Merge ~ballot/content-cache-charm/+git/content-cache-charm:request_per_ip_action into content-cache-charm:master

[Benjamin Allot]

Unmerged commits

f5e98b3... by Benjamin Allot on 2021-03-08

WIP

diff --git a/actions.yaml b/actions.yaml
 new file mode 100644
index 0000000..6dcdfc3
--- /dev/null
+++ b/actions.yaml
@@ -0,0 +1,11 @@
+get-requests-per-ip:
+  description: Get a quick view of the number of requests per ip on the last 15 minutes
+  params:
+    minutes:
+      type: int
+      description: Look in the last X minutes to gather the requests
+    range:
+      type: string
+      description: |
+        Pass a string in the regexp format to express a time range.
+        i.e: "16:(3[5-9]|4[0-3]):" would look at all the logs between 16:35 and 16:43
diff --git a/actions/get-requests-per-ip b/actions/get-requests-per-ip
 new file mode 100644
index 0000000..9e9e758
--- /dev/null
+++ b/actions/get-requests-per-ip
@@ -0,0 +1,67 @@
+#!/usr/bin/env python3
+
+
+import subprocess
+import json
+
+
+from charmhelpers.core import hookenv
+
+
+LAST_MINUTES = 20
+TOP_IPS = 20
+
+
+def extract_logs_from_last_minutes(minutes: int) -> str:
+    """Extract the logs from the last `minutes` minutes.
+
+    Return the logs written in the last minutes.
+
+    :param minutes: Number of last minutes used to extract relevant logs
+    """
+    # TODO: use subprocess or pure python. TDB
+    pass
+
+
+def extract_logs_from_time_range(time_range: str) -> str:
+    """Extract the logs for the selected time_range.
+
+    This time range is converted in the right format depending on the haproxy
+    or nginx logs and return the associated logs.
+
+    :param time_range: String in the regexp form to filter the log lines
+    """
+    # TODO: use subprocess or pure python. TDB
+
+
+def extract_top_ips(logs_data: str, num: int) -> str:
+    """Extract the number of requests per IP.
+
+    Return the `num` IPs with the greatest number of requests
+    in the provided logs.
+    """
+    # TODO: probably use pure python here.
+
+
+def main():
+    """Main entry point."""
+
+    action_data = hookenv.function_get()
+    # Set the default to LAST_MINUTES if no arguments
+    if "minutes" not in action_data and "range" not in action_data:
+        action_data["minutes"] = 20
+
+    # if we have range, we should ignore last_minutes
+    elif "range" in action_data:
+        logs_data = extract_logs_from_time_range(action_data["range"])
+    else:
+        logs_data = extract_logs_from_last_minutes(action_data["minutes"])
+
+    print(extract_top_ips(logs_data, TOP_IPS))
+
+
+if __name__ == "__main__":
+    try:
+        main()
+    finally:
+        print("End of action")