From 5fd16b340c397c6e628b2eb879b6844ecffad56b Mon Sep 17 00:00:00 2001
From: mulhern <amulhern@redhat.com>
Date: Fri, 29 Mar 2024 12:47:46 -0400
Subject: [PATCH] Add a simple prediction script for metadatat size

Signed-off-by: mulhern <amulhern@redhat.com>
---
 .github/workflows/main.yml                  |   1 +
 .github/workflows/weekly.yml                |   1 +
 misc_scripts/pool_metadata_size_estimate.py | 117 ++++++++++++++++++++
 3 files changed, 119 insertions(+)
 create mode 100755 misc_scripts/pool_metadata_size_estimate.py

diff --git a/.github/workflows/main.yml b/.github/workflows/main.yml
index 96ba572..4beb4ca 100644
--- a/.github/workflows/main.yml
+++ b/.github/workflows/main.yml
@@ -21,6 +21,7 @@ jobs:
               pylint
               python3-dbus
               python3-dbus-python-client-gen
+              python3-justbytes
               python3-matplotlib
               python3-numpy
               python3-pygithub
diff --git a/.github/workflows/weekly.yml b/.github/workflows/weekly.yml
index 081b927..849d451 100644
--- a/.github/workflows/weekly.yml
+++ b/.github/workflows/weekly.yml
@@ -19,6 +19,7 @@ jobs:
               pylint
               python3-dbus
               python3-dbus-python-client-gen
+              python3-justbytes
               python3-matplotlib
               python3-numpy
               python3-pygithub
diff --git a/misc_scripts/pool_metadata_size_estimate.py b/misc_scripts/pool_metadata_size_estimate.py
new file mode 100755
index 0000000..ded1542
--- /dev/null
+++ b/misc_scripts/pool_metadata_size_estimate.py
@@ -0,0 +1,117 @@
+#!/usr/bin/python3
+"""
+pool_metadata_size_estimate: estimate the inrease of pool level metadata due to 
+addition of blockdevs.
+"""
+
+# isort: STDLIB
+import argparse
+
+# isort: THIRDPARTY
+from justbytes import Range
+
+STRING = 255
+UUID = 36
+
+
+def metadata_str(value):
+    """
+    Parse a value representing the size of a metadata string.
+    """
+    value = int(value)
+    if value > 255:
+        raise argparse.ArgumentTypeError(
+            f"Exceeds maximum length {STRING} for metadata string."
+        )
+
+    if value < 0:
+        raise argparse.ArgumentTypeError("Negative lengths prohbited.")
+
+    return value
+
+
+def nat(value):
+    """
+    Parse a value representing a natural number.
+    """
+    value = int(value)
+    if value < 0:
+        raise argparse.ArgumentTypeError("Must be a natural number.")
+    return value
+
+
+def gen_parser():
+    """
+    Generate parser.
+    """
+    parser = argparse.ArgumentParser(
+        description=(
+            "Calculate size increase in Stratis pool-level metadata due to "
+            "addition of new devices."
+        )
+    )
+
+    parser.add_argument(
+        "num_devices",
+        help="Number of devices to be added",
+        type=nat,
+    )
+
+    parser.add_argument(
+        "--hardware-info",
+        dest="hardware_info",
+        help="Bytes required to represent hardware info.",
+        type=metadata_str,
+    )
+
+    parser.add_argument(
+        "--user-info",
+        dest="user_info",
+        help="Bytes required to represent user info.",
+        type=metadata_str,
+    )
+
+    return parser
+
+
+def f(*, hardware_info=None, user_info=None):
+    """
+    Calculate the bytes required to store num_devices in metadata.
+
+    :param hardware_info: the len of the hardware info
+    :param user_info: the len of the user info
+    """
+    commas = (0 if hardware_info is None else 1) + (0 if user_info is None else 1)
+
+    m = (
+        commas
+        + (0 if hardware_info is None else hardware_info + 18)
+        + (0 if user_info is None else user_info + 13)
+        + (UUID + 9)
+        + 3
+    )
+
+    return m
+
+
+def main():
+    """
+    Main method
+    """
+
+    parser = gen_parser()
+    args = parser.parse_args()
+
+    num_devices = args.num_devices
+
+    user_info = args.user_info
+    hardware_info = args.hardware_info
+
+    m = f(hardware_info=hardware_info, user_info=user_info)
+    print(f"Slope: {m} bytes")
+    prediction = m * num_devices
+    print(f"Predicted use: {prediction} bytes, approx {Range(prediction)}")
+
+
+if __name__ == "__main__":
+    main()