haproxy job from haproxy/14.1.0+2.8.10
The HAProxy server can be used to terminate SSL in front of the Routers. Each HAProxy instance should point to multiple Routers.
Github source:
3204fc8
or
master branch
Properties¶
ha_proxy
¶
accept_proxy
¶Turned off by default. Enforces the use of the PROXY protocol for all incoming connections to all frontends, with the exception of local requests to the health check endpoint, since Monit does not support PROXY protocol. When enabled, standard TCP connections to these ports no longer work.
- Default
false
additional_unrestricted_volumes
¶Map of Additional Unrestricted volumes to added so HA Proxy to give the ability to load external volumes under BPM. These will need to be provided in a Map based on bpm unrestricted volume spec
always_allow_body_http10
¶Always allow a body to be sent when using HTTP/1.0. By default HAProxy denies GET/HEAD/DELETE requests with a body when using HTTP/1.0 due to potential request smuggling attacks. See https://github.com/haproxy/haproxy/commit/e136bd12a32970bc90d862d5fe09ea1952b62974
- Default
false
backend_ca_file
¶Optional SSL CA certificate chain (PEM file) concatenated together for backend SSL servers, only used when one of the
backend_ssl
options is set toverify
backend_config
¶Raw HAProxy config that will be added to the default HTTP + routed HTTP backend definitions, provided either as a multiline text blob or as an array of lines.
backend_crt
¶Provides client certificate to backend server to do mutual ssl. Note this only configures the client cert for HTTP backends configured via the backend_servers property or through BOSH links. It is not used with backend servers configured via routed_backend_servers or TCP backends
- Example
|+ -----BEGIN CERTIFICATE----- ****** -----END CERTIFICATE----- -----BEGIN PRIVATE KEY----- ****** -----END PRIVATE KEY-----
backend_health_fall
¶Number of consecutive unsuccessful health checks required before the server is considered unhealthy from a healthy state. The default value of 3 matches the default if the parameter is undefined. This parameter will be ignored if ha_proxy.backend_use_http_health is false.
- Default
3
backend_health_rise
¶Number of consecutive successful health checks required before the server is considered healthy from an unhealthy state. The default value of 2 matches the default if the parameter is undefined. This parameter will be ignored if ha_proxy.backend_use_http_health is false.
- Default
2
backend_http_health_port
¶Port to check the backend health on
- Default
8080
backend_http_health_uri
¶URI to request when checking the health of a backend
- Default
/health
backend_match_http_protocol
¶Uses the same version of HTTP for backend connections that was used for frontend connections (ie HTTP 1.1 or HTTP 2). Ignores the value of enable_http2. HTTP2 backend connections require that
ha_proxy.backend_ssl
is notoff
.
- Default
false
backend_port
¶Listening port for Router
- Default
80
backend_prefer_local_az
¶Prefer backend servers which are located on the same availability zone. Note that this only affects servers provided via the http_backend link property. Servers provided via the tcp backend_link will automatically prefer the local AZ.
- Default
false
backend_servers
¶Array of the router IPs acting as the HTTP/TCP backends (should include servers all Availability Zones being used)
- Default
[]
backend_ssl
¶Optionally enable SSL verification for backend servers, one of
verify
,noverify
, any other value assumes no ssl backend. Settingverify
requiresha_proxy.backend_ca_file
key to be set. Note thatoff
will disable all backend HTTP2 support regardless of other properties.
- Default
"off"
backend_ssl_verifyhost
¶Optional hostname to verify in the x509 certificate subject for SSL-enabled backend servers. Requires
ha_proxy.backend_ssl
is set toverify
when using this.
backend_use_http_health
¶Enables HTTP based health checks for the default backend
- Default
false
binding_ip
¶If there are multiple ethernet interfaces, specify which one to bind. Set to
::
to bind to all IPv6 interfaces (no IPv4). IPv6 must be enabled on the HAProxy VM in the deployment manifest.
- Default
""
block_all
¶Optionally block all incoming traffic to http(s). Use in conjunction with whitelist.
- Default
false
buffer_size_bytes
¶Buffer size to use for requests, any requests larger than this (large cookies or query strings) will result in a gateway error
- Default
16384
cidr_blacklist
¶List of CIDRs to block for http(s). Format is string array of CIDRs or single string of base64 encoded gzip.
- Example
cidr_blacklist: - 10.0.0.0/8 - 192.168.2.0/24
cidr_whitelist
¶List of CIDRs to allow for http(s). Format is string array of CIDRs or single string of base64 encoded gzip. Note that unless ha_proxy.block_all is true, non-whitelisted traffic will still be allowed, provided that traffic is not also blacklisted
- Example
cidr_whitelist: - 172.168.4.1/32 - 10.2.0.0/16
cidrs_in_file
¶List of cidrs that will be placed in /var/vcap/jobs/haproxy/config/cidrs/. Useful for acl’s that reference a long list of cidrs (invoke the file with -f /var/vcap/jobs/haproxy/config/cidrs/).
- Example
cidrs_in_file: - cidrs: - 5.22.1.3 - 5.22.12.3 name: sample_cidrs
client_ca_file
¶path for CA certs to validate client certificate
- Example
|+ -----BEGIN CERTIFICATE----- ****** -----END CERTIFICATE----- -----BEGIN PRIVATE KEY----- ****** -----END PRIVATE KEY-----
client_cert
¶Enable haproxy mutual auth
- Default
false
client_cert_ignore_err
¶Error code(s) to ignore from verifying a client cert during a mutual ssl handshake, in a comma-separated list. For example, 2 is if it cannot get the issuer certificate, 10 if the certificate has expired and 18 if the certificate is self-signed. The keyword ‘all’ will ignore all possible errors. Note that the errors will be ignored on both the certificate and the CA verification. See the openssl verify documentation [https://www.openssl.org/docs/manmaster/man3/X509_STORE_CTX_get_error.html] for a full list of all error codes and their meanings. See https://github.com/openssl/openssl/blob/master/include/openssl/x509_vfy.h.in#L206 for a mapping of error codes to symbols.
- Example
2,10,18
client_revocation_list
¶provide a list of revocation certs
client_timeout
¶Timeout (in floating point seconds) used on connections from a client to haproxy that have gone inactive
- Default
30
compress_types
¶If this property is set, gzip compression will be activated for the mime types named in this property. definition like ‘text/html text/plain text/css’
- Default
""
config_mode
¶‘auto’ - utilizes raw_config if defined and mixes it with raw_blocks; otherwise, it uses traditional configuration mixed with raw_blocks ‘raw_blocks_only’ - uses only raw_blocks, ignoring other configurations.
- Default
auto
connect_timeout
¶Timeout (in floating point seconds) used on connections from haproxy to a backend, while waiting for the TCP handshake to complete + connection to establish
- Default
5
connections_rate_limit
¶
block
¶Whether or not to block connections. See docs/rate_limiting.md
- Default
false
connections
¶How many connections are allowed in the given time window from one IP address. See docs/rate_limiting.md
table_size
¶Size of the stick table in which the IPs and counters are stored. See docs/rate_limiting.md
window_size
¶Window size for counting connections. See docs/rate_limiting.md
crt_list
¶Array of private keys and certificates used for TLS handshakes with downstream clients. Each element in the array is an object containing at least the field ‘ssl_pem’. The field ‘ssl_pem’ itself is either an object containing fields ‘cert_chain’ and ‘private_key’, or a single string containing the cert chain and the private key. The following fields are optional: - ‘alpn’ (a optional array of strings). If both HTTP/2 and HTTP/1.1 are expected to be supported, both versions can be advertised, in order of preference - ‘client_ca_file’ (replaces ha_proxy.client_ca_file) - ‘verify’ (allowed values: [none|optional|required]) - ‘ssl_ciphers’ (overrides ha_proxy.ssl_ciphers) - ‘ssl_ciphersuites’ (overrides ha_proxy.ssl_ciphersuites) - ‘ssl_min_version’ (allowed values: [SSLv3 | TLSv1.0 | TLSv1.1 | TLSv1.2 | TLSv1.3]) - ‘ssl_max_version’ (allowed values: [SSLv3 | TLSv1.0 | TLSv1.1 | TLSv1.2 | TLSv1.3]) - ‘client_revocation_list’ (replaces ha_proxy.client_revocation_list) - ‘snifilter’ (either a string or an array of strings) The global option ha_proxy.client_cert has to be set to ‘true’, if there are crt_list entries with mutual auth configuration (‘client_ca_file’, ‘client_revocation_list’ and ‘verify’!=‘none’) To avoid confusing configurations, it’s not allowed to specify ‘client_ca_file’ and ‘client_revocation_list’ both globally AND in crt_list entries.
- Example
crt_list: - ssl_pem: |+ -----BEGIN CERTIFICATE----- -----END CERTIFICATE----- -----BEGIN CERTIFICATE----- -----END CERTIFICATE----- -----BEGIN RSA PRIVATE KEY----- -----END RSA PRIVATE KEY----- - alpn: - h2 - http/1.1 client_ca_file: |+ -----BEGIN CERTIFICATE----- -----END CERTIFICATE----- -----BEGIN CERTIFICATE----- -----END CERTIFICATE----- client_revocation_list: |+ -----BEGIN X509 CRL----- -----END X509 CRL----- -----BEGIN X509 CRL----- -----END X509 CRL----- snifilter: - '*.domain.tld' - '!secure.domain.tld' ssl_ciphers: AES:ALL:!aNULL:!eNULL:+RC4:@STRENGTH ssl_ciphersuites: TLS_AES_128_GCM_SHA256:TLS_AES_256_GCM_SHA384:TLS_CHACHA20_POLY1305_SHA256 ssl_max_version: TLSv1.3 ssl_min_version: TLSv1.2 ssl_pem: cert_chain: |+ -----BEGIN CERTIFICATE----- -----END CERTIFICATE----- -----BEGIN CERTIFICATE----- -----END CERTIFICATE----- private_key: |+ -----BEGIN RSA PRIVATE KEY----- -----END RSA PRIVATE KEY----- verify: required
custom_http_error_files
¶A map of status codes to errorfile contents
- Default
"503": |+ HTTP/1.1 503 Service Unavailable Cache-Control: no-cache Connection: close Content-Type: text/html <html><body><h1>503 Service Unavailable</h1> No server is available to handle this request. </body></html>
default_config
¶Raw HAProxy config that will be added to the HA proxy default section, provided either as a multiline text blob or as an array of lines.
default_dh_param
¶Maximum size of DH params when generating ephemeral keys during key exchange
- Default
2048
disable_backend_http2_websockets
¶Forward websockets to the backend servers using HTTP/1.1, never HTTP/2. Does not apply to custom routed_backend_servers. Works around https://github.com/cloudfoundry/routing-release/issues/230. Overrides backend_match_http_protocol for websockets.
- Default
false
disable_domain_fronting
¶Must be one of true, false, or “mtls_only” If set to true, it will prevent clients from setting a host header different from the SNI value for HTTPS and WSS (secured websockets) connections. This is called domain fronting and is mostly used by CDNs. If domain fronting is disabled, such requests will result in a 421 Misdirected Request error. If set to “mtls_only”, the host header will only be checked against the SNI for mtls connections Example curl -H “Host: bob.com” https://alice.com <– This will result in a 421 Misdirected Request
- Default
false
disable_health_check_proxy
¶Disables the use of the PROXY protocol for health checks. Only applies if
ha_proxy.accept_proxy
is enabled.
- Default
false
disable_http
¶Disable port 80 traffic
- Default
false
disable_monit_health_check_http
¶The HAProxy health check endpoint returns a healthy status if at least one backend server is responding. By default when enable_health_check_http: true, Bosh will consider the HAProxy VM unhealthy if the HAProxy health check returns an unhealthy status. In some cases this might not be desired, for example when deploying HAProxy before deploying the backend servers. To prevent Bosh from considering the HAProxy VM unhealthy when all backend servers are unhealthy set disable_monit_health_check_http: true. Note that this flag is ignored unless enable_health_check_http: true.
- Default
false
disable_tcp_accept_proxy
¶Disables the PROXY protocol on tcp backends. Only applies if
ha_proxy.accept_proxy
is enabled.
- Default
false
disable_tls_10
¶Disable TLS 1.0 in HA Proxy. Use ‘ssl_min_ver’ and ‘ssl_max_ver’ instead.
- Default
false
disable_tls_11
¶Disable TLS 1.1 in HA Proxy. Use ‘ssl_min_ver’ and ‘ssl_max_ver’ instead.
- Default
false
disable_tls_12
¶Disable TLS 1.2 in HA Proxy. Use ‘ssl_min_ver’ and ‘ssl_max_ver’ instead.
- Default
false
disable_tls_13
¶Disable TLS 1.3 in HA Proxy. Use ‘ssl_min_ver’ and ‘ssl_max_ver’ instead.
- Default
false
disable_tls_tickets
¶Improve (Perfect) Forward Secrecy by disabling TLS tickets.
- Default
true
dns_hold
¶DNS Hold time
- Default
10s
drain_enable
¶Send SIGUSR1 signal to all haproxy processes in a drain script in order to gracefully shutdown
- Default
false
drain_frontend_grace_time
¶Time in seconds after health checks have been shut down until SIGUSR1 signal is sent to make the frontends stop accepting connections
- Default
0
drain_timeout
¶Time in seconds after SIGUSR1 signal is sent in the drain script until monit stops the processes
- Default
30
enable_4443
¶Enables port 4443 for backwards compatibility with WSS-based apps using the old CF haproxy. If true you must provide a valid SSL config via ssl_pem or crt_list
- Default
false
enable_health_check_http
¶Optionally enable http health-check on
haproxy_ip:8080/health
. It shows200 OK
if >0 backend servers are up. If used with ext_crt_list_timeout you should make sure that the deployment canary_watch_time and update_watch_time are configured to wait at least the number of seconds defined by ext_crt_list_timeout.
- Default
false
enable_http2
¶Enables ingress (frontend) and egress (backend) HTTP/2 ALPN negotiation. Egress (backend) HTTP protocol version may be overridden by
ha_proxy.backend_ssl
,ha_proxy.disable_backend_http2_websockets
andha_proxy.backend_match_http_protocol
.
- Default
false
enable_redispatch
¶When enabled, HAProxy will try to connect to another server if a connect attempt fails. Best used in conjunction with retries.
- Default
false
ext_crt_list
¶A flag denoting the use of additional certificates from external sources. If set to true the contents of an external crt-list file located at
ha_proxy.ext_crt_list_file
are added to the crt-list described by theha_proxy.crt_list
property. Please be aware that reloading external certificates does only work if haproxy is started in daemon mode. So this feature does not work if “haproxy.syslog_server” is set to “stdout”. If using this feature but not using internal certs, you should set ha_proxy.crt_list to be an empty array
- Default
false
ext_crt_list_file
¶The location from which to load additional external certificates list
- Default
/var/vcap/jobs/haproxy/config/ssl/ext/crt-list
ext_crt_list_policy
¶What to do if the external certificates list located at
ha_proxy.ext_crt_list_file
does not appear within the time denoted byha_proxy.ext_crt_list_timeout
. Set to either ‘fail’ (HAproxy will not start) or ‘continue’ (HAproxy will start without external certificates)
- Default
fail
ext_crt_list_timeout
¶Timeout (in seconds) to wait for the external certificates list located at
ha_proxy.ext_crt_list_file
to appear during HAproxy startup
- Default
60
forwarded_client_cert
¶This option lets you decide how to handle the X-Forwarded-Client-Cert (XFCC) http header on any https frontend. On http frontends the
always_forward_only
option is active by default and can’t be changed. On https frontends your options are:
always_forward_only: Least secure option. Always forward the X-Forwarded-Client-Cert header in the request, regardless of whether the client connection is mTLS. The following headers will also be forwarded if they are present in the original request: X-SSL-Client, X-SSL-Client-Session-ID, X-SSL-Client-Verify, X-SSL-Client-Subject-DN, X-SSL-Client-Subject-CN, X-SSL-Client-Issuer-DN, X-SSL-Client-NotBefore, X-SSL-Client-NotAfter. Use this value when your load balancer is forwarding the client certificate and requests are not forwarded to HAProxy over mTLS. In the case where the connection between load balancer and HAProxy is mTLS, the client certificate received by HAProxy in the mTLS handshake will not be forwarded.
forward_only: Secure version of
always_forward_only
. Forward the X-Forwarded-Client-Cert header received from the client only when the client connection is mTLS. The following headers will also be forwarded for mTLS connections if they are present in the original request: X-SSL-Client, X-SSL-Client-Session-ID, X-SSL-Client-Verify, X-SSL-Client-Subject-DN, X-SSL-Client-Subject-CN, X-SSL-Client-Issuer-DN, X-SSL-Client-NotBefore, X-SSL-Client-NotAfter. If the client connection does not use mTLS, these X-Forwarded-Client-Cert and X-SSL-Client-* headers will be removed if they are present. In the case where the connection between load balancer and HAProxy is mTLS, the client certificate received by HAProxy in the mTLS handshake will not be forwarded.sanitize_set: Most secure option. Strip any instances of XFCC headers from the client request.
When the client connection is mTLS, the following headers will be overwritten in the request
X-Forwarded-Client-Cert: Contains the client certificate in binary DER format (Base64 encoded). Backends should use this header to authenticate incoming requests.
X-SSL-Client: Contains the number 1 if the request was made using a client certificate, 0 otherwise. For easy checks on the backend.
X-SSL-Client-Session-ID: The SSL session ID of the client connection. Useful for debugging purposes.
X-SSL-Client-Verify: Contains the number 0 if the client certificate verification was successful. Otherwise it contains the appropriate OpenSSL return code (see https://github.com/openssl/openssl/blob/master/include/openssl/x509_vfy.h.in#L206)
X-SSL-Client-Subject-DN: Contains the base64-encoded subject distinguished name of the client certificate
X-SSL-Client-Subject-CN: Contains the base64-encoded subject common name of the client certificate
X-SSL-Client-Issuer-DN: Contains the base64-encoded issuer distinguished name of the client certificate
X-SSL-Client-Root-CA-DN: X-SSL-Client-Root-CA-DN: Contains base64-encoded subject DN of the root CA which signed the client certificate
X-SSL-Client-NotBefore: Contains the start date of the client certificate in YYMMDDhhmmss[Z] format.
X-SSL-Client-NotAfter: Contains the expiration date of the client certificate in YYMMDDhhmmss[Z] format.
forward_only_if_route_service: This option is useful to support Mutual TLS with CF Route Services. When the client connection is not mTLS, the X-Forwarded-Client-Cert and X-SSL-Client-* headers will be removed UNLESS there is an X-Cf-Proxy-Signature header. When the client connection is mTLS, the following headers will be overwritten in the request
X-Forwarded-Client-Cert: Contains the client certificate in binary DER format (Base64 encoded). Backends should use this header to authenticate incoming requests.
X-SSL-Client: Contains the number 1 if the request was made using a client certificate, 0 otherwise. For easy checks on the backend.
X-SSL-Client-Session-ID: The SSL session ID of the client connection. Useful for debugging purposes.
X-SSL-Client-Verify: Contains the number 0 if the client certificate verification was successful. Otherwise it contains the appropriate OpenSSL return code (see https://github.com/openssl/openssl/blob/master/include/openssl/x509_vfy.h.in#L206)
X-SSL-Client-Subject-DN: Contains the base64-encoded subject distinguished name of the client certificate
X-SSL-Client-Subject-CN: Contains the base64-encoded subject common name of the client certificate
X-SSL-Client-Issuer-DN: Contains the base64-encoded issuer distinguished name of the client certificate
X-SSL-Client-NotBefore: Contains the start date of the client certificate in YYMMDDhhmmss[Z] format.
X-SSL-Client-NotAfter: Contains the expiration date of the client certificate in YYMMDDhhmmss[Z] format.
This option is only secure if Gorouter is deployed behind Haproxy to validate that X-Cf-Proxy-Signature is coming from a route service.
- Default
sanitize_set
frontend_config
¶Raw HAProxy config that will be added to each HA proxy frontend definition, provided either as a multiline text blob or as an array of lines.
global_config
¶Raw HAProxy config that will be added to the HA proxy global section, provided either as a multiline text blob or as an array of lines.
headers
¶Hash of custom headers you wish you have set on each request. Spaces are automatically escaped, but any other haproxy delimiters will need to be escaped manually
- Example
|+ headers: X-Application-ID: my-custom-header MyCustomHeader: 3
health_check_port
¶port for http health-check
- Default
8080
hsts_enable
¶Enables HSTS(Strict-Transport-Security Header) for all the SSL/TLS listeners
- Default
false
hsts_include_subdomains
¶This enables the includeSubDomains flag for HSTS.
- Default
false
hsts_max_age
¶max-age value for the Strict-Transport-Security header
- Default
3.1536e+07
hsts_preload
¶This enables the preload flag for HSTS
- Default
false
http_request_deny_conditions
¶List of conditions to block http requests. Each condition consists of multiple rules combined with the AND operator. Setting the negate flag to true will negate the acl condition.
- Example
http_request_deny_conditions: - condition: - acl_name: block_host acl_rule: hdr_beg(host) -i login - acl_name: block_reset_password_url acl_rule: path_beg,url_dec -m beg -i /reset_password - acl_name: whitelist_ips acl_rule: src 5.22.5.11 5.22.5.12 negate: true
https_redirect_all
¶If this is set to ‘true’, a https redirect rule for all http calls will be put in the config file
- Default
false
https_redirect_domains
¶For each domain in this array, a HTTPS redirect rule will be put in the config file. Redirect will be applied for all subdomains
- Default
[]
internal_only_domains
¶Array of domains for internal-only apps/services (not hostnames for the apps/services)
- Default
[]
keepalive_timeout
¶Timeout (in floating point seconds) applied to any connection that is in an http-keepalive state, waiting for the next request to occur
- Default
6
legacy_xfcc_header_mapping
¶
- Default
false
log_format
¶The log format used when generating syslog messages.
- Default
raw
log_level
¶Log level
- Default
info
log_max_length
¶Optional maximum line length. Log lines larger than this value will be truncated before being sent.
- Default
1024
lua_scripts
¶A list of LUA scripts that HA Proxy should load. These will need to be provided by other boshreleases, as none are supplied in
haproxy_boshrelease
. Specify each script with the full path to the script (likely/var/vcap/packages/something/something/darkside.lua
)Inside these scripts, any variable set is visible from any thread. This is the easiest and recommended way to load Lua programs but it will not scale well if a lot of Lua calls are performed, as only one thread may be running on the global state at a time. A program loaded this way will always see 0 in the “core.thread” variable. This directive can be used multiple times.
- Default
[]
lua_scripts_per_thread
¶A list of LUA scripts that HA Proxy should load per thread. These will need to be provided by other boshreleases, as none are supplied in
haproxy_boshrelease
. Specify each script with the full path to the script (likely/var/vcap/packages/something/something/darkside.lua
).Inside these scripts, any global variable has a thread-local visibility so that each thread could see a different value. As such it is strongly recommended not to use global variables in programs loaded this way. An independent copy is loaded and initialized for each thread, everything is done sequentially and in the thread’s numeric order from 1 to nbthread. If some operations need to be performed only once, the program should check the “core.thread” variable to figure what thread is being initialized. Programs loaded this way will run concurrently on all threads and will be highly scalable. This is the recommended way to load simple functions that register sample-fetches, converters, actions or services once it is certain the program doesn’t depend on global variables. For the sake of simplicity, the directive is available even if only one thread is used and even if threads are disabled (in which case it will be equivalent to
lua_scripts
).
- Default
[]
master_cli_bind
¶IP and port or UNIX socket to bind master CLI to
- Default
127.0.0.1:9001
master_cli_enable
¶If true, enables the master CLI which can be used to manage HAProxy
- Default
false
max_connections
¶Number of simultaneous connections HAProxy supports handling
- Default
64000
max_open_files
¶The number of file descriptors HAProxy can have open at one time
- Default
256000
max_rewrite
¶Buffer size to use for header rewriting or appending. The default of haproxy is min(1024,buffer_size_bytes/2). Will be set to buffer_size_bytes/2 by haproxy if it is set to a larger value
nbthread
¶Optional number of threads per VM
- Default
1
pre_start_script
¶This script will be appended to the pre-start script and run before the job starts. The pre-start script allows the job to prepare the machine before starting HAProxy, for example, by setting the MTU to a custom value.
- Example
|+ # customize MTU CUST_MTU=((custom_mtu)) INTERFACE=$(ip -4 route get 8.8.8.8 | grep -Po '(?<=dev )\S+') #' CURR_MTU=$(ip link show "$INTERFACE" | grep -Po 'mtu \K\d+') if [[ $CURR_MTU -ne $CUST_MTU ]]; then sudo ip link set dev "$INTERFACE" mtu "$CUST_MTU" echo "MTU changed from $CURR_MTU to $CUST_MTU, interface: $INTERFACE." fi CURR_MTU=$(ip link show "$INTERFACE" | grep -Po 'mtu \K\d+') echo "MTU: $CURR_MTU, interface: $INTERFACE"
queue_timeout
¶Timeout (in floating point seconds) used on any connection sitting in the pending queue, waiting to be sent to the backend, to limit its time being queued
- Default
30
raw_blocks
¶A hash of block types, where each type contains either a configuration or a hash of specific block names with their respective configurations. The configurations are provided as either multiline text blobs or arrays of lines. This structure will be appended to the end of the HAProxy configuration file. Use at your own risk.
- Example
defaults: |+ log global timeout http-request 10s listen: my-listen-x: |+ bind :81 mode http server-template srv 1-3 q-s0.web.default.deployment-x.bosh:8080 check inter 1000 my-listen-y: - bind :82 - mode http - server-template srv 1-3 q-s0.web.default.deployment-y.bosh:8080 check inter 1000
raw_config
¶A multiline text blob of an entire haproxy config. Overrides every other option available, so you can provide your own config, and do whatever you want. Use at your own risk.
reload_hard_stop_after
¶Defines the maximum time allowed to perform a clean soft-stop. This is used when issuing a reload via the “bin/reload” script. It limits the time for the old instance to finish existing connections before the instance is stopped.
- Default
5m
reload_idle_close_on_response
¶This option makes HAproxy wait for another request on idle connections during reloads or restarts. Once the response is received, a “Connection: close” header is injected to signal to the client that this connection no longer usable and permitting a more graceful handling on the client’s side. This has the downside that HAproxy may linger around for longer, waiting for a request on idle connections, so it should be used in conjunction with the “reload_hard_stop_after” option to limit that time. If set to false, idle connections will be terminated immediately during reloads or restarts.
- Default
true
reload_max_instances
¶Sets the maximum number of instances to exist at the same time. In conjunction with “reload_hard_stop_after” this limits the number of reloads that can occur during a given period of time. Set this to the number of instances your machine can fit into memory at a time, minus a safety buffer. Set to 0 for no limit.
- Default
4
request_timeout
¶Timeout (in floating point seconds) applied to any connection to limit the maximum time for a complete HTTP request (headers only). Used to limit DoS attacks that send data slowly to not trigger the client/server timeouts
- Default
5
requests_rate_limit
¶
block
¶Whether or not to block requests. See docs/rate_limiting.md
- Default
false
requests
¶How many requests are allowed in the given time window from one IP address. See docs/rate_limiting.md
table_size
¶Size of the stick table in which the IPs and counters are stored. See docs/rate_limiting.md
window_size
¶Window size for counting requests. See docs/rate_limiting.md
resolve_retries
¶Number of dns queries to send to resolve a server name before giving up
- Default
3
resolve_retry_timeout
¶Timeouts related to resolver name resolution
- Default
1s
resolvers
¶List of DNS servers
- Example
resolvers: - private: 10.0.0.2 - public: 8.8.8.8
retries
¶HAProxy will retry this many times on failed connections. When redispatch is enabled, the retries may occur on different servers. In combination with connect_timeout this defines the maximum response time of HAProxy to clients. e.g. 0.5s connect_timeout * 10 retries = 5s max response time
- Default
0
routed_backend_servers
¶Hash of the URL prefixes -> array of the router IPs acting as the HTTP/TCP backends (should include servers all Availability Zones being used)
- Default
{}- Example
routed_backend_servers: /images: additional_acls: - method GET backend_health_fall: 3 backend_health_rise: 2 backend_http_health_port: 80 backend_http_health_uri: /health backend_ssl: verify backend_use_http_health: true backend_verifyhost: example.com port: 4443 servers: - 10.0.0.2 - 10.0.0.3
rsp_headers
¶Hash of custom headers you wish you have set on each request. Spaces are automatically escaped, but any other haproxy delimiters will need to be escaped manually
- Example
|+ rsp_headers: X-Application-ID: my-custom-header MyCustomHeader: 3
server_timeout
¶Timeout (in floating point seconds) used on connections from haproxy to a backend, while waiting for data from the backend
- Default
30
ssl_ciphers
¶List of TLSv<=1.2 Ciphers for that are passed to HAProxy
- Default
ECDHE-ECDSA-CHACHA20-POLY1305:ECDHE-RSA-CHACHA20-POLY1305:ECDHE-ECDSA-AES128-GCM-SHA256:ECDHE-RSA-AES128-GCM-SHA256:ECDHE-ECDSA-AES256-GCM-SHA384:ECDHE-RSA-AES256-GCM-SHA384:DHE-RSA-AES128-GCM-SHA256:DHE-RSA-AES256-GCM-SHA384:ECDHE-ECDSA-AES128-SHA256:ECDHE-RSA-AES128-SHA256:ECDHE-ECDSA-AES128-SHA:ECDHE-RSA-AES256-SHA384:ECDHE-RSA-AES128-SHA:ECDHE-ECDSA-AES256-SHA384:ECDHE-ECDSA-AES256-SHA:ECDHE-RSA-AES256-SHA:DHE-RSA-AES128-SHA256:DHE-RSA-AES128-SHA:DHE-RSA-AES256-SHA256:DHE-RSA-AES256-SHA:ECDHE-ECDSA-DES-CBC3-SHA:ECDHE-RSA-DES-CBC3-SHA:EDH-RSA-DES-CBC3-SHA:AES128-GCM-SHA256:AES256-GCM-SHA384:AES128-SHA256:AES256-SHA256:AES128-SHA:AES256-SHA:DES-CBC3-SHA:!DSS
ssl_ciphersuites
¶List of TLSv1.3 Ciphers that are passed to HAProxy, requires at least openssl version 1.1.1
- Example
TLS_AES_128_GCM_SHA256:TLS_AES_256_GCM_SHA384:TLS_CHACHA20_POLY1305_SHA256
ssl_max_ver
¶This option enforces the use of ‘version’ or lower for SSL connections initiated from this listener. It will only be set if ‘ssl_min_ver’ is specified, as the default HAProxy ssl-min-ver may change in future.
- Example
TLSv1.3
ssl_min_ver
¶This option enforces the use of ‘version’ or higher for SSL connections initiated from this listener. The recommended value is ‘TLSv1.2’. It is not the default due to backward compatibility concerns with the ‘disabletls’ options. If this option is set, the ‘disable_tls_1’ options will be ignored.
- Example
TLSv1.2
ssl_pem
¶Array of private keys and certificates used for TLS handshakes with downstream clients. Each element in the array is an object containing fields ‘cert_chain’ and ‘private_key’, each of which supports a PEM block. Each element can also be a single string containing the cert chain and the private key.
- Example
ssl_pem: - cert_chain: |+ -----BEGIN CERTIFICATE----- -----END CERTIFICATE----- -----BEGIN CERTIFICATE----- -----END CERTIFICATE----- private_key: |+ -----BEGIN RSA PRIVATE KEY----- -----END RSA PRIVATE KEY----- - |+ -----BEGIN CERTIFICATE----- -----END CERTIFICATE----- -----BEGIN CERTIFICATE----- -----END CERTIFICATE----- -----BEGIN RSA PRIVATE KEY----- -----END RSA PRIVATE KEY-----
stats_bind
¶Define listening address and port for the stats frontend. If multithreading is enabled (
ha_proxy.threads > 1
) multiple stat pages are available - one for each thread. You can see the stat page for each thread on a separate port - starting at the defined port number.
- Default
'*:9000'
stats_enable
¶If true, haproxy will enable a socket for stats. You can see the stats on
haproxy_ip:9000/haproxy_stats
. If multithreading is enabled (ha_proxy.threads > 1
) haproxy will create a separate socket and stat page for each thread. Each stat page is reachable on a different port ranging from9000
to9000 + ha_proxy.threads - 1
.
- Default
false
stats_password
¶Password to authenticate haproxy stats, ignored when ‘stats_user’ is empty
stats_promex_enable
¶If true, haproxy will enable native prometheus exporter.
- Default
false
stats_promex_path
¶Define prometheus exporter path.
- Default
/metrics
stats_uri
¶URI used to access the stats UI.
- Default
haproxy_stats
stats_user
¶User name to authenticate haproxy stats, leave empty if you want the statistics to be available without authentication
strict_sni
¶Optional setting to decide whether the SSL/TLS negotiation is allowed only if the client provided an SNI which strict match a certificate. If set to true, the default certificate is not used
- Default
false
syslog_server
¶An IPv4 address optionally followed by a colon and a UDP port. It can also be an IPv6 address or filesystem path to a UNIX domain socket.
- Default
stdout
tcp
¶List of mappings to perform tcp-based proxying on. See example for mapping datastructure and keys
- Default
[]- Example
tcp: - backend_port: 80 backend_servers: - 10.20.10.10 - 10.20.10.11 backend_servers_local: - 10.20.10.10 backend_ssl: verify backend_verifyhost: example.com balance: roundrobin health_check_http: 4444 name: wss port: 4443 ssl: true
tcp_backend_config
¶Raw HAProxy config that will be added to the CF TCP Router + Generic TCP backend definitions, provided either as a multiline text blob or as an array of lines.
tcp_link_check_port
¶Optional port for tcp_backend health checks. Will use ha_proxy.tcp_link_port if not set.
tcp_link_health_check_http
¶Optional port for http health check when using the tcp_backend link.
tcp_link_port
¶Port haproxy should listen on when using the tcp_backend link
tcp_routing
¶
port_range
¶A range of ports for haproxy to listen on to enable CF TCP Routing. Used only if ‘tcp_router’ link is present.
- Default
1024-1123
trusted_domain_cidrs
¶Space separated trusted cidr blocks for internal_only_domains. You may alternatively provide a base64-encoded gzipped HAProxy cidr file, with each CIDR on a new line.
- Default
0.0.0.0/32
trusted_stats_cidrs
¶Trusted ip range that can access the stats UI
- Default
0.0.0.0/32
v4v6
¶Boolean, disabled by default. Enables binding to all IPv4 and IPv6 interfaces. Only applies if
ha_proxy.binding_ip
is set to::
.
- Default
false
websocket_timeout
¶Timeout (in floating point seconds) used on websocket/tunnel traffic, when both ends of the conversation have become inactive
- Default
3600
Templates¶
Templates are rendered and placed onto corresponding
instances during the deployment process. This job's templates
will be placed into /var/vcap/jobs/haproxy/
directory
(learn more).
bin/drain
(fromdrain.erb
)bin/haproxy_wrapper
(fromhaproxy_wrapper.erb
)bin/pre-start
(frompre-start.erb
)bin/reload
(fromreload.erb
)config/backend-ca-certs.pem
(frombackend-ca-certs.erb
)config/backend-crt.pem
(frombackend-crt.erb
)config/blacklist_cidrs.txt
(fromblacklist_cidrs.txt.erb
)config/bpm.yml
(frombpm.yml
)config/certs.ttar
(fromcerts.ttar.erb
)config/cidrs.ttar
(fromcidrs.ttar.erb
)config/client-ca-certs.pem
(fromclient-ca-certs.erb
)config/client-revocation-list.pem
(fromclient-revocation-list.erb
)config/haproxy.config
(fromhaproxy.config.erb
)config/ssl_redirect.map
(fromssl_redirect.map.erb
)config/trusted_domain_cidrs.txt
(fromtrusted_domain_cidrs.txt.erb
)config/whitelist_cidrs.txt
(fromwhitelist_cidrs.txt.erb
)
Packages¶
Packages are compiled and placed onto corresponding
instances during the deployment process. Packages will be
placed into /var/vcap/packages/
directory.