From 61ef3ce76c50142c7fbd5e71f0fb82e1377b530f Mon Sep 17 00:00:00 2001
From: Patrick Double <pat@patdouble.com>
Date: Tue, 13 Aug 2024 14:18:50 -0500
Subject: [PATCH] find_media_errors.py: run daily

---
 Dockerfile                      |  1 +
 dvrprocess/common/config.py     |  2 +-
 dvrprocess/find_media_errors.py | 18 +++++++++++-------
 logrotate.conf                  |  8 ++++++++
 media-errors.sh                 |  6 ++++++
 5 files changed, 27 insertions(+), 8 deletions(-)
 create mode 100755 media-errors.sh

diff --git a/Dockerfile b/Dockerfile
index 7ee407d..121777a 100644
--- a/Dockerfile
+++ b/Dockerfile
@@ -72,6 +72,7 @@ COPY tvshow-summary.sh /etc/cron.daily/tvshow-summary
 #COPY comchap-apply.sh /etc/cron.daily/comchap-apply
 COPY comtune-apply.sh /etc/cron.daily/comtune-apply
 COPY transcode-apply.sh /etc/cron.hourly/transcode-apply
+COPY media-errors.sh /etc/cron.daily/media-errors
 COPY profanity-filter-apply.sh /etc/cron.daily/profanity-filter-apply
 COPY logrotate.conf /etc/logrotate.d/dvr
 COPY sendmail-log.sh /usr/sbin/sendmail
diff --git a/dvrprocess/common/config.py b/dvrprocess/common/config.py
index e776f1f..cef3b85 100644
--- a/dvrprocess/common/config.py
+++ b/dvrprocess/common/config.py
@@ -289,7 +289,7 @@ def get_file_config(path: str) -> ConfigParser:
     config = ConfigParser()
     if os.path.exists(config_path):
         config.read(config_path)
-        if config.get('general', 'fingerprint') != _mkv_fingerprint(path):
+        if config.get('general', 'fingerprint', fallback=None) != _mkv_fingerprint(path):
             config = ConfigParser()
     return config
 
diff --git a/dvrprocess/find_media_errors.py b/dvrprocess/find_media_errors.py
index 589a24c..abccf1e 100755
--- a/dvrprocess/find_media_errors.py
+++ b/dvrprocess/find_media_errors.py
@@ -31,6 +31,8 @@ def usage():
     Directory containing media. Defaults to {common.get_media_roots()}
 --nagios
     Output for Nagios monitoring. Also human readable with statistics and estimates of transcode time.
+--cache-only
+    Only report cached results, do not look for new media errors.
 --time-limit={config.get_global_config_option('background_limits', 'time_limit')}
     Limit runtime. Set to 0 for no limit.
 --ignore-compute
@@ -44,10 +46,11 @@ def find_media_errors_cli(argv):
     nagios_output = False
     time_limit = config.get_global_config_time_seconds('background_limits', 'time_limit')
     check_compute = True
+    cache_only = False
 
     try:
         opts, args = getopt.getopt(argv, "t:d:",
-                                   ["terminator=", "dir=", "nagios", "time-limit=", "ignore-compute"])
+                                   ["terminator=", "dir=", "nagios", "time-limit=", "ignore-compute", "cache-only"])
     except getopt.GetoptError:
         usage()
         return 2
@@ -70,6 +73,8 @@ def find_media_errors_cli(argv):
             time_limit = config.parse_seconds(arg)
         elif opt == '--ignore-compute':
             check_compute = False
+        elif opt == '--cache-only':
+            cache_only = True
 
     if not roots:
         roots = common.get_media_roots()
@@ -84,7 +89,7 @@ def find_media_errors_cli(argv):
         return 0
 
     generator = media_errors_generator(media_paths=media_paths, media_roots=roots,
-                                       time_limit=time_limit, check_compute=check_compute)
+                                       time_limit=time_limit, check_compute=check_compute, cache_only=cache_only)
 
     if nagios_output:
         corrupt_files = list(generator)
@@ -122,9 +127,8 @@ def __init__(self, file_name: str, host_file_path: str, size: float, error_count
 
 def media_errors_generator(media_paths: list[str], media_roots: list[str],
                            time_limit=config.get_global_config_time_seconds('background_limits', 'time_limit'),
-                           check_compute=True) -> Iterable[MediaErrorFileInfo]:
+                           check_compute=True, cache_only=False) -> Iterable[MediaErrorFileInfo]:
     time_start = time.time()
-    only_cached = False
 
     for media_path in media_paths:
         for root, dirs, files in os.walk(media_path, topdown=True):
@@ -133,19 +137,19 @@ def media_errors_generator(media_paths: list[str], media_roots: list[str],
                 cached_error_count = config.get_file_config_option(filepath, 'error', 'count')
                 if cached_error_count:
                     error_count = int(cached_error_count)
-                elif only_cached:
+                elif cache_only:
                     continue
                 else:
                     duration = time.time() - time_start
                     if 0 < time_limit < duration:
                         logger.debug(
                             f"Time limit expired after processing {common.s_to_ts(int(duration))}, limit of {common.s_to_ts(time_limit)} reached, only using cached data")
-                        only_cached = True
+                        cache_only = True
                         continue
                     if check_compute and common.should_stop_processing():
                         # when compute limit is reached, use cached data
                         logger.debug("not enough compute available, only using cached data")
-                        only_cached = True
+                        cache_only = True
                         continue
 
                     error_count = len(tools.ffmpeg.check_output(
diff --git a/logrotate.conf b/logrotate.conf
index 58a7fed..bedd99a 100644
--- a/logrotate.conf
+++ b/logrotate.conf
@@ -69,3 +69,11 @@
   missingok
   notifempty
 }
+
+/var/log/find_media_errors.log {
+  rotate 2
+  daily
+  compress
+  missingok
+  notifempty
+}
diff --git a/media-errors.sh b/media-errors.sh
new file mode 100755
index 0000000..578401a
--- /dev/null
+++ b/media-errors.sh
@@ -0,0 +1,6 @@
+#!/usr/bin/env bash
+
+# random sleep to prevent multiple programs running at the exact same time
+sleep $((5 + RANDOM % 30))
+
+exec ionice -c 3 nice -n 15 /usr/local/bin/find_media_errors.py "$@" >>/var/log/find_media_errors.log 2>&1