Login | Register


Documentation -> Manuals -> Manual devel -> Core Parameters

Pages for other versions: devel 3.3 3.2 3.1 Older versions: 3.0 2.4 2.3 2.2 2.1 1.11 1.10 1.9 1.8 1.7 1.6 1.5 1.4

Core Parameters v3.4

Table of Contents (hide)

  1. 1. Core parameters
    1. 1.1 abort_on_assert
    2. 1.2 advertised_address
    3. 1.3 advertised_port
    4. 1.4 alias
    5. 1.5 auto_aliases
    6. 1.6 auto_scaling_cycle
    7. 1.7 auto_scaling_profile
    8. 1.8 check_via
    9. 1.9 chroot
    10. 1.10 debug_mode
    11. 1.11 db_version_table
    12. 1.12 db_default_url
    13. 1.13 db_max_async_connections
    14. 1.14 disable_503_translation
    15. 1.15 disable_core_dump
    16. 1.16 disable_dns_blacklist
    17. 1.17 disable_dns_failover
    18. 1.18 disable_stateless_fwd
    19. 1.19 dns
    20. 1.20 dns_retr_time
    21. 1.21 dns_retr_no
    22. 1.22 dns_servers_no
    23. 1.23 dns_try_ipv6
    24. 1.24 dns_try_naptr
    25. 1.25 dns_use_search_list
    26. 1.26 dst_blacklist
    27. 1.27 enable_asserts
    28. 1.28 event_pkg_threshold
    29. 1.29 event_shm_threshold
    30. 1.30 exec_dns_threshold
    31. 1.31 exec_msg_threshold
    32. 1.32 include_file
    33. 1.33 import_file
    34. 1.34 listen (Replaced in OpenSIPS 3.1)
    35. 1.35 log_facility
    36. 1.36 log_level
    37. 1.37 log_name
    38. 1.38 log_stdout
    39. 1.39 log_stderror
    40. 1.40 log_prefix
    41. 1.41 max_while_loops
    42. 1.42 maxbuffer
    43. 1.43 mem-group
    44. 1.44 mem_warming
    45. 1.45 mem_warming_percentage
    46. 1.46 mem_warming_pattern_file
    47. 1.47 memdump | mem_dump
    48. 1.48 memlog | mem_log
    49. 1.49 mcast_loopback
    50. 1.50 mcast_ttl
    51. 1.51 mhomed
    52. 1.52 mpath
    53. 1.53 open_files_limit
    54. 1.54 poll_method
    55. 1.55 port
    56. 1.56 pv_print_buf_size
    57. 1.57 query_buffer_size
    58. 1.58 query_flush_time
    59. 1.59 restart_persistency_cache_file
    60. 1.60 restart_persistency_size
    61. 1.61 rev_dns
    62. 1.62 server_header
    63. 1.63 server_signature
    64. 1.64 shm_hash_split_percentage
    65. 1.65 shm_secondary_hash_size
    66. 1.66 sip_warning
    67. 1.67 socket
    68. 1.68 tcp_workers
    69. 1.69 tcp_accept_aliases
    70. 1.70 tcp_connect_timeout
    71. 1.71 tcp_connection_lifetime
    72. 1.72 tcp_max_connections
    73. 1.73 tcp_max_msg_time
    74. 1.74 tcp_no_new_conn_bflag
    75. 1.75 tcp_parallel_read_on_workers
    76. 1.76 tcp_socket_backlog
    77. 1.77 tcp_threshold
    78. 1.78 tcp_keepalive
    79. 1.79 tcp_keepcount
    80. 1.80 tcp_keepidle
    81. 1.81 tcp_keepinterval
    82. 1.82 timer_workers
    83. 1.83 tos
    84. 1.84 udp_workers
    85. 1.85 user_agent_header
    86. 1.86 wdir
    87. 1.87 xlog_buf_size
    88. 1.88 xlog_force_color
    89. 1.89 xlog_level
    90. 1.90 xlog_print_level

This section lists the all the parameters exported by OpenSIPS core for script usage (to be used in opensips.cfg).

1. Core parameters

Global parameters that can be set in configuration file. Accepted values are, depending on the actual parameters strings, numbers and yes/ no. If you need to specify either "yes" or "no" as part of a string, wrap this in double quotes.

1.1 abort_on_assert 🔗

Default value: false

Only relevant if asserts are enabled. Set to true in order to make OpenSIPS shut down immediately in case a script assert fails.

Example of usage:

    abort_on_assert = true

1.2 advertised_address 🔗

It can be an IP address or string and represents the address advertised in Via header and other destination lumps (e.g RR header). If empty or not set (default value) the socket address from where the request will be sent is used.


  - don't set it unless you know what you are doing (e.g. nat traversal)
  - you can set anything here, no check is made (e.g. foo.bar will be
  accepted even if foo.bar doesn't exist)

Example of usage:


NOTE: Aside this global approach, you can also define an advertise IP and port in a per-interface manner (see the socket parameter). When advertise values are defined per interface, they will be used only for traffic leaving that interface only.

1.3 advertised_port 🔗

The port advertised in Via header and other destination lumps (e.g. RR). If empty or not set (default value) the port from where the message will be sent is used. Same warnings as for 'advertised_address'.

Example of usage:


NOTE: Aside this global approach, you can also define an advertise IP and port in a per-interface manner (see the socket parameter). When advertise values are defined per interface, they will be used only for traffic leaving that interface only.

1.4 alias 🔗

Parameter to set alias hostnames for the server. It can be set many times, each value being added in a list to match the hostname when 'myself' is checked.

If the ":port" part is omitted, all ports of the given "hostname" will be considered an alias (similar behavior to port 0).

It is necessary to include the port (the port value used in the "socket=" definitions) in the alias definition otherwise the loose_route() function will not work as expected for local forwards!

Example of usage:


1.5 auto_aliases 🔗

This parameter controls if aliases should be automatically discovered and added during fixing listening sockets. The auto discovered aliases are result of the DNS lookup (if the 'socket' definition has a name and not IP) or of a reverse DNS lookup on the socket IP.

Far backward compatibility reasons, the default value is "off"/0.

Example of usage:


1.6 auto_scaling_cycle 🔗

The number of seconds defining a auto-scaling cycle - the auto-scaling engine, at each cycle, is evaluating the internal load of the groups and decided if more processes needs to be created or if existing processes need to be terminated. Also see auto_scaling_profile for more details on how the auto-scaling works.
The default value is 1 second. Example of usage:

    auto_scaling_cycle=3  # do auto-scaling checks once every 3 seconds

1.7 auto_scaling_profile 🔗

Defines the behavior of the auto-scaling support, in terms of how many processes should be allowed and when to terminate or create new processes. These profiles may be used for the UDP processes (see udp_workers or socket options) , TCP processes (see tcp_workers option) or TIMER processes (see timer_workers option).
For more, see this external description of auto-scaling.

Example of usage:

    auto_scaling_profile = PROFILE_SIP
     scale up to 6 on 70% for 4 cycles within 5   
     scale down to 2 on 18% for 10 cycles

This profile will allow the group to fork up to 6 processes. A new process will be forked when the overall load of the group will be higher than 70% for more than 4 cycles during a 5 cycles monitoring window. A cycle is a time unit used for monitoring (like 2 seconds).
Also the profile will allow the group to scale down to a minimum of 2 processes. A process will be terminated when the overall load of the group will be lower than 20% during 10 cycles. The down scaling part of the profile is optional. If not defined, OpenSIPS will never down scale, but only up scale.

1.8 check_via 🔗

Check if the address in top most via of replies is local. Default value is 0 (check disabled).

Example of usage:


1.9 chroot 🔗

The value must be a valid path in the system. If set, OpenSIPS will chroot (change root directory) to its value.

Example of usage:


1.10 debug_mode 🔗

Enabling the debug_mode option is a fast way to debug your OpenSIPS. This option will automatically force:

  • staying in foreground (do not detach from console)
  • set logging level to 4 (debug)
  • set logging to standard error
  • enable core dumping
  • set UDP worker processes to 2
  • set TCP worker processes to 2

Default value is false/0 (disabled).

NOTE that enabling this option will override all the other individual parameters like foreground mode, log level, udp_workers, tcp_workers, etc.

1.11 db_version_table 🔗

The name of the table version to be used by the DB API to check the version of the used tables.
Default value is "version"

Example of usage:


1.12 db_default_url 🔗

The default DB URL to be used by modules if no per-module URL is given. Default is NULL (not defined)

Example of usage:


1.13 db_max_async_connections 🔗

Maximum number of TCP connections opened from a single OpenSIPS worker to each individual SQL backend. Default value is 10.

Individual backends are determined from DB URLs as follows:

        [ scheme, user, pass, host, port, database ]

Example of usage:


1.14 disable_503_translation 🔗

If 'yes', OpenSIPS will not translate the received 503 replies into 500 replies (RFC 3261 clearly states that a proxy should never relay a 503 response, but instead it must transform it into a 500).

Default value is 'no' (do translation).

1.15 disable_core_dump 🔗

Can be 'yes' or 'no'. By default core dump limits are set to unlimited or a high enough value. Set this config variable to 'yes' to disable core dump-ing (will set core limits to 0).

Default value is 'no'.

Example of usage:


1.16 disable_dns_blacklist 🔗

The DNS resolver, when configured with failover, can automatically store in a temporary blacklist the failed destinations. This will prevent (for a limited period of time) OpenSIPS to send requests to destination known as failed. So, the blacklist can be used as a memory for the DNS resolver.

The temporary blacklist created by DNS resolver is named "dns" and it is by default selected for usage (no need use the use_blacklist()) function. The rules from this list have a life time of 4 minutes - you can change it at compile time, from resolve.c .

Can be 'yes' or 'no'. By default the blacklist is disabled (Default value is 'yes').

Example of usage:


1.17 disable_dns_failover 🔗

Can be 'yes' or 'no'. By default DNS-based failover is enabled. Set this config variable to 'yes' to disable the DNS-based failover. This is a global option, affecting the core and the modules also.

Default value is 'no'.

Example of usage:


1.18 disable_stateless_fwd 🔗

Can be 'yes' or 'no'. This parameter controls the handling of stateless replies:

    yes - drop stateless replies if stateless fwd functions (like forward) are not used in script
    no - forward stateless replies

Default value is 'yes'.

1.19 dns 🔗

This parameter controls if the SIP server should attempt to lookup its own domain name in DNS. If this parameter is set to yes and the domain name is not in DNS a warning is printed on syslog and a "received=" field is added to the via header.

Default is no.

1.20 dns_retr_time 🔗

Time in seconds before retrying a dns request. Default value is system specific, depends also on the '/etc/resolv.conf' content (usually 5s).

Example of usage:


1.21 dns_retr_no 🔗

Number of dns retransmissions before giving up. Default value is system specific, depends also on the '/etc/resolv.conf' content (usually 4).

Example of usage:


1.22 dns_servers_no 🔗

How many dns servers from the ones defined in '/etc/resolv.conf' will be used. Default value is to use all of them.

Example of usage:


1.23 dns_try_ipv6 🔗

Can be 'yes' or 'no'. If it is set to 'yes' and a DNS lookup fails, it will retry it for ipv6 (AAAA record). Default value is 'no'.

Example of usage:


1.24 dns_try_naptr 🔗

Disables the NAPTR lookups when doing DNS based routing for SIP requests - if disabled, the DNS lookup will start with SRV lookups. Can be 'yes' or 'no'. By default it is enabled, value 'yes'.

Example of usage:


1.25 dns_use_search_list 🔗

Can be 'yes' or 'no'. If set to 'no', the search list in '/etc/resolv.conf' will be ignored (=> fewer lookups => gives up faster). Default value is 'yes'.

HINT: even if you don't have a search list defined, setting this option to 'no' will still be "faster", because an empty search list is in fact search "" (so even if the search list is empty/missing there will still be 2 dns queries, eg. foo+'.' and foo+""+'.')

Example of usage:


1.26 dst_blacklist 🔗

Definition of a IP/destination blacklist. These lists can be selected from script (at runtime) to filter the outgoing requests, based on IP, protocol, port, etc.

Its primary purposes is be to prevent sending requests to critical IPs (like GWs), due bad DNS entries or to avoid sending to destinations that are known to be unavailable (temporary or permanent).

The grammar to specify a list is as it follows:

  "dst_blacklist" = id [/bl_flags] [: bl_rules] 
  • id is a unique identifier of the blacklist
  • bl_flags contains a set of optional modifiers:
  bl_flags = bl_flag [, bl_flag]*
  bl_flag = "expire" | "default" | "readonly"
  • bl_rules contains one or more blacklists rules
  bl_rules = [!] ipnet | { bl_rule [, bl_rule]* }
  bl_rule = [!] ( [bl_proto, ] ipnet [, port [, bl_pattern]] )

The blacklist modifiers meanings are as follows:

  • "expire": the blacklist may contain entries that expire
  • "default": the blacklist is used by default when sending requests, without having to explicitly set it (using the use_blacklist function)
  • "readonly": the blacklist is statically defined in script and cannot change at runtime

When dst_flags are missing, the "readonly" flag is explicitly set.

A rule is defined of the following properties:

  • if "!" is at the beginning of the rule, it negates the entire rule
  • bl_proto : any supported protocol, or "any" for any protocol; if missing, default is "any"
  • ipnet: IP or IP/MASK that should match the rule
  • port : number or 0 for any
  • bl_pattern - is a filename like matching (see "man 3 fnmatch") applied on the outgoing request buffer (first_line+hdrs+body)

Example of usage:

   # filter out requests going to ips of my gws
   dst_blacklist = gw:{( tcp , , 5060 , "" ),( any , , 0 , "" )}
   # block requests going to "evil" networks
   dst_blacklist = net_filter:{ ( any , , 0 , "" )}
   # block message requests with nasty words
   dst_blacklist = msg_filter:{ ( any , , 0 , "MESSAGE*ugly_word" )}
   # block requests not going to a specific subnet
   dst_blacklist = net_filter2:{ !( any , 193.468.30.0/ , 0 , "" )}
   # define a dynamic list that is built at runtime and has expire entries
   dst_blacklist = net_dynamic/expire

1.27 enable_asserts 🔗

Default value: false

Set to true in order to enable the assert script statement.

Example of usage:

    enable_asserts = true

1.28 event_pkg_threshold 🔗

A number representing the percentage threshold above which the E_CORE_PKG_THRESHOLD event is raised, warning about low amount of free private memory. It accepts integer values between 0 and 100.

Default value is 0 ( event disabled ).

Example of usage:

    event_pkg_threshold = 90

1.29 event_shm_threshold 🔗

A number representing the percentage threshold above which the E_CORE_SHM_THRESHOLD event is raised, warning about low amount of free shared memory. It accepts integer values between 0 and 100.

Default value is 0 ( event disabled ).

Example of usage:

    event_shm_threshold = 90

1.30 exec_dns_threshold 🔗

A number representing the maximum number of microseconds a DNS query is expected to last. Anything above the set number will trigger a warning message to the logging facility.

Default value is 0 ( logging disabled ).

Example of usage:

    exec_dns_threshold = 60000

1.31 exec_msg_threshold 🔗

A number representing the maximum number of microseconds the processing of a SIP msg is expected to last. Anything above the set number will trigger a warning message to the logging facility. Aside from the message and the processing time, the most time consuming function calls from the script will also be logged.

Default value is 0 ( logging disabled ).

Example of usage:

    exec_msg_threshold = 60000

1.32 include_file 🔗

Can be called from outside route blocks to load additional routes/blocks or from inside them to simply perform more functions. The file path can be relative or absolute. If it is a relative path, first attempt to locate it is relative to the directory from which OpenSIPS is started. If that fails, second try is relative to directory of the file that includes it. Will throw an error if file is not found.

Example of usage:

    include_file "proxy_regs.cfg"

1.33 import_file 🔗

Same as include_file.

Example of usage:

    import_file "proxy_regs.cfg"

1.34 listen (Replaced in OpenSIPS 3.1) 🔗

This parameter was replaced by the [#socket|socket]] parameter, preserving exactly the same format and behavior.

1.35 log_facility 🔗

If OpenSIPS logs to syslog, you can control the facility for logging. Very useful when you want to divert all OpenSIPS logs to a different log file. See the man page syslog(3) for more details.

Default value is LOG_DAEMON.

Example of usage:


1.36 log_level 🔗

Set the logging level (how verbose OpenSIPS should be). Higher values make OpenSIPS print more messages.

Examples of usage:

    log_level=1 -- print only important messages (like errors or more critical situations) 
    - recommended for running proxy as daemon

    log_level=4 -- print a lot of debug messages - use it only when doing debugging sessions

Actual values are:

  • -3 - Alert level
  • -2 - Critical level
  • -1 - Error level
  • 1 - Warning level
  • 2 (default) - Notice level
  • 3 - Info level
  • 4 - Debug level

The log_level parameter is usually used in concordance with the log_stderror parameter.

The value of the log_level parameter can also be get and set dynamically using the log_level Core MI function or $log_level script variable.

1.37 log_name 🔗

Set the id to be printed in syslog. The value must be a string and has effect only when OpenSIPS runs in daemon mode (fork=yes), after daemonize. Default value is argv[0].

Example of usage:


1.38 log_stdout 🔗

Although all OpenSIPS logs are done via standard error, enabling this parameter may be still be useful when trying to extract logs from 3rd party libraries.

- "no" (default) - drop all standard output logs

- "yes" - let all standard output logs pass through

Example of usage:

    log_stdout = yes

1.39 log_stderror 🔗

With this parameter you can make OpenSIPS write log messages to standard error. Possible values are:

- "no" (default) - write the messages to syslog

- "yes" - write the messages to standard error

Example of usage:

    log_stderror = yes

1.40 log_prefix 🔗

A string prefix which will be prepended to all logs produced by OpenSIPS (from both C code and script xlog() statements). Default: ""

Example of usage:

    log_prefix = "opensips-backup"

1.41 max_while_loops 🔗

The parameters set the value of maximum loops that can be done within a "while". Comes as a protection to avoid infinite loops in config file execution. Default is 100.

Example of usage:


1.42 maxbuffer 🔗

The size in bytes not to be exceeded during the auto-probing procedure of discovering the maximum buffer size for receiving UDP messages. Default value is 262144.

Example of usage:


1.43 mem-group 🔗

Defines a group of modules (by name) to get separate memory statistics. OpenSIPS will provide per-group memory information - the number of allocated fragments, the amount of used memory and the amount of real used memory (with memory manager overhead). This is useful if you want to monitor the memory usage of a certain module (or group of modules).

In order for the feature to work you have to run "make generate-mem-stats" and complile with the variable SHM_EXTRA_STATS defined.

Usage example:

    mem-group = "interest": "core" "tm"
    mem-group = "runtime": "dialog" "usrloc" "tm"

For the above example the generated statistics will be named: shmem_group_interest:fragments, shmem_group_interest:memory_used, shmem_group_interest:real_used.

Multiple groups can be defined, but they must not have the same name.

If you want to generate the statistics for the default group (all the other modules not included in a group) you have to complile with the variable SHM_SHOW_DEFAULT_GROUP defined.

1.44 mem_warming 🔗

Default value: off

Only relevant when the HP_MALLOC compile flag is enabled. If set to "on", on each startup, OpenSIPS will attempt to restore the memory fragmentation pattern it had before the stop/restart. If no pattern_file from a previous run is found, memory warming is skipped, and the memory allocator simply starts with a big chunk of memory, like all other allocators.

Memory warming is useful when dealing with high volumes of traffic (thousands of cps on multi-core machines - the more cores, the more useful), because processes must mutually exclude themselves when chopping up the initial big memory chunk. By performing fragmentation on startup, OpenSIPS will also behave optimally in the first minute(s) after a restart. Fragmentation usually lasts a few seconds (e.g. ~5 seconds on an 8GB shm pool and 3.4Ghz CPU) - traffic will not be processed at all during this period.

Example of usage:

    mem_warming = on

1.45 mem_warming_percentage 🔗

Default value: 75

How much of OpenSIPS's memory should be fragmented with the pattern of the previous run, upon a restart. Used at startup, if mem_warming is enabled.

Example of usage:

    mem_warming_percentage = 50

1.46 mem_warming_pattern_file 🔗

Default value: "CFG_DIR/mem_warming_pattern"

Only relevant if mem_warming is enabled. It contains the memory fragmentation pattern of a previous OpenSIPS run. This file is overwritten during each OpenSIPS shutdown, and is used during startup in order to restore the service behavior as soon as possible.

Example of usage:

    mem_warming_pattern_file = "/var/tmp/my_memory_pattern"


1.47 memdump | mem_dump mem_dump|🔗

Log level to print memory status information (runtime and shutdown). It has to be less than the value of 'log_level' parameter if you want memory info to be logged. Default: memdump=L_DBG (4)

Example of usage:


NOTE that setting memlog (see below), will also set the memdump parameter - if you want different values for memlog and memdump, you need to first set memlog and then memdump.


1.48 memlog | mem_log mem_log|🔗

Log level to print memory debug info. It has to be less than the value of 'log_level' parameter if you want memory info to be logged. Default: memlog=L_DBG (4)

Example of usage:


NOTE: by setting memlog parameter, the memdump will automatically be set to the same value (see memdump docs).

1.49 mcast_loopback 🔗

It can be 'yes' or 'no'. If set to 'yes', multicast datagram are sent over loopback. Default value is 'no'.

Example of usage:


1.50 mcast_ttl 🔗

Set the value for multicast ttl. Default value is OS specific (usually 1).

Example of usage:


1.51 mhomed 🔗

Set the server to try to locate outbound interface on multihomed host. By default is not (0) - it is rather time consuming.

Example of usage:


1.52 mpath 🔗

Set the module search path. This can be used to simplify the loadmodule parameter

Example of usage:

    loadmodule "mysql.so"
    loadmodule "uri.so"
    loadmodule "uri_db.so"
    loadmodule "sl.so"
    loadmodule "tm.so"

1.53 open_files_limit 🔗

If set and bigger than the current open file limit, OpenSIPS will try to increase its open file limit to this number. Note: OpenSIPS must be started as root to be able to increase a limit past the hard limit (which, for open files, is 1024 on most systems).

Example of usage:


1.54 poll_method 🔗

The poll method to be used by the I/O internal reactor - by default the best one for the current OS is selected. The available types are: poll, epoll, sigio_rt, select, kqueue, /dev/poll.

Example of usage:


1.55 port 🔗

The port the SIP server listens to. The default value for it is 5060.

Example of usage:


1.56 pv_print_buf_size 🔗

The maximum size of an expanded formatted string containing variables and/or pseudo-variables. Default: 20,000 bytes.

Example of usage:

    pv_print_buf_size = 60000

1.57 query_buffer_size 🔗

If set to a value greater than 1, inserts to DB will not be flushed one by one. Rows to be inserted will be kept in memory until until they gather up to query_buffer_size rows, and only then they will be flushed to the database.

Example of usage:


1.58 query_flush_time 🔗

If query_buffer_size is set to a value greater than 1, a timer will trigger once every query_flush_time seconds, ensuring that no row will be kept for too long in memory.

Example of usage:


1.59 restart_persistency_cache_file 🔗

This parameter controls the name of the cache file that is used to store restart persistence memory.

Default value is ".restart_persistency.cache".

1.60 restart_persistency_size 🔗

This parameter controls the size of the cache file. If this parameter is not specified, it defaults to the size of the shared memory.

Default value is the value of the shared memory, 32MB.

1.61 rev_dns 🔗

This parameter controls if the SIP server should attempt to lookup its own IP address in DNS. If this parameter is set to yes and the IP address is not in DNS a warning is printed on syslog and a "received=" field is added to the via header.

Default is no.

1.62 server_header 🔗

The body of Server header field generated by OpenSIPS when it sends a request as UAS. It defaults to "OpenSIPS (<version> (<arch>/<os>))".

Example of usage:

server_header="Server: My Company SIP Proxy"

Please note that you have to add the header name "Server:", otherwise OpenSIPS will just write a header like:

My Company SIP Proxy

1.63 server_signature 🔗

This parameter controls the "Server" header in any locally generated message.

Example of usage:


If it is enabled (default=yes) a header is generated as in the following example:

     Server: OpenSIPS (0.9.5 (i386/linux))

1.64 shm_hash_split_percentage 🔗

Only relevant when the HP_MALLOC compile flag is enabled. It controls how many memory buckets will be optimized. (e.g. setting it to 2% will optimize the first 81 most used buckets as frequency). The default value is 1.

1.65 shm_secondary_hash_size 🔗

Only relevant when the HP_MALLOC compile flag is enabled. It represents the optimization factor of a single bucket (e.g. setting it to 4 will cause the optimized buckets to be further split into 4). The default value is 8.

1.66 sip_warning 🔗

Can be 0 or 1. If set to 1 (default value is 0) a 'Warning' header is added to each reply generated by OpenSIPS. The header contains several details that help troubleshooting using the network traffic dumps.

Example of usage:


1.67 socket 🔗

Set the network addresses/sockets the OpenSIPS server should listen on. Its syntax is protocol:address[:port], where:

  • protocol: should be one of the transport modules loaded in the config file (e.g., udp, tcp, tls, bin, hep)
  • address: can be an IP address, a hostname, a network interface id, or the * wildcard which makes OpenSIPS listen on all possible interfaces for that protocol
  • port: optional, the port used by the listening socket - if absent, the default port exported by the transport module is used.

This parameter can be set multiple times in same configuration file, the server listening on all specified sockets.

The socket definition may accept several optional parameters:

  • "AS ip:port" - to configure an advertised IP and port only for an interface. Example "AS"
  • "USE_WORKERS n" - to set a different number of workers for this socket only (for UDP, SCTP and HEP_UDP interfaces only). This will override the global "udp_worker" parameter. Example "use_workers 5"
  • "ANYCAST" - to marke the socket as an anycast IP
  • "USE_AUTO_SCALING_PROFILE" - to enforce a certain governing policy on how many UDP workers you have at runtime. Dynamically, UDP processes my be created or terminated, depending on the load/traffic. This parameter may be used ONLY for UDP sockets. Note that the per-socket defined auto-scaling profile will override this global UDP auto-scaling profile.
  • "TAG" - this is a non SIP name/tag of the socket to be used when replicating the socket identify across an OpenSIPS cluster, with other OpenSIPS nodes. By using same TAG value, you can correlate/link listening sockets with different IPs on different OpenSIPS nodes. This is useful when replicating dialogs between OpenSIPS instances with different IPs.
  • "FRAG" - indicates that the socket should not use PMTU (Path MTU) discovery to determine whether fragmentation should be done, but always allow fragmentation (i.e. do not force DF bit to 1 in UDP packets).

Remember that the above parameters only affect the sockets they are configured for; if they are not defined for a given socket, the global values will be used instead.

Examples of usage:

    socket = udp:*
    socket = udp:eth1
    socket = tcp:eth1:5062
    socket = tls:localhost:5061
    socket = hep_udp:
    socket = ws: use_workers 5
    socket = sctp: as use_workers 3
    socket = udp: anycast
    socket = udp: use_workers 4 use_auto_scaling_profile PROFILE_SIP

On startup, OpenSIPS reports all the sockets that it is listening on.

1.68 tcp_workers 🔗

Number of worker processes to be created for reading from TCP connections. These workers are responsible for handling any traffic over any TCP based protocol, like SIP-TCP, SIP-TLS, SIP-WS, SIP-WSS, BIN or HEP. If no value is explicitly set, 8 TCP workers will be created. Optionally, you can define a auto-scaling profile to govern in a dynamic way the number of TCP workers (by creating or terminating processes, depending on load). See auto_scaling_profile parameter for more.

Example of usage:

    tcp_workers= 4
    tcp_workers= 3 use_auto_scaling_profile PROFILE_SIP

1.69 tcp_accept_aliases 🔗

Default value 0 (disabled). If enabled, OpenSIPS will enforce RFC 5923 behaviour when detecting an ";alias" Via header field parameter and will reuse any TCP (or TLS, WS, WSS) connection opened for such SIP requests (source IP + Via port + proto) when sending other SIP requests backwards, towards the same (source IP + Via port + proto) pair. The final purpose of RFC 5923, after all, is to minimize the number of TLS connections a SIP proxy must open, due to the large CPU overhead of the connection setup phase.

On top of RFC 5923's connection reusage (aliasing) mechanism, TCP connections in OpenSIPS are also persistent across multiple SIP dialogs. This can be controlled with the tcp_connection_lifetime global parameter.

WARNING! Enabling the global tcp_accept_aliases parameter (RFC 5923) for end-user initiated connections (who are most likely grouped by one or more public IPs) is an open vector for call hijacking! In such platforms, we recommend using the force_tcp_alias() core function, in order to employ RFC 5923 behaviour only in conjunction with adjacent SIP proxies.

1.70 tcp_connect_timeout 🔗

Time in milliseconds before an ongoing blocking attempt to connect will be aborted. Default value is 100ms.

Example of usage:

    tcp_connect_timeout = 5

1.71 tcp_connection_lifetime 🔗

Lifetime in seconds for TCP sessions. TCP sessions which are inactive for >tcp_connection_lifetime will be closed by OpenSIPS. Default value is defined in tcp_conn.h: #define DEFAULT_TCP_CONNECTION_LIFETIME 120. Setting this value to 0 will close the TCP connection pretty quick ;-). You can also set the TCP lifetime to the expire value of the REGISTER by using the tcp_persistent_flag parameter of the registrar module.

Example of usage:

    tcp_connection_lifetime = 3600

1.72 tcp_max_connections 🔗

Maximum number of active TCP accepted connections (i.e. initiated by remote endpoints). Once the limit is reached, any new incoming TCP connections will be rejected. The default is 2048. For outgoing TCP connections (initiated by OpenSIPS), there is currently no limit.

Example of usage:

    tcp_max_connections = 4096

1.73 tcp_max_msg_time 🔗

The maximum number of seconds that a SIP message is expected to arrive via TCP. If a single SIP packet is still not fully received after this number of seconds, the connection is dropped ( either the connection is very overloaded and this leads to high fragmentation - or we are the victim of an ongoing attack where the attacker is sending the traffic very fragmented in order to decrease our performance ). Default value is 4

Example of usage:

    tcp_max_msg_time = 8

1.74 tcp_no_new_conn_bflag 🔗

A branch flag to be used as marker to instruct OpenSIPS not to attempt to open a new TCP connection when delivering a request, but only to reuse an existing one (if available). If no existing conn, a generic send error will be returned.

This is intended to be used in NAT scenarios, where makes no sense to open a TCP connection towards a destination behind a NAT (like TCP connection created during registration was lost, so there is no way to contact the device until it re-REGISTER). Also this can be used to detect when a NATed registered user lost his TCP connection, so that opensips can disable his registration as useless.

Example of usage:

     tcp_no_new_conn_bflag = TCP_NO_CONNECT
     route {
         if (isflagset(DST_NATED) && $socket_in(proto) == "TCP")
         t_relay("0x02"); # no auto error reply
         $var(retcode) = $rc;
         if ($var(retcode) == -6) {
             #send error
             xlog("unable to send request to destination");
             send_reply("404", "Not Found");
         } else if ($var(retcode) < 0) {

1.75 tcp_parallel_read_on_workers 🔗

This option will allow a TCP conn to perform read operations from different processes, not only from one. So far, upon creation, a TCP conn was assigned to a TCP workers which was doing all the reading for that TCP conn. This may become a bootleneck. With "tcp_parallel_read_on_workers", after a read is completed, the TCP conn is passed back to the TCP Main processes, which will perform a re-balancing for the next read operations, passing the TCP conn potentially to another worker. NOTE: at TCP conn level, the read ops are still performed in serial way, one at a time (even if from different processes)

1.76 tcp_socket_backlog 🔗

The backlog argument defines the maximum length to which the queue of pending connections for the TCP listening sockets may grow. If a connection request arrives when the queue is full, the client may receive an error with an indication of ECONNREFUSED or, if the underlying protocol supports retransmission, the request may be ignored so that a later reattempt at connection succeeds.

Default configured value is 10.

1.77 tcp_threshold 🔗

A number representing the maximum number of microseconds sending of a TCP request is expected to last. Anything above the set number will trigger a warning message to the logging facility.

Default value is 0 ( logging disabled ).

Example of usage:

    tcp_threshold = 60000

1.78 tcp_keepalive 🔗

Enable or disable TCP keepalive (OS level).

Enabled by default.

Example of usage:

    tcp_keepalive = 1

1.79 tcp_keepcount 🔗

Number of keepalives to send before closing the connection (Linux only). Default value is Operating System dependent and can be found using cat /proc/sys/net/ipv4/tcp_keepalive_probes. Common value is 9.

Setting tcp_keepcount to any value will enable tcp_keepalive.

Example of usage:

    tcp_keepcount = 5

1.80 tcp_keepidle 🔗

Amount of time before OpenSIPS will start to send keepalives if the connection is idle (Linux only). Default value is Operating System dependent and can be found using cat /proc/sys/net/ipv4/tcp_keepalive_time. Common value is 7200 seconds.

Setting tcp_keepidle to any value will enable tcp_keepalive.

Example of usage:

    tcp_keepidle = 30

1.81 tcp_keepinterval 🔗

Interval between keepalive probes, if the previous one failed (Linux only). Default value is Operating System dependent and can be found using cat /proc/sys/net/ipv4/tcp_keepalive_intvl. Common value is 75 seconds.

Setting tcp_keepinterval to any value will enable tcp_keepalive.

Example of usage:

    tcp_keepinterval = 10

1.82 timer_workers 🔗

The number of worker processes to be created exclusively for timer related tasks/processing. The default and minimum number is '1'. Optionally, you can define a auto-scaling profile to govern in a dynamic way the number of timer workers (by creating or terminating processes, depending on load). See auto_scaling_profile parameter for more.

Example of usage:

    timer_workers = 3
    timer_workers = 3 use_auto_scaling_profile PROFILE_TIMER

1.83 tos 🔗

The TOS (Type Of Service) to be used for the sent IP packages (both TCP and UDP).

Example of usage:


1.84 udp_workers 🔗

Number of worker processes to be created for each UDP or SCTP interface you have defined. Default value is 8. Optionally, you can define a auto-scaling profile to govern in a dynamic way the number of UDP workers (by creating or terminating processes, depending on load). Note that the per-interface defined auto-scaling profile will override this global UDP auto-scaling profile. See auto_scaling_profile parameter for more.

Example of usage:

    udp_workers=4 use_auto_scaling_profile PROFILE_SIP 

NOTE: this global value (applicable for all UDP/SCTP interfaces) can be override if you set a different number of workers in the definition of a specific interface - so actually you can define a different number of workers for each interface (see the listen parameter for syntax).

1.85 user_agent_header 🔗

The body of User-Agent header field generated by OpenSIPS when it sends a request as UAC. It defaults to "OpenSIPS (<version> (<arch>/<os>))".

Example of usage:

user_agent_header="User-Agent: My Company SIP Proxy"

Please note that you have to include the header name "User-Agent:" as OpenSIPS does not add it and you will get an erroneous header like:

My Company SIP Proxy

1.86 wdir 🔗

The working directory used by OpenSIPS at runtime. You might find it usefull when come to generating core files :)

Example of usage:


1.87 xlog_buf_size 🔗

Default value: 4096

Size of the buffer used to print a single line on the chosen logging facility of OpenSIPS. If the buffer is too small, an overflow error will be printed, and the concerned line will be skipped.

Usage example:

    xlog_buf_size = 8388608 #given in bytes

1.88 xlog_force_color 🔗

Default value: false

Only relevant when xlog is set to true. Enables the use of the color escape sequences, otherwise they will have no effect.

Usage example:

    xlog_force_color = true

1.89 xlog_level 🔗

Similar to log_level this parameter independently controls (from the rest of the OpenSIPS code) the verbosity of the xlog() functions. This give you the possibility to separately control the verbosity level for logs from code versus logs from xlog().

Default value is 2 / L_NOTICE

Usage example:

    xlog_level = 3 #L_DBG

1.90 xlog_print_level 🔗

Default value: -1 (L_ERR)

Default level for printing the logs generated by xlog core function, when the log_level parameter is omitted.

Usage example:

    xlog_print_level = 2 #L_NOTICE

↑ Contents

Page last modified on October 28, 2022, at 03:11 PM