such a server will never be considered unavailable.
address … [valid=time] [ipv4=on|off] [ipv6=on|off] [status_zone=zone]
Configures name servers used to resolve names of upstream servers\ninto addresses, for example:
\n\nresolver 127.0.0.1 [::1]:5353;\nThe address can be specified as a domain name or IP address,\nwith an optional port.\nIf port is not specified, the port 53 is used.\nName servers are queried in a round-robin fashion.
\n\nBy default, nginx will look up both IPv4 and IPv6 addresses while resolving.\nIf looking up of IPv4 or IPv6 addresses is not desired,\nthe ipv4=off (1.23.1) or\nthe ipv6=off parameter can be specified.
By default, nginx caches answers using the TTL value of a response.\nAn optional valid parameter allows overriding it:
resolver 127.0.0.1 [::1]:5353 valid=30s;\n\n\n\nTo prevent DNS spoofing, it is recommended\nconfiguring DNS servers in a properly secured trusted local network.
\n
The optional status_zone parameter\nenables\ncollection\nof DNS server statistics of requests and responses\nin the specified zone.
\n\n" + "description_md": "Configures name servers used to resolve names of upstream servers\ninto addresses, for example:\n```\nresolver 127.0.0.1 [::1]:5353;\n```\nThe address can be specified as a domain name or IP address,\nwith an optional port.\nIf port is not specified, the port 53 is used.\nName servers are queried in a round-robin fashion.\n\nBy default, nginx will look up both IPv4 and IPv6 addresses while resolving.\nIf looking up of IPv4 or IPv6 addresses is not desired,\nthe `ipv4=off` (1.23.1) or\nthe `ipv6=off` parameter can be specified.\n\nBy default, nginx caches answers using the TTL value of a response.\nAn optional `valid` parameter allows overriding it:\n```\nresolver 127.0.0.1 [::1]:5353 valid=30s;\n```\n> To prevent DNS spoofing, it is recommended\n> configuring DNS servers in a properly secured trusted local network.\n\nThe optional `status_zone` parameter (1.17.5)\nenables\n[collection](https://nginx.org/en/docs/http/ngx_http_api_module.html#resolvers_)\nof DNS server statistics of requests and responses\nin the specified *`zone`*.\nThe parameter is available as part of our\n[commercial subscription](https://nginx.com/products/).\n\n> Since version 1.17.5 and prior to version 1.27.3,\n> this directive was available only as part of our\n> [commercial subscription](https://nginx.com/products/).", + "description_html": "This directive is available as part of our\ncommercial subscription.
\n
Configures name servers used to resolve names of upstream servers\ninto addresses, for example:
\n\nresolver 127.0.0.1 [::1]:5353;\nThe address can be specified as a domain name or IP address,\nwith an optional port.\nIf port is not specified, the port 53 is used.\nName servers are queried in a round-robin fashion.
\n\nBy default, nginx will look up both IPv4 and IPv6 addresses while resolving.\nIf looking up of IPv4 or IPv6 addresses is not desired,\nthe ipv4=off (1.23.1) or\nthe ipv6=off parameter can be specified.
By default, nginx caches answers using the TTL value of a response.\nAn optional valid parameter allows overriding it:
resolver 127.0.0.1 [::1]:5353 valid=30s;\n\n\n\nTo prevent DNS spoofing, it is recommended\nconfiguring DNS servers in a properly secured trusted local network.
\n
The optional status_zone parameter (1.17.5)\nenables\ncollection\nof DNS server statistics of requests and responses\nin the specified zone.\nThe parameter is available as part of our\ncommercial subscription.
\n\n" }, { "name": "resolver_timeout", @@ -9744,8 +9744,8 @@ "Since version 1.17.5 and prior to version 1.27.3,\nthis directive was available only as part of our\ncommercial subscription.
\n
time
Sets a timeout for name resolution, for example:
\n\nresolver_timeout 5s;\n\n\n" + "description_md": "Sets a timeout for name resolution, for example:\n```\nresolver_timeout 5s;\n```\n\n> Since version 1.17.5 and prior to version 1.27.3,\n> this directive was available only as part of our\n> [commercial subscription](https://nginx.com/products/).", + "description_html": "This directive is available as part of our\ncommercial subscription.
\n
Sets a timeout for name resolution, for example:
\n\nresolver_timeout 5s;\n\n\n" }, { "name": "sticky", @@ -15858,8 +15858,8 @@ "Since version 1.17.5 and prior to version 1.27.3,\nthis directive was available only as part of our\ncommercial subscription.
\n
address [parameters]
Defines the address and other parameters\nof a server.\nThe address can be specified as a domain name or IP address\nwith an obligatory port, or as a UNIX-domain socket path\nspecified after the “unix:” prefix.\nA domain name that resolves to several IP addresses defines\nmultiple servers at once.
The following parameters can be defined:
\n\nweight=number
sets the weight of the server, by default, 1.
max_conns=number
limits the maximum number of simultaneous\nconnections to the proxied server (1.11.5).\nDefault value is zero, meaning there is no limit.\nIf the server group does not reside in the shared memory,\nthe limitation works per each worker process.
\n\nPrior to version 1.11.5, this parameter was available as part of our\ncommercial subscription.
\n
max_fails=number
sets the number of unsuccessful attempts to communicate with the server\nthat should happen in the duration set by the fail_timeout\nparameter to consider the server unavailable for a duration also set by the\nfail_timeout parameter.\nBy default, the number of unsuccessful attempts is set to 1.\nThe zero value disables the accounting of attempts.\nHere, an unsuccessful attempt is an error or timeout\nwhile establishing a connection with the server.
fail_timeout=time
sets
\n\nBy default, the parameter is set to 10 seconds.
backup
marks the server as a backup server.\nConnections to the backup server will be passed\nwhen the primary servers are unavailable.
\n\n\n\nThe parameter cannot be used along with the\n
\nhashandrandomload balancing methods.
down
marks the server as permanently unavailable.
Additionally,\nthe following parameters are available as part of our\ncommercial subscription:
\n\nresolve
monitors changes of the IP addresses\nthat correspond to a domain name of the server,\nand automatically modifies the upstream configuration\nwithout the need of restarting nginx.\nThe server group must reside in the shared memory.
\n\nIn order for this parameter to work,\nthe resolver directive\nmust be specified in the\nstream block\nor in the corresponding upstream block.
service=name
enables resolving of DNS\nSRV\nrecords and sets the service name (1.9.13).\nIn order for this parameter to work, it is necessary to specify\nthe resolve parameter for the server\nand specify a hostname without a port number.
If the service name does not contain a dot (“.”), then\nthe RFC-compliant name\nis constructed\nand the TCP protocol is added to the service prefix.\nFor example, to look up the\n_http._tcp.backend.example.com SRV record,\nit is necessary to specify the directive:
server backend.example.com service=http resolve;\nIf the service name contains one or more dots, then the name is constructed\nby joining the service prefix and the server name.\nFor example, to look up the _http._tcp.backend.example.com\nand server1.backend.example.com SRV records,\nit is necessary to specify the directives:
server backend.example.com service=_http._tcp resolve;\nserver example.com service=server1.backend resolve;\nHighest-priority SRV records\n(records with the same lowest-number priority value)\nare resolved as primary servers,\nthe rest of SRV records are resolved as backup servers.\nIf the backup parameter is specified for the server,\nhigh-priority SRV records are resolved as backup servers,\nthe rest of SRV records are ignored.
slow_start=time
sets the time during which the server will recover its weight\nfrom zero to a nominal value, when unhealthy server becomes\nhealthy,\nor when the server becomes available after a period of time\nit was considered unavailable.\nDefault value is zero, i.e. slow start is disabled.
\n\nThe parameter cannot be used along with the\n
\nhashandrandomload balancing methods.
\n\n" + "description_md": "Defines the *`address`* and other *`parameters`*\nof a server.\nThe address can be specified as a domain name or IP address\nwith an obligatory port, or as a UNIX-domain socket path\nspecified after the “`unix:`” prefix.\nA domain name that resolves to several IP addresses defines\nmultiple servers at once.\n\nThe following parameters can be defined:\n- `weight`=*`number`*\n\n sets the weight of the server, by default, 1.\n- `max_conns`=*`number`*\n\n limits the maximum *`number`* of simultaneous\n connections to the proxied server (1.11.5).\n Default value is zero, meaning there is no limit.\n If the server group does not reside in the [shared memory](https://nginx.org/en/docs/stream/ngx_stream_upstream_module.html#zone),\n the limitation works per each worker process.\n > Prior to version 1.11.5, this parameter was available as part of our\n > [commercial subscription](https://nginx.com/products/).\n- `max_fails`=*`number`*\n\n sets the number of unsuccessful attempts to communicate with the server\n that should happen in the duration set by the `fail_timeout`\n parameter to consider the server unavailable for a duration also set by the\n `fail_timeout` parameter.\n By default, the number of unsuccessful attempts is set to 1.\n The zero value disables the accounting of attempts.\n Here, an unsuccessful attempt is an error or timeout\n while establishing a connection with the server.\n- `fail_timeout`=*`time`*\n\n sets\n - the time during which the specified number of unsuccessful attempts to\n communicate with the server should happen to consider the server unavailable;\n - and the period of time the server will be considered unavailable.\n \n By default, the parameter is set to 10 seconds.\n- `backup`\n\n marks the server as a backup server.\n Connections to the backup server will be passed\n when the primary servers are unavailable.\n > The parameter cannot be used along with the\n > [`hash`](https://nginx.org/en/docs/stream/ngx_stream_upstream_module.html#hash) and [`random`](https://nginx.org/en/docs/stream/ngx_stream_upstream_module.html#random) load balancing methods.\n- `down`\n\n marks the server as permanently unavailable.\n- `resolve`\n\n monitors changes of the IP addresses\n that correspond to a domain name of the server,\n and automatically modifies the upstream configuration\n without the need of restarting nginx.\n The server group must reside in the [shared memory](https://nginx.org/en/docs/stream/ngx_stream_upstream_module.html#zone).\n \n In order for this parameter to work,\n the `resolver` directive\n must be specified in the\n [stream](https://nginx.org/en/docs/stream/ngx_stream_core_module.html#resolver) block\n or in the corresponding [upstream](https://nginx.org/en/docs/stream/ngx_stream_upstream_module.html#resolver) block.\n \n \n \n > Prior to version 1.27.3, this parameter was available only as part of our\n > [commercial subscription](https://nginx.com/products/).\n- `service`=*`name`*\n\n enables resolving of DNS\n [SRV](https://datatracker.ietf.org/doc/html/rfc2782)\n records and sets the service *`name`* (1.9.13).\n In order for this parameter to work, it is necessary to specify\n the [`resolve`](https://nginx.org/en/docs/stream/ngx_stream_upstream_module.html#resolve) parameter for the server\n and specify a hostname without a port number.\n \n If the service name does not contain a dot (“`.`”), then\n the [RFC](https://datatracker.ietf.org/doc/html/rfc2782)-compliant name\n is constructed\n and the TCP protocol is added to the service prefix.\n For example, to look up the\n `_http._tcp.backend.example.com` SRV record,\n it is necessary to specify the directive:\n ```\n server backend.example.com service=http resolve;\n ```\n If the service name contains one or more dots, then the name is constructed\n by joining the service prefix and the server name.\n For example, to look up the `_http._tcp.backend.example.com`\n and `server1.backend.example.com` SRV records,\n it is necessary to specify the directives:\n ```\n server backend.example.com service=_http._tcp resolve;\n server example.com service=server1.backend resolve;\n ```\n \n \n \n Highest-priority SRV records\n (records with the same lowest-number priority value)\n are resolved as primary servers,\n the rest of SRV records are resolved as backup servers.\n If the [`backup`](https://nginx.org/en/docs/stream/ngx_stream_upstream_module.html#backup) parameter is specified for the server,\n high-priority SRV records are resolved as backup servers,\n the rest of SRV records are ignored.\n \n \n \n > Prior to version 1.27.3, this parameter was available only as part of our\n > [commercial subscription](https://nginx.com/products/).\n\nAdditionally,\nthe following parameters are available as part of our\n[commercial subscription](https://nginx.com/products/):\n- `slow_start`=*`time`*\n\n sets the *`time`* during which the server will recover its weight\n from zero to a nominal value, when unhealthy server becomes\n [healthy](https://nginx.org/en/docs/stream/ngx_stream_upstream_hc_module.html#health_check),\n or when the server becomes available after a period of time\n it was considered [unavailable](https://nginx.org/en/docs/stream/ngx_stream_upstream_module.html#fail_timeout).\n Default value is zero, i.e. slow start is disabled.\n > The parameter cannot be used along with the\n > [`hash`](https://nginx.org/en/docs/stream/ngx_stream_upstream_module.html#hash) and [`random`](https://nginx.org/en/docs/stream/ngx_stream_upstream_module.html#random) load balancing methods.\n\n> If there is only a single server in a group, `max_fails`,\n> `fail_timeout` and `slow_start` parameters\n> are ignored, and such a server will never be considered unavailable.", + "description_html": "If there is only a single server in a group,
\nmax_fails,\nfail_timeoutandslow_startparameters\nare ignored, and such a server will never be considered unavailable.
Defines the address and other parameters\nof a server.\nThe address can be specified as a domain name or IP address\nwith an obligatory port, or as a UNIX-domain socket path\nspecified after the “unix:” prefix.\nA domain name that resolves to several IP addresses defines\nmultiple servers at once.
The following parameters can be defined:
\n\nweight=number
sets the weight of the server, by default, 1.
max_conns=number
limits the maximum number of simultaneous\nconnections to the proxied server (1.11.5).\nDefault value is zero, meaning there is no limit.\nIf the server group does not reside in the shared memory,\nthe limitation works per each worker process.
\n\nPrior to version 1.11.5, this parameter was available as part of our\ncommercial subscription.
\n
max_fails=number
sets the number of unsuccessful attempts to communicate with the server\nthat should happen in the duration set by the fail_timeout\nparameter to consider the server unavailable for a duration also set by the\nfail_timeout parameter.\nBy default, the number of unsuccessful attempts is set to 1.\nThe zero value disables the accounting of attempts.\nHere, an unsuccessful attempt is an error or timeout\nwhile establishing a connection with the server.
fail_timeout=time
sets
\n\nBy default, the parameter is set to 10 seconds.
backup
marks the server as a backup server.\nConnections to the backup server will be passed\nwhen the primary servers are unavailable.
\n\n\n\nThe parameter cannot be used along with the\n
\nhashandrandomload balancing methods.
down
marks the server as permanently unavailable.
resolve
monitors changes of the IP addresses\nthat correspond to a domain name of the server,\nand automatically modifies the upstream configuration\nwithout the need of restarting nginx.\nThe server group must reside in the shared memory.
\n\nIn order for this parameter to work,\nthe resolver directive\nmust be specified in the\nstream block\nor in the corresponding upstream block.
\n\nPrior to version 1.27.3, this parameter was available only as part of our\ncommercial subscription.
\n
service=name
enables resolving of DNS\nSRV\nrecords and sets the service name (1.9.13).\nIn order for this parameter to work, it is necessary to specify\nthe resolve parameter for the server\nand specify a hostname without a port number.
If the service name does not contain a dot (“.”), then\nthe RFC-compliant name\nis constructed\nand the TCP protocol is added to the service prefix.\nFor example, to look up the\n_http._tcp.backend.example.com SRV record,\nit is necessary to specify the directive:
server backend.example.com service=http resolve;\nIf the service name contains one or more dots, then the name is constructed\nby joining the service prefix and the server name.\nFor example, to look up the _http._tcp.backend.example.com\nand server1.backend.example.com SRV records,\nit is necessary to specify the directives:
server backend.example.com service=_http._tcp resolve;\nserver example.com service=server1.backend resolve;\nHighest-priority SRV records\n(records with the same lowest-number priority value)\nare resolved as primary servers,\nthe rest of SRV records are resolved as backup servers.\nIf the backup parameter is specified for the server,\nhigh-priority SRV records are resolved as backup servers,\nthe rest of SRV records are ignored.
\n\nPrior to version 1.27.3, this parameter was available only as part of our\ncommercial subscription.
\n
Additionally,\nthe following parameters are available as part of our\ncommercial subscription:
\n\nslow_start=time
sets the time during which the server will recover its weight\nfrom zero to a nominal value, when unhealthy server becomes\nhealthy,\nor when the server becomes available after a period of time\nit was considered unavailable.\nDefault value is zero, i.e. slow start is disabled.
\n\nThe parameter cannot be used along with the\n
\nhashandrandomload balancing methods.
\n\n" }, { "name": "zone", @@ -15970,8 +15970,8 @@ "If there is only a single server in a group,
\nmax_fails,\nfail_timeoutandslow_startparameters\nare ignored, and such a server will never be considered unavailable.
address … [valid=time] [ipv4=on|off] [ipv6=on|off] [status_zone=zone]
Configures name servers used to resolve names of upstream servers\ninto addresses, for example:
\n\nresolver 127.0.0.1 [::1]:5353;\nThe address can be specified as a domain name or IP address,\nwith an optional port.\nIf port is not specified, the port 53 is used.\nName servers are queried in a round-robin fashion.
\n\nBy default, nginx will look up both IPv4 and IPv6 addresses while resolving.\nIf looking up of IPv4 or IPv6 addresses is not desired,\nthe ipv4=off (1.23.1) or\nthe ipv6=off parameter can be specified.
By default, nginx caches answers using the TTL value of a response.\nThe optional valid parameter allows overriding it:
resolver 127.0.0.1 [::1]:5353 valid=30s;\n\n\n\nTo prevent DNS spoofing, it is recommended\nconfiguring DNS servers in a properly secured trusted local network.
\n
The optional status_zone parameter\nenables\ncollection\nof DNS server statistics of requests and responses\nin the specified zone.
\n\n" + "description_md": "Configures name servers used to resolve names of upstream servers\ninto addresses, for example:\n```\nresolver 127.0.0.1 [::1]:5353;\n```\nThe address can be specified as a domain name or IP address,\nwith an optional port.\nIf port is not specified, the port 53 is used.\nName servers are queried in a round-robin fashion.\n\nBy default, nginx will look up both IPv4 and IPv6 addresses while resolving.\nIf looking up of IPv4 or IPv6 addresses is not desired,\nthe `ipv4=off` (1.23.1) or\nthe `ipv6=off` parameter can be specified.\n\nBy default, nginx caches answers using the TTL value of a response.\nThe optional `valid` parameter allows overriding it:\n```\nresolver 127.0.0.1 [::1]:5353 valid=30s;\n```\n> To prevent DNS spoofing, it is recommended\n> configuring DNS servers in a properly secured trusted local network.\n\nThe optional `status_zone` parameter (1.17.5)\nenables\n[collection](https://nginx.org/en/docs/http/ngx_http_api_module.html#resolvers_)\nof DNS server statistics of requests and responses\nin the specified *`zone`*.\nThe parameter is available as part of our\n[commercial subscription](https://nginx.com/products/).\n\n> Since version 1.17.5 and prior to version 1.27.3,\n> this directive was available only as part of our\n> [commercial subscription](https://nginx.com/products/).", + "description_html": "This directive is available as part of our\ncommercial subscription.
\n
Configures name servers used to resolve names of upstream servers\ninto addresses, for example:
\n\nresolver 127.0.0.1 [::1]:5353;\nThe address can be specified as a domain name or IP address,\nwith an optional port.\nIf port is not specified, the port 53 is used.\nName servers are queried in a round-robin fashion.
\n\nBy default, nginx will look up both IPv4 and IPv6 addresses while resolving.\nIf looking up of IPv4 or IPv6 addresses is not desired,\nthe ipv4=off (1.23.1) or\nthe ipv6=off parameter can be specified.
By default, nginx caches answers using the TTL value of a response.\nThe optional valid parameter allows overriding it:
resolver 127.0.0.1 [::1]:5353 valid=30s;\n\n\n\nTo prevent DNS spoofing, it is recommended\nconfiguring DNS servers in a properly secured trusted local network.
\n
The optional status_zone parameter (1.17.5)\nenables\ncollection\nof DNS server statistics of requests and responses\nin the specified zone.\nThe parameter is available as part of our\ncommercial subscription.
\n\n" }, { "name": "resolver_timeout", @@ -15986,8 +15986,8 @@ "Since version 1.17.5 and prior to version 1.27.3,\nthis directive was available only as part of our\ncommercial subscription.
\n
time
Sets a timeout for name resolution, for example:
\n\nresolver_timeout 5s;\n\n\n" + "description_md": "Sets a timeout for name resolution, for example:\n```\nresolver_timeout 5s;\n```\n\n> Since version 1.17.5 and prior to version 1.27.3,\n> this directive was available only as part of our\n> [commercial subscription](https://nginx.com/products/).", + "description_html": "This directive is available as part of our\ncommercial subscription.
\n
Sets a timeout for name resolution, for example:
\n\nresolver_timeout 5s;\n\n\n" } ], "variables": [ @@ -16385,5 +16385,5 @@ ] } ], - "version": "https://github.com/nginx/nginx.org/commit/3aad6b0f8f073986c4114a57cb42395a0d1b18ad" + "version": "https://github.com/nginx/nginx.org/commit/c9f5a565b315bf58bd564ab028ba9173d4208cc3" }Since version 1.17.5 and prior to version 1.27.3,\nthis directive was available only as part of our\ncommercial subscription.
\n