Sampling in Python SDK

You can reduce the amount of data that the Python agent sends to the Thundra Web Console by enabling sampling. You can either add built-in sampling rules or you can provide a custom sampler and implement the sampling logic by yourself. Either option allows you to sample Trace, Metric, and/or Log data using the Sampling API.
Sampling API
A sampler should implement the abstract class shown below:
1
from abc import ABC, abstractmethod
2
​
3
class BaseSampler(ABC):
4
​
5
    @abstractmethod
6
    def is_sampled(self, args):
7
        raise Exception("should be implemented")
Copied!
Custom Sampler
A custom sampler is a class with the is_sampled function. The is_sampled function should return as true if you want to send data to Thundra, and should return as false if you want to omit the data.
1
from thundra.samplers import BaseSampler
2
​
3
class CustomSampler(BaseSampler):
4
  def is_sampled(self, data=None):
5
    # TODO: Your custom logic here
Copied!
Count Aware Sampler
The Count Aware Sampler enables you to sample metric data with a count frequency. For example, if the count frequency is 500, data will be sampled with every 500 Lambda invocations.
1
from thundra.samplers import CountAwareSampler
2
​
3
# Sample for every 500 call
4
sampler = CountAwareSampler(count_freq=500)
Copied!
You can update count frequency without re-deploying Lambda functions. After setting Count Aware Sampler as the sampler, you can change the count frequency by setting the thundra_agent_lambda_metric_sample_sampler_countAware_countFreq environment variable in AWS Lambda.
Time Aware Sampler
The Time Aware Sampler enables you to sample metric data with a time frequency. For example, if the time frequency is 300,000, data will be sampled every 300 milliseconds within consecutive Lambda invocations.
1
from thundra.samplers import TimeAwareSampler
2
​
3
# Sample for every 1000 milliseconds
4
sampler = TimeAwareSampler(time_freq=1000)
Copied!
You can update time frequency without re-deploying Lambda functions. After setting Time Aware Sampler as the sampler, you can change the time frequency by setting the thundra_agent_lambda_metric_sample_sampler_timeAware_timeFreq environment variable in AWS Lambda.
Composite Sampler
The Composite Sampler enables you to combine multiple Sampler using the specified composition operation (and or or).
An example using TimeAwareSampler and CountAwareSampler with or can be seen below:
1
from thundra.samplers import (
2
    TimeAwareSampler, CountAwareSampler,
3
    CompositeSampler
4
)
5
​
6
​
7
sampler = CompositeSampler(
8
    samplers=[
9
        TimeAwareSampler(),
10
        CountAwareSampler()
11
    ],
12
    operator='or'
13
)
Copied!
Metric Sampling
To use a sampler for metric sampling, you should set the sampler using the set_sampler function from thundra.plugins.metric.metric_support.
Default Metric Sampling
By default, metrics are sampled every 100 invocation or every 5 minutes, whichever comes first. Default invocation count and time frequencies can be configured through the thundra_agent_lambda_sampler_countAware_countFreq and thundra_agent_lambda_sampler_timeAware_timeFreq environment variables. You can also set your own custom sampler to be the preferred method over default configuration.
The samplers available include:
Composite Sampler
Count Aware Sampler
Time Aware Sampler
Count Aware Sampler
To use the Count Aware Sampler for metric sampling, you should import thundra.samplers.CountAwareSampler and set it as the sampler using the set_sampler function from thundra.plugins.metric.metric_support. See the example below:
1
from thundra.thundra_agent import Thundra
2
from thundra.samplers import CountAwareSampler
3
from thundra.plugins.metric import metric_support
4
​
5
metric_support.set_sampler(CountAwareSampler(count_freq=500))
6
thundra = Thundra()
7
​
8
# Main handler
9
@thundra
10
def handler(event, context):  
11
    return {"result": "hello"}
Copied!
You can update count frequency without re-deploying Lambda functions. After setting Count Aware Sampler as the sampler, you can change the count frequency by setting the thundra_agent_lambda_metric_sample_sampler_countAware_countFreq environment variable in AWS Lambda.
Time Aware Sampler
To use Time Aware Sampler for metric sampling, you should import thundra.samplers.TimeAwareSampler and set it as the sampler using the set_sampler function from thundra.plugins.metric.metric_support. See the example below:
1
from thundra.thundra_agent import Thundra
2
from thundra.samplers import TimeAwareSampler
3
from thundra.plugins.metric import metric_support
4
​
5
metric_support.set_sampler(TimeAwareSampler(time_freq=1000))
6
thundra = Thundra()
7
​
8
# Main handler
9
@thundra
10
def handler(event, context):  
11
    return {"result": "hello"}
Copied!
You can update time frequency without re-deploying Lambda functions. After setting Time Aware Sampler as the sampler, you can change the time frequency by setting the thundra_agent_lambda_metric_sample_sampler_timeAware_timeFreq environment variable in AWS Lambda.
Custom sampler
An example usage of a custom sampler for the metric plugin is shown below:
1
import requests
2
​
3
from thundra.thundra_agent import Thundra
4
​
5
from thundra.samplers import BaseSampler
6
from thundra.plugins.metric import metric_support
7
​
8
class CustomSampler(BaseSampler):
9
  def is_sampled(self, data=None):
10
    # TODO: Your custom logic here
11
​
12
metric_support.set_sampler(CustomSampler())
13
​
14
thundra = Thundra()
15
# Main handler
16
@thundra
17
def handler(event, context):
18
  return {"result": "hello"}
Copied!
Log Sampling
To use a sampler for log sampling, you should set the sampler using the set_sampler function from thundra.plugins.log.log_support.
The samplers available include:
Composite Sampler
Count Aware Sampler
Time Aware Sampler
Custom sampler
An example usage of a custom sampler for the log plugin is shown below:
1
import requests
2
​
3
from thundra.thundra_agent import Thundra
4
​
5
from thundra.samplers import BaseSampler
6
from thundra.plugins.log import log_support
7
​
8
class CustomSampler(BaseSampler):
9
  def is_sampled(self, log=None):
10
    # TODO: Your custom logic here
11
​
12
log_support.set_sampler(CustomSampler())
13
​
14
thundra = Thundra()
15
# Main handler
16
@thundra
17
def handler(event, context):
18
  return {"result": "hello"}
Copied!
Trace Sampling
To use a sampler for trace sampling, you should set the sampler using the set_sampler function from thundra.plugins.trace.trace_support.
Sampling Target
For traces, sampling is only applied to the root span (not the other internal spans) of the invocation which represents the actual Lambda invocation. According to decision, if the root span is sampled, then all of the spans in the invocation are sampled.
The samplers available include:
Composite Sampler
Duration Aware Sampler
Error Aware Sampler
Duration Aware Sampler
The Duration Aware Sampler enables you to sample trace data according to the duration of the Lambda. This sampler is ideal if, for example, you want to send trace data only if the duration of the Lambda is longer than 500 milliseconds.
See the example below:
Duration Aware Sampler
1
import requests
2
​
3
from thundra.thundra_agent import Thundra
4
from thundra.plugins.log.thundra_log_handler import ThundraLogHandler
5
​
6
from thundra.samplers import DurationAwareSampler
7
from thundra.plugins.trace import trace_support
8
​
9
trace_support.set_sampler(DurationAwareSampler(500, True))
10
​
11
thundra = Thundra()
12
# Main handler
13
@thundra
14
def handler(event, context):  
15
    return {"result": "hello"}
Copied!
Error Aware Sampler
The Error Aware Sampler enables you to sample trace data according to erroneous invocations of a Lambda. For example, this would be ideal to use if you want to send trace data only from a Lambda whose execution results in an error.
See an example below:
Error Aware Sampler
1
import requests
2
​
3
from thundra.thundra_agent import Thundra
4
​
5
from thundra.samplers import ErrorAwareSampler
6
from thundra.plugins.trace import trace_support
7
​
8
trace_support.set_sampler(ErrorAwareSampler())
9
​
10
thundra = Thundra()
11
# Main handler
12
@thundra
13
def handler(event, context):  
14
    return {"result": "hello"}
Copied!
Custom sampler
A custom sampler can be added for the trace plugin by setting the sampler using the set_sampler function from thundra.plugins.trace.trace_support.
Custom Sampler
1
import requests
2
​
3
from thundra.thundra_agent import Thundra
4
​
5
from thundra.samplers import BaseSampler
6
from thundra.plugins.trace import trace_support
7
​
8
class CustomSampler(BaseSampler):
9
  def is_sampled(self, span=None):
10
    # TODO: Your custom logic here
11
​
12
trace_support.set_sampler(CustomSampler())
13
​
14
thundra = Thundra()
15
# Main handler
16
@thundra
17
def handler(event, context):
18
  return {"result": "hello"}
Copied!
​