increase sampling rate

Summary:
This CLI is one of the biggest CLIs logging to the `cli_tool_usage` (1) Scuba table. We have recently rolled this logging out to all machines but it was enabled on devservers, ondemands, desktops, and laptops before for almost 10 years. Because of this, we want to increase sampling for the CLIs that are running millions to billions of times a day, which this CLI is one of. The per-host continuous calling leads to more data use in the Scuba table and does not add much value. We have therefore changed the sampling rate for automation from the default of 100 to 10 000, sampling for calls by humans remain unsampled. See CLI Foundation's Observability guidelines for more information on this usage logging: https://www.internalfb.com/intern/wiki/CLI_Foundation/Guidelines_&_How_To/Observability/

If you are wondering what the `cli.py.par_telemetry` library is, it is a way of communicating with the usage logging part that is integrated in every Python binary for many years. We are here leveraging it to tell it to change the sample rate. See https://fburl.com/cli.py.par_telemetry for its documentation.

(1): https://fburl.com/scuba/cli_tool_usage/qd6bnwzj

Reviewed By: genevievehelsel

Differential Revision: D50026796

fbshipit-source-id: 6a5c241635497ff7a5cf62b18ab0228bf3c007f4

eden/fs/cli/main.py

@@ -27,6 +27,8 @@
 from typing import Dict, List, Optional, Set, Tuple, Type
 
 import thrift.transport
+
+from cli.py import par_telemetry
 from eden.fs.cli.buck import get_buck_command, run_buck_command
 from eden.fs.cli.telemetry import TelemetrySample
 from eden.fs.cli.util import (
@@ -2502,6 +2504,9 @@ async def async_main(parser: argparse.ArgumentParser, args: argparse.Namespace)
 
 
 def main() -> int:
+    # This is called hundreds of millions of times on unique hosts.
+    # Increase how often it's sampled.
+    par_telemetry.set_sample_rate(automation=10000)
     parser = create_parser()
     args = parser.parse_args()