This section details all built-in Lua functions provided by the router.
The router provides Lua logging functionality that is convenient when creating custom Lua functions. A prefix can be added to the log message which is useful to differentiate log messages from different lua files. At the top of the Lua source file, add the line
local log = log.add_prefix("my_lua_file")
to prepend all log messages with "my_lua_file"
.
The logging functions support formatting and common log levels:
log.critical('A log message with number %d', 1.5)
log.error('A log message with string %s', 'a string')
log.warning('A log message with integer %i', 1)
log.info('A log message with a local number variable %d', some_local_number)
log.debug('A log message with a local string variable %s', some_local_string)
log.trace('A log message with a local integer variable %i', some_local_integer)
log.message('A log message')
Many of the router’s built-in Lua functions use the logging functions.
The router provides Lua logging functionality that is convenient when creating custom Lua functions. A prepend can be added to the log message which is useful to differentiate log messages from different lua files. At the top of the file, add the line
local log = log.add_prefix("my_lua_file")
to prepend all log messages with "my_lua_file"
.
The logging functions support formatting and common log levels:
log.critical('A log message with number %d', 1.5)
log.error('A log message with string %s', 'a string')
log.warning('A log message with integer %i', 1)
log.info('A log message with a local number variable %d', some_local_number)
log.debug('A log message with a local string variable %s', some_local_string)
log.trace('A log message with a local integer variable %i', some_local_integer)
log.message('A log message')
Many of the router’s built in Lua functions use the logging functions.
Predictive load balancing is a tool that can be used to avoid overloading hosts with traffic. Consider the case where a popular event starts at a certain time, let’s say 12 PM. A spike in traffic will be routed to the hosts that are streaming the content at 12 PM, most of them starting at low bitrates. A host might have sufficient bandwidth left to take on more clients but when the recently connected clients start ramping up in video quality and increase their bitrate, the host can quickly become overloaded, possibly dropping incoming requests or going offline. Predictive load balancing solves this issue by considering how many times a host recently been redirected to.
Four functions for predictive load balancing are provided by the router
that can be used when constructing conditions/weight functions: host_bitrate()
, host_bitrate_custom()
, host_has_bw()
and host_has_bw_custom()
.
All require data to be supplied to the selection input API and apply
only to leaf nodes in the routing tree. In order for predictive load balancing
to work properly the data must be updated at regular intervals. The data needs
to be supplied by the target system.
These functions are suitable to used as host health checks. To configure host health checks, see configuring CDNs and hosts.
Note that host_bitrate()
and host_has_bw()
rely on data supplied by metrics
agents, detailed in Cache hardware metrics: monitoring and routing.
host_bitrate_custom()
and host_has_bw_custom()
rely on
manually supplied selection input data, detailed in selection input API. The
bitrate unit depends on the data submitted to the selection input API.
The data supplied to the selection input API by the metrics agents uses the following structure:
{
"streamer-1": {
"hardware_metrics": {
"/": {
"free": 1741596278784,
"total": 1758357934080,
"used": 16761655296,
"used_percent": 0.9532561585516977
},
"cpu_load1": 0.02,
"cpu_load15": 0.12,
"cpu_load5": 0.02,
"mem_available": 4895789056,
"mem_available_percent": 59.551760354263074,
"mem_total": 8221065216,
"mem_used": 2474393600,
"n_cpus": 4
},
"per_interface_metrics": {
"eths1": {
"link": 1,
"interface_up": true,
"megabits_sent": 22322295739.378456,
"megabits_sent_rate": 8085.2523952,
"speed": 100000
}
}
}
}
Note that all built-in functions interacting with selection input values support indexing into nested selection input data. Consider the selection input data in above. The nested values can be accessed by using dots between the keys:
si('streamer-1.per_interface_metrics.eths1.megabits_sent_rate')
Note that the whole selection input variable name must be within single quotes.
The function si()
is documented under
general purpose functions.
host_bitrate({})
host_bitrate()
returns the predicted bitrate (in megabits per second) of
the host after the recently connected clients start ramping up in streaming
quality. The function accepts an argument table with the following keys:
interface
: The name of the interface to use for bitrate prediction.avg_bitrate
: the average bitrate per client,
defaults to 6 megabits per second.num_routers
: the number of routers that can route to this host,
defaults to 1. This is important to accurately predict the incoming load if
multiple routers are used.host
: The name of the host to use for bitrate prediction.
Defaults to the current host if not provided.This function relies on the field megabits_sent_rate
, supplied by the Telegraf
metrics agent, as seen in example metrics. If these fields
are missing from your selection input data, this function will not work.
Examples of usage:
host_bitrate({interface='eths0'})
host_bitrate({avg_bitrate=1, interface='eths0'})
host_bitrate({num_routers=2, interface='eths0'})
host_bitrate({avg_bitrate=1, num_routers=4, interface='eths0'})
host_bitrate({avg_bitrate=1, num_routers=4, host='custom_host', interface='eths0'})
host_bitrate({})
calculates the predicted bitrate as:
predicted_host_bitrate = current_host_bitrate + (recent_connections * avg_bitrate * num_routers)
host_bitrate_custom({})
Same functionality as host_bitrate()
but uses a custom selection input
variable as bitrate input instead of accessing hardware metrics. The function
accepts an argument table with the following keys:
custom_bitrate_var
: The name of the selection input variable to be used for
accessing current host bitrate.avg_bitrate
: see host_bitrate()
documentation above.num_routers
: see host_bitrate()
documentation above.host_bitrate_custom({custom_bitrate_var='host1_current_bitrate'})
host_bitrate_custom({avg_bitrate=1, custom_bitrate_var='host1_current_bitrate'})
host_bitrate_custom({num_routers=4, custom_bitrate_var='host1_current_bitrate'})
host_has_bw({})
Instead of accessing the predicted bitrate of a host through host_bitrate()
,
host_has_bw()
returns 1 if the host is predicted to have enough
bandwidth left to take on more clients after recent connections ramp up in
bitrate, otherwise it returns 0. The function accepts an argument table with the
following keys:
interface
: see host_bitrate()
documentation above.avg_bitrate
: see host_bitrate()
documentation above.num_routers
: see host_bitrate()
documentation above.host
: see host_bitrate()
documentation above.margin
: the bitrate (megabits per second) headroom that
should be taken into account during calculation, defaults to 0.host_has_bw({})
returns whether or not the following statement is true:
predicted_host_bitrate + margin < host_bitrate_capacity
host_has_bw({})
relies on the fields megabits_sent_rate
and speed
,
supplied by the Telegraf metrics agent, as seen in
example metrics. If these fields are missing from your
selection input data, this function will not work.
Examples of usage:
host_has_bw({interface='eths0'})
host_has_bw({margin=10, interface='eth0'})
host_has_bw({avg_bitrate=1, interface='eth0'})
host_has_bw({num_routers=4, interface='eth0'})
host_has_bw({host='custom_host', interface='eth0'})
host_has_bw_custom({})
Same functionality as host_has_bw()
but uses a custom selection input
variable as bitrate. It also uses a number or a custom selection input
variable for the capacity. The function accepts an argument table
with the following keys:
custom_capacity_var
: a number representing the capacity of the network
interface OR the name of the selection input variable to be used for
accessing host capacity.custom_bitrate_var
: see host_bitrate_custom()
documentationmargin
: see host_has_bw()
documentation above.
above.avg_bitrate
: see host_bitrate()
documentation above.num_routers
: see host_bitrate()
documentation above.Examples of usage:
host_has_bw_custom({custom_capacity_var=10000, custom_bitrate_var='streamer-1.per_interface_metrics.eths1.megabits_sent_rate'})
host_has_bw_custom({custom_capacity_var='host1_capacity', custom_bitrate_var='streamer-1.per_interface_metrics.eths1.megabits_sent_rate'})
host_has_bw_custom({margin=10, custom_capacity_var=10000, custom_bitrate_var='streamer-1.per_interface_metrics.eths1.megabits_sent_rate'})
host_has_bw_custom({avg_bitrate=1, custom_capacity_var=10000, custom_bitrate_var='streamer-1.per_interface_metrics.eths1.megabits_sent_rate'})
host_has_bw_custom({num_routers=4, custom_capacity_var=10000, custom_bitrate_var='streamer-1.per_interface_metrics.eths1.megabits_sent_rate'})
This section details built-in Lua functions that are meant to be used for host health checks. Note that these functions rely on data supplied by metric agents detailed in Cache hardware metrics: monitoring and routing. Make sure cache hardware metrics are supplied to the router before using any of these functions.
cpu_load_ok({})
The function accepts an optional argument table with the following keys:
host
: The name of the host. Defaults to the name of the
selected host if not provided.cpu_load5_limit
: The acceptable limit for the 5-minute CPU
load. Defaults to 0.9 if not provided.The function returns 1 if the five minute CPU load average is below their respective limits, and 0 otherwise.
Examples of usage:
cpu_load_ok()
cpu_load_ok({host = 'custom_host'})
cpu_load_ok({cpu_load5_limit = 0.8})
cpu_load_ok({host = 'custom_host', cpu_load5_limit = 0.8})
memory_usage_ok({})
The function accepts an optional argument table with the following keys:
host
: The name of the host. Defaults to the host of the
selected host if not provided.memory_usage_limit
: The acceptable limit for the memory usage.
Defaults to 0.9 if not provided.The function returns 1 if the memory usage is below the limit, and 0 otherwise.
Examples of usage:
memory_usage_ok()
memory_usage_ok({host = 'custom_host'})
memory_usage_ok({memory_usage_limit = 0.7})
memory_usage_ok({host = 'custom_host', memory_usage_limit = 0.7})
interfaces_online({})
The function accepts an argument table with the following keys:
interfaces
: A string or a table of strings representing the
network interfaces to check.host
: The name of the host. Defaults to the host of the
selected host if not provided.The function returns 1 if all the specified interfaces are online, and 0 otherwise.
This function relies on the fields link
and interface_up
, supplied by
the Telegraf metrics agent, as seen in example metrics. If
these fields are missing from your selection input data, this function will not
work.
Examples of usage:
interfaces_online({interfaces = 'eth0'})
interfaces_online({interfaces = {'eth0', 'eth1'}})
interfaces_online({host = 'custom_host', interfaces = 'eth0'})
interfaces_online({host = 'custom_host', interfaces = {'eth0', 'eth1'}})
health_check({})
The function accepts an optional argument table with the following keys:
interfaces
: A string or a table of strings representing the
network interfaces to check.host
: The name of the host. Defaults to the host of the
selected host if not provided.cpu_load5_limit
: The acceptable limit for the 5-minute CPU
load. Defaults to 0.9 if not provided.memory_usage_limit
: The acceptable limit for the memory usage.
Defaults to 0.9 if not provided.The function calls the health check functions cpu_load_ok({})
,
memory_usage_ok({})
and interfaces_online({})
. The functions returns 1 if
all these functions returned 1, otherwise it returns 0.
Examples of usage:
health_check({interfaces = 'eths0'})
health_check({host = 'custom_host', interfaces = 'eths0'})
health_check({cpu_load5_limit = 0.7, memory_usage_limit = 0.8, interfaces = 'eth0'})
health_check({host = 'custom_host', cpu_load5_limit = 0.7, memory_usage_limit = 0.8, interfaces = {'eth0', 'eth1'}})
The router supplies a number of general purpose Lua functions.
always()
Always returns 1.
never()
Always returns 0. Useful for temporarily disabling caches by using it as a health check.
Examples of usage:
always()
never()
si(si_name)
The function reads the value of the selection input variable si_name
and
returns it if it exists, otherwise it returns 0. The function accepts a string
argument for the selection input variable name.
Examples of usage:
si('some_selection_input_variable_name')
si('streamer-1.per_interface_metrics.eths1.megabits_sent_rate')
All comparison functions use the form function(si_name, value)
and compares
the selection input value with the name si_name
with value
.
ge(si_name, value)
- greater than or equalgt(si_name, value)
- greater thanle(si_name, value)
- less than or equallt(si_name, value)
- less thaneq(si_name, value)
- equal toneq(si_name, value)
- not equal toExamples of usage:
ge('streamer-1.hardware_metrics.mem_available_percent', 30)
gt('streamer-1.hardware_metrics./.free', 1000000000)
le('streamer-1.hardware_metrics.cpu_load5', 0.8)
lt('streamer-1.per_interface_metrics.eths1.megabits_sent_rate', 9000)
eq('streamer-1.per_interface_metrics.eths1.link.', 1)
neq('streamer-1.hardware_metrics.n_cpus', 4)
in_subnet(subnet)
Returns 1 if the current session belongs to subnet
, otherwise it returns 0.
See Subnets API for more details on how to use
subnets in routing. The function accepts a string argument for the subnet name.
Examples of usage:
in_subnet('stockholm')
in_subnet('unserviced_region')
in_subnet('some_other_subnet')
These functions checks the current session’s session groups.
in_session_group(session_group)
Returns 1 if the current session has been classified into session_group
,
otherwise it returns 0. The function accepts a string argument for the session
group name.
in_any_session_group({})
Returns 1 if the current session has been classified into any of
session_groups
, otherwise it returns 0. The function accepts a table array of
strings as argument for the session group names.
in_all_session_groups({})
Returns 1 if the current session has been classified into all of
session_groups
, otherwise it returns 0. The function accepts a table array of
strings as argument for the session group names.
Examples of usage:
in_session_group('safari_browser')
in_any_session_group({ 'in_europe', 'in_asia'})
in_all_session_group({ 'vod_content', 'in_america'})
Many of the functions documented are suitable to use in host health checks. To configure host health checks, see configuring CDNs and hosts. Here are some configuration examples of using the built-in Lua functions, utilizing the example metrics:
"healthChecks": [
"gt('streamer-1.hardware_metrics.mem_available_percent', 20)", // More than 20% memory is left
"lt('streamer-1.per_interface_metrics.eths1.megabits_sent_rate', 9000)" // Current bitrate is lower than 9000 Mbps
"host_has_bw({host='streamer-1', interface='eths1', margin=1000})", // host_has_bw() uses 'streamer-1.per_interface_metrics.eths1.speed' to determine if there is enough bandwidth left with a 1000 Mbps margin
"interfaces_online({host='streamer-1', interfaces='eths1'})",
"memory_usage_ok({host='streamer-1'})",
"cpu_load_ok({host='streamer-1'})",
"health_check({host='streamer-1', interfaces='eths1'})" // Combines interfaces_online(), memory_usage_ok(), cpu_load_ok()
]