From 9099d07dfa6dc6436f8af44972a1d5e43b3e57da Mon Sep 17 00:00:00 2001
From: Scott Lamb <slamb@slamb.org>
Date: Wed, 10 Mar 2021 08:12:49 -0800
Subject: [PATCH] improve panic messages and docs (#112)

---
 guide/troubleshooting.md | 23 +++++++++++++++++++----
 server/src/main.rs       | 31 +++++++++++++++++++++++++++++++
 2 files changed, 50 insertions(+), 4 deletions(-)

diff --git a/guide/troubleshooting.md b/guide/troubleshooting.md
index 8b2e949..f4a52df 100644
--- a/guide/troubleshooting.md
+++ b/guide/troubleshooting.md
@@ -7,7 +7,7 @@ need more help.
 * [Troubleshooting](#troubleshooting)
     * [Viewing Moonfire NVR's logs](#viewing-moonfire-nvrs-logs)
         * [Flushes](#flushes)
-        * [`thread '...' panicked` errors](#thread--panicked-errors)
+        * [Panic errors](#panic-errors)
         * [Slow operations](#slow-operations)
         * [Camera stream errors](#camera-stream-errors)
     * [Problems](#problems)
@@ -76,6 +76,17 @@ PPPP = module path
 ...  = message body
 ```
 
+Moonfire NVR names a few important thread types as follows:
+
+*   `main`: during `moonfire-nvr run`, the main thread does initial setup then
+    just waits for the other threads. In other subcommands, it does everything.
+*   `s-CAMERA-TYPE`: there is one of these threads for every recorded stream
+    (up to two per camera, where `TYPE` is `main` or `sub`). These threads read
+    frames from the cameras via RTSP and write them to disk.
+*   `sync-PATH`: there is one of these threads for every sample file directory.
+    These threads call `fsync` to commit sample files to disk, delete old sample
+    files, and flush the database.
+
 You can use the following command to teach [`lnav`](http://lnav.org/) Moonfire
 NVR's log format:
 
@@ -149,17 +160,21 @@ This log message is packed with debugging information:
     of recordings, and the IDs of each recording. For GCed recordings, the
     sizes are omitted (as this information is not stored).
 
-### `thread '...' panicked` errors
+### Panic errors
 
 Errors like the one below indicate a serious bug in Moonfire NVR. Please
 file a bug if you see one. It's helpful to set the `RUST_BACKTRACE`
 environment variable to include more information.
 
 ```
-thread 's-peck_west-main' panicked at 'should always be an unindexed sample', /usr/local/src/moonfire-nvr/server/db/writer.rs:750:54
-note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace
+E20210304 11:09:29.230 main s-peck_west-main] panic at 'src/moonfire-nvr/server/db/writer.rs:750:54': should always be an unindexed sample
+
+(set environment variable RUST_BACKTRACE=1 to see backtraces)"
 ```
 
+In this case, a stream thread (one starting with `s-`) panicked. That stream
+won't record again until Moonfire NVR is restarted.
+
 ### Slow operations
 
 Warnings like the following indicate that some operation took more than 1
diff --git a/server/src/main.rs b/server/src/main.rs
index cc81553..16ed058 100644
--- a/server/src/main.rs
+++ b/server/src/main.rs
@@ -5,6 +5,7 @@
 #![cfg_attr(all(feature = "nightly", test), feature(test))]
 
 use log::{debug, error};
+use std::fmt::Write;
 use std::str::FromStr;
 use structopt::StructOpt;
 
@@ -107,6 +108,34 @@ impl Args {
     }
 }
 
+/// Custom panic hook that logs instead of directly writing to stderr.
+///
+/// This means it includes a timestamp and is more recognizable as a serious
+/// error (including console color coding by default, a format `lnav` will
+/// recognize, etc.).
+fn panic_hook(p: &std::panic::PanicInfo) {
+    let mut msg;
+    if let Some(l) = p.location() {
+        msg = format!("panic at '{}'", l);
+    } else {
+        msg = "panic".to_owned();
+    }
+    if let Some(s) = p.payload().downcast_ref::<&str>() {
+        write!(&mut msg, ": {}", s).unwrap();
+    }
+    let b = failure::Backtrace::new();
+    if b.is_empty() {
+        write!(
+            &mut msg,
+            "\n\n(set environment variable RUST_BACKTRACE=1 to see backtraces)"
+        )
+        .unwrap();
+    } else {
+        write!(&mut msg, "\n\nBacktrace:\n{}", b).unwrap();
+    }
+    error!("{}", msg);
+}
+
 fn main() {
     let args = Args::from_args();
     let mut h = mylog::Builder::new()
@@ -126,6 +155,8 @@ fn main() {
         .build();
     h.clone().install().unwrap();
 
+    std::panic::set_hook(Box::new(&panic_hook));
+
     let r = {
         let _a = h.async_scope();
         args.run()