From 1a4d8ef4718355db6d4d486e613e5c392bfe04f6 Mon Sep 17 00:00:00 2001
From: "Peter J. Holzer" <hjp@hjp.at>
Date: Sat, 20 Aug 2022 17:40:16 +0200
Subject: [PATCH] Use a lossy timeseries to predict when a filesystem will be
 full (PoC)

---
 predict_disk_full            | 64 ++++++++++++++++++++++++++++++++++++
 predict_disk_full_home_bytes |  4 +++
 predict_disk_full_var_bytes  |  4 +++
 3 files changed, 72 insertions(+)
 create mode 100755 predict_disk_full
 create mode 100755 predict_disk_full_home_bytes
 create mode 100755 predict_disk_full_var_bytes

diff --git a/predict_disk_full b/predict_disk_full
new file mode 100755
index 0000000..d394f0b
--- /dev/null
+++ b/predict_disk_full
@@ -0,0 +1,64 @@
+#!/usr/bin/python3
+
+# PoC
+
+# add a new entry and remove a random old entry with a probability
+# depending on the target size
+
+import argparse
+import datetime
+import json
+import random
+import sys
+import time
+import os
+
+ap = argparse.ArgumentParser()
+
+ap.add_argument("--targetsize", type=int, default=1000)
+ap.add_argument("filesystem")
+
+args = ap.parse_args()
+
+st = os.statvfs(args.filesystem)
+
+blocks_reserved = st.f_bfree - st.f_bavail
+blocks_usable = st.f_blocks - blocks_reserved
+blocks_used = st.f_blocks -  st.f_bfree
+
+# Just for readability
+bytes_usable = blocks_usable * st.f_bsize
+bytes_used = blocks_used * st.f_bsize
+
+
+filename = args.filesystem.replace("/", "_") + ".bytes.json"
+
+ts = []
+try:
+    with open(filename) as f:
+        ts = json.load(f)
+except FileNotFoundError as e:
+    pass # Ok, will create
+
+now = time.time()
+ts.append((now, bytes_used, bytes_usable))
+if random.random() > 1 - len(ts) / args.targetsize:
+    i = random.randrange(0, len(ts))
+    ts.pop(i)
+
+with open(filename, "w") as f:
+    json.dump(ts, f)
+
+for d in reversed(ts):
+    bytes_predicted = bytes_used * bytes_used / d[1]
+    print("Predicted use", bytes_predicted,
+          "in", now - d[0], "seconds",
+          "at", datetime.datetime.fromtimestamp(now + now - d[0]))
+    if bytes_predicted > bytes_usable:
+        print("Predicted disk full",
+              "in", now - d[0], "seconds",
+              "at", datetime.datetime.fromtimestamp(now + now - d[0]))
+        sys.exit(1)
+else:
+    print("safe for", now - ts[0][0], "seconds",
+          "at", datetime.datetime.fromtimestamp(now + now - ts[0][0]))
diff --git a/predict_disk_full_home_bytes b/predict_disk_full_home_bytes
new file mode 100755
index 0000000..960f9f0
--- /dev/null
+++ b/predict_disk_full_home_bytes
@@ -0,0 +1,4 @@
+#!/bin/sh
+set -e
+cd $(dirname $0)
+./predict_disk_full /home
diff --git a/predict_disk_full_var_bytes b/predict_disk_full_var_bytes
new file mode 100755
index 0000000..69e2e46
--- /dev/null
+++ b/predict_disk_full_var_bytes
@@ -0,0 +1,4 @@
+#!/bin/sh
+set -e
+cd $(dirname $0)
+./predict_disk_full /var