Key: name
Required: yes
Default: ``
|
String; may consist of Latin letters and digits and must start with a letter
|
Key: sitetype
Required: yes
Default: —
|
Site type.
Possible values:
|
Key: domains
Required: yes
Default: —
|
List of domains served on the site. Specified by names, not FQDNs.
The site also serves all domains from the child trees of the domains specified in the domains property.
One domain can be served on several sites.
If a domain is specified on site A and not specified on site B, while one of its parent domains is specified on site B, then this domain and the domains from its child tree are served only on site B.
Thus every domain of the full domain tree is served on at least one site: where it is specified explicitly, or where its parent domain is specified explicitly, and so on up to the master domain.
The master domain is automatically bound exclusively to the master site and does not require separate specification.
|
Key: descr
Required: no
Default: ``
|
Site description. Default: not set
|
Key: ctrl_domains
Required: no
Default: []
|
List of domains that can be managed through a connection to the current site.
The specification method is the same as for the domains field.
|
Key: num
Required: no
Default: —
|
Numeric site number consisting of three digits, for use in site element names
|
Key: prefix
Required: no
Default: —
|
Three-character alphabetic prefix for use in site element names
|
Key: check_quorum
Required: no
Default: false
|
Switch for quorum accounting on the site.
If the value is not set or is empty, the corresponding global parameter is used.
If the parameter is not set anywhere, the quorum accounting mechanism is not applied on the site.
Quorum is a group of active, interconnected, and mutually available site servers totaling more than half.
Quorum ensures data integrity when a site splits into several parts, because operation is maintained only in the part that has quorum, and there is at most one such part at any time.
The quorum determination algorithm runs in nodes where the master instance of any role made redundant in Active-Passive mode is started or running: at activation and every 10 seconds during operation.
If quorum unavailability is detected, 2 more attempts are made at 2-second intervals, after which the node restarts and enters the waiting-for-quorum mode from its server.
A special case is splitting a site strictly into two equal halves, each a candidate for quorum.
Under high load, servers in the candidate subgroup may spend more time checking connectivity due to additional check timeouts when polling other servers (10 seconds).
To apply the quorum concept, the resulting equal islands require an external arbiter (50%+1). The island that has communication with the arbiter obtains quorum.
|
Key: check_offline
Required: no
Default: false
|
Switch for accounting the offline state of servers (unavailability of arbitration servers).
If the value is not set or is empty, the corresponding global parameter is used.
If the parameter is not set anywhere, the offline state accounting mechanism is not applied on the site.
Offline state is a condition when a server has no access to arbitration servers (the odd_referee parameter) located in the same zone as users and the outside world (outside the datacenter).
The check can be used, for example, when a site is placed in two datacenters connected by a separate direct channel and having independent external channels. More about the mode.
The mode protects against the loss of the external channel by one of the datacenters by removing from service microservices that interact with external services and subscribers:
im, email, mware, mg, svc, ivrscript.
|
Key: odd_referee
Required: no
Default: —
|
Address or domain name of a third-party server acting as an odd arbitration server when determining quorum among an even number of servers on the site.
The arbitration server is checked by sending a ping request.
If the value is not set or is empty, the corresponding global parameter is used.
If the parameter is not set anywhere, then when checking quorum on the site, the arbitration server is considered available and quorum is considered present.
Several addresses or domain names can be specified, separated by commas (for example, "8.8.8.8, my.domain.ru"). Then availability is considered successful if at least one of the arbitration servers is available.
Arbitration server availability is checked:
-
When check_quorum is enabled. If the number of configured servers on the site is even and exactly half of them are available in the connectivity group. Requests are made from all servers of that group at intervals of 2, 10, and 30 seconds under different conditions.
-
When check_offline is enabled. If the site is deployed in two datacenters connected by a direct channel. Requests are made from all servers of the site where microservices interacting with external services and subscribers are deployed: im, email, mware, mg, script, ivrscript.
|
Key: leader_site
Required: no
Default: —
|
Name of another system site that is the leader for the group of domains assigned to the current site.
Specified when the current site must be a standby site, that is, provide functionality only when the specified site is unavailable.
During the passive state, some microservices are forcibly deactivated. Other microservices remain active, provide data synchronization, and monitor the availability of the active site.
The setting assumes that the current standby site serves a subset of the specified site’s domains.
|
Key: logging_default_level
Required: no
Default: empty
|
Default logging level applied in nodes served by servers of this site.
Applied to nodes for which the default logging level is not set by the configuration settings of a role or server.
-
CRASH
-
ERROR
-
WARNING
-
INFO
-
TRACE
-
DEBUG
|
Key: logging_store_mode
Required: no
Default: empty
|
Mode of transferring logs to the logstore storage for nodes served by servers of this site.
Applied to nodes for which the mode is not set by the configuration settings of a server.
-
text — files are transferred as is
-
gzip — files are transferred compressed (<name>.gz); the compressed copy is created on the node next to the source file and deleted together with it; the source logs on the nodes remain uncompressed
|
Key: takeover
Required: no
Default: global
|
Takeover mode switch used by default for all microservices of this site.
Failover - activation of a lower-priority service instance when the higher-priority service instance (leader of an active-passive group) is detected as unavailable.
Takeover - activation of the higher-priority service instance (leader of an active-passive group) during its recovery, accompanied by stopping the service on the lower-priority service instance.
Applied only to microservice instances for which no individual value is set in the role configuration parameter.
* If the mode is enabled, takeover is performed automatically.
* If the mode is disabled, takeover is not performed; the recovered group leader temporarily becomes a follower and waits for the service to be unloaded on the lower-priority instance.
* If the option is not set, the value from the general section is used. If it is not set anywhere, takeover is enabled for more accurate load distribution according to the configured priorities.
|
Key: sip_stat_timeout_callinfo
Required: no
Default: 1200
|
Call statistics retention time in seconds. Used for monitoring requests /api/monitor/v1/trace/calls.
By default, data is retained for 20 minutes.
The longer the time, the longer after the call the data can be provided by a monitoring request. At the same time, more RAM is required in SIP signaling nodes.
Possible values from 0 to 86400. 0 - no retention.
|
Key: sip_stat_timeout_closure
Required: no
Default: 12600
|
Retention time in seconds of statistics about links between Call-Ids of telephone calls in one session for closure calculation. Used when building call diagrams.
It also stores information about links between call Call-Ids and system domains to enable building diagrams in working domains for their own calls.
By default, data is retained for 3.5 hours.
The longer the time, the longer after the call the closures and SIP signaling diagrams of the session can be built. At the same time, more RAM is required in SIP signaling nodes. When the callid log is enabled (the log_closure role parameter), closures are also built from it: the time can be reduced, but not below sip_stat_timeout_callinfo (domain links are used in the current calls list of working domains), volume estimate.
Possible values from 0 to 86400. 0 - no retention.
|
Key: virtualip_automation
Required: no
Default: false
|
Switch for the service that assigns virtual IP addresses to cluster servers.
The value applies to servers for which the corresponding server parameter is not set directly.
If the parameter is set neither for the server nor for the site, the corresponding global parameter is used for the server.
The operating principle is described in the servers section.
|