From 01543d857bfe3a56a8d4fc49b80c8556a4228b29 Mon Sep 17 00:00:00 2001
From: "Peter J. Holzer" <hjp@hjp.at>
Date: Sun, 21 Aug 2022 12:00:07 +0200
Subject: [PATCH] Record disk usage in LTsDb

---
 record_df  | 56 ++++++++++++++++++++++++++++++++++++++++++++++++++++++
 record_dus | 35 ++++++++++++++++++++++++++++++++++
 2 files changed, 91 insertions(+)
 create mode 100755 record_df
 create mode 100755 record_dus

diff --git a/record_df b/record_df
new file mode 100755
index 0000000..b5eb9ec
--- /dev/null
+++ b/record_df
@@ -0,0 +1,56 @@
+#!/usr/bin/python3
+
+import os
+import socket
+import time
+
+from ltsdb_json import LTS
+
+# Use a whitelist since there are a lot of uninteresting file system
+# types on modern linux systems
+interesting_fstypes = {"ext4", "tmpfs", "vfat"}
+
+# It's a bit weird that this works.
+hostname = socket.gethostbyaddr(socket.gethostname())[0]
+
+
+with open("/proc/mounts") as mfh:
+    for ln in mfh:
+        (dev, mountpoint, fstype, options, _, _) = ln.split()
+        if fstype not in interesting_fstypes:
+            continue
+        st = os.statvfs(mountpoint)
+        now = time.time()
+
+        # We record the capacity usable by unprivileged processes, not the
+        # total size of the filesystem. This is more relevant and also what df
+        # uses as the "100%" mark.
+        blocks_reserved = st.f_bfree - st.f_bavail
+        blocks_usable = st.f_blocks - blocks_reserved
+        blocks_used = st.f_blocks - st.f_bfree
+
+        bytes_usable = blocks_usable * st.f_bsize
+        bytes_used = blocks_used * st.f_bsize
+
+        ts = LTS(
+            {
+                "hostname": hostname,
+                "mountpoint": mountpoint,
+                "fstype": fstype,
+                "measure": "bytes_usable",
+                "unit": "bytes",
+            }
+        )
+        ts.add(now, bytes_usable)
+        ts.save()
+        ts = LTS(
+            {
+                "hostname": hostname,
+                "mountpoint": mountpoint,
+                "fstype": fstype,
+                "measure": "bytes_used",
+                "unit": "bytes",
+            }
+        )
+        ts.add(now, bytes_used)
+        ts.save()
diff --git a/record_dus b/record_dus
new file mode 100755
index 0000000..6e77810
--- /dev/null
+++ b/record_dus
@@ -0,0 +1,35 @@
+#!/usr/bin/python3
+
+import argparse
+import os
+import socket
+import subprocess
+import time
+
+from ltsdb_json import LTS
+
+ap = argparse.ArgumentParser()
+ap.add_argument("directory")
+args = ap.parse_args()
+
+# It's a bit weird that this works.
+hostname = socket.gethostbyaddr(socket.gethostname())[0]
+
+
+p = subprocess.run(["/usr/bin/dus", args.directory], capture_output=True, text=True)
+t0 = time.time()
+for ln in p.stdout.split("\n")[:-1]:
+    (size, directory) = ln.split("\t")
+
+    ts = LTS(
+        {
+            "hostname": hostname,
+            "directory": directory,
+            "measure": "diskusage",
+            "unit": "bytes",
+        }
+    )
+    ts.add(t0, int(size)*1024)
+    ts.save()
+t1 = time.time()
+print("recorded in", t1 - t0, "seconds")