railwatch 0.1.1 → 0.1.2

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 27a48b972fe7616fbf46c861f8b0f1a9ffb4d99146ec2121cb05274ea3dee031
-  data.tar.gz: 45c879afc12f731e446be6aa43f8a78b61d078a8f5a7cdc6294bd94c4dfe864e
+  metadata.gz: 50160a5f39e8b422581a3b6e6cf0b07eddc3565cfc17a8ba522bc49e15cd93df
+  data.tar.gz: 4f2e2fdac0c85df89a9d595fb33f02ebcf5c02a6f88ed81cb6dc19a163a2b90d
 SHA512:
-  metadata.gz: 29512ddbb519a85a1334531289be33d59d4ceb3cf0a8c7ef55d3db04ab41ae884adf65b64ddecd40d1b70254ca76ed0ce0bbb38ec03eee0125dbae5a8b7730de
-  data.tar.gz: 67d7a5573b95aa032a227a8dc10e440bfaee6681628de90d2948f074ceb15ffb9ea839968984beaaec545d0db161a266ba94e99cd2b3a0908c14c9e98969c2c7
+  metadata.gz: 584a8545680ad9c2d24ef05f6fcfd331a933cb87a310859a6c6e7df6c447b1d1f1d1ace62ccf11b899ae9bbcf0255d3b397562e57ba402e2410ba5790a99b144
+  data.tar.gz: 0fd7c798ae6d8b745e5e19ed9a894331b09eb6b15a179d9d6cda6d6bbb4c1d8a6b8bf15eefecdb24db57fd2b956b1eabba1085e0aba7fa074810aa2d5f41374c

data/CHANGELOG.md

@@ -1,5 +1,25 @@
 # Changelog
 
+## 0.1.2 (2026-09-14)
+
+- A failed job's exception is reported once. Solid Queue re-raises it out
+  of the worker thread, where its app executor reports the same error object
+  to `Rails.error` again after the `job_attempt` execution has finished; that
+  second report (source `application.solid_queue`, unlinked) doubled every
+  failed job's occurrence count. An error object now remembers that its
+  unhandled report has shipped, across executions.
+- `require "railwatch/minitest"` includes the assertions into
+  `ActiveSupport::TestCase` through its load hook, so `assert_railwatch_queries`
+  works after the generator's one-line edit to `test/test_helper.rb` without
+  a manual `include`.
+- Solid Queue's supervisor, dispatcher, scheduler, and forked workers report
+  role `worker` after Solid Queue renames the process. They were classified
+  `web` by every health sample taken after boot, with empty Puma thread
+  stats.
+- Troubleshooting entries for `json` 3.0 against Rails 8.1.3.1 (`bin/jobs`
+  crash loop, not a Railwatch fault) and for deprecations that are counted
+  but never listed because the app's deprecation behavior lacks `:notify`.
+
 ## 0.1.1 (2026-09-14)
 
 - Token prefixes are `rw_` for an environment's ingest token and `rwp_`

data/docs/records.md

@@ -652,7 +652,9 @@ ship. Forked workers start with an empty cache.
 
 ### `deprecation`
 
-`deprecation.rails`. See `lib/railwatch/subscribers/deprecations.rb`.
+`deprecation.rails`. See `lib/railwatch/subscribers/deprecations.rb`. Rails
+only emits that notification when `config.active_support.deprecation`
+includes `:notify`; see [Troubleshooting](troubleshooting.md#deprecations-are-counted-but-never-listed).
 
 | Field | Meaning |
 |---|---|

data/docs/replacing-sentry.md

@@ -12,12 +12,12 @@ the other.
 
 ## Decide whether Railwatch covers your workload
 
-Nightwatch parity and Sentry parity are different targets. Laravel
-Nightwatch is an application-monitoring product built around framework
-executions; Railwatch deliberately maps that model onto Rails. Sentry is a
-broader, multi-language managed platform with browser replay, native/mobile
-SDKs, a large integration catalog, and generic tracing. Those broader
-capabilities are not implied by Rails/Nightwatch parity.
+Railwatch and Sentry aim at different targets. Railwatch is an
+application-monitoring product built around Rails executions: a request,
+job, scheduled task, command, or channel action and everything it did.
+Sentry is a broader, multi-language managed platform with browser replay,
+native/mobile SDKs, a large integration catalog, and generic tracing.
+Railwatch does not claim those broader capabilities.
 
 | Workload or capability | Status | Boundary |
 |---|---|---|

data/docs/testing.md

@@ -18,15 +18,12 @@ require "railwatch/rspec"
 That requires `railwatch/spec_helper`, includes `Railwatch::SpecHelper` into every
 example group, and defines the matchers below.
 
-Minitest: the same thing in `test/test_helper.rb`.
+Minitest: the same thing in `test/test_helper.rb`. The require includes
+`Railwatch::Minitest` into `ActiveSupport::TestCase` on its own.
 
 ```ruby
 require "rails/test_help"
 require "railwatch/minitest"
-
-class ActiveSupport::TestCase
-  include Railwatch::Minitest
-end
 ```
 
 Railwatch must be *enabled* in the test environment or every block would look

data/docs/troubleshooting.md

@@ -81,6 +81,42 @@ to stderr prefixed `[railwatch]`. They never go to `Rails.logger`, so they
 can't become `log` records about themselves. `Railwatch.on_unrecoverable
 { |e| ... }` gets the same failures as a callback.
 
+## `bin/jobs` dies in a loop with `wrong number of arguments (given 2, expected 1)`
+
+**Symptom.** Every Solid Queue process crashes at boot from
+`json/common.rb` in `parse`, and Railwatch reports the same
+`ArgumentError` from source `application.solid_queue` hundreds of times.
+The web server still answers until it touches a session or a JSON column.
+
+**Cause.** Not Railwatch. `json` 3.0 (September 2026) made `JSON.parse`'s
+options keyword-only, and Active Support up to 8.1.3.1 still passes them
+positionally ([rails/rails#58685](https://github.com/rails/rails/issues/58685)).
+A fresh `rails new` resolves the newest `json`, so a new app hits this
+before Railwatch is even installed.
+
+**Fix.** Pin `json` below 3 until a Rails release includes the fix:
+
+```sh
+bundle add json --version "< 3"
+```
+
+## Deprecations are counted but never listed
+
+**Symptom.** The `deprecations` counter on an execution is non-zero, but
+the Deprecations page stays empty.
+
+**Cause.** Railwatch listens to the `deprecation.rails` notification, and
+Rails only emits it when the app's deprecation behavior includes
+`:notify`. The default in development is `:log` and in production
+`:silence`, neither of which notifies.
+
+**Fix.** Add `:notify` alongside whatever the environment already does:
+
+```ruby
+# config/environments/production.rb
+config.active_support.deprecation = [:silence, :notify]
+```
+
 ## Doubled scheduled_task records
 
 **Symptom.** Every recurring task shows twice on the Scheduled tasks

data/lib/railwatch/minitest.rb

@@ -7,10 +7,9 @@ module Railwatch
   # assert_/refute_. Add to test/test_helper.rb:
   #
   #   require "railwatch/minitest"
-  #   class ActiveSupport::TestCase
-  #     include Railwatch::Minitest
-  #   end
   #
+  # That includes the module into ActiveSupport::TestCase (through its load
+  # hook, so the require order against rails/test_help does not matter).
   # Includes Railwatch::SpecHelper, so `railwatch_records(:query)` is available
   # too. See docs/testing.md.
   module Minitest
@@ -41,3 +40,5 @@ module Railwatch
     end
   end
 end
+
+ActiveSupport.on_load(:active_support_test_case) { include Railwatch::Minitest } if defined?(ActiveSupport.on_load)

data/lib/railwatch/subscribers/exceptions.rb

@@ -87,6 +87,13 @@ module Railwatch
       def capture(error, handled:, severity:, context: {}, source: nil, fingerprint: nil)
         return unless Railwatch.enabled?
         return if ignored?(error)
+        # Solid Queue re-raises a failed job's error out of the worker thread,
+        # where its executor reports it to Rails.error a second time
+        # (source application.solid_queue) after the job_attempt execution
+        # has already finished and reported it. The execution's own
+        # first_exception_report? bookkeeping is gone by then, so the error
+        # object itself remembers that its unhandled report has shipped.
+        return if !handled && reported_unhandled?(error)
 
         exe = execution
         if exe&.first_exception_observation?(error, handled)
@@ -151,9 +158,22 @@ module Railwatch
           # children (Railwatch.tail_keep?).
           exe.exception_reported = true if exe
           Railwatch.record_now(:exception, group: group, **rec)
+          remember_reported(error)
         end
       end
 
+      def reported_unhandled?(error)
+        error.instance_variable_defined?(:@__railwatch_reported)
+      rescue StandardError
+        false
+      end
+
+      def remember_reported(error)
+        error.instance_variable_set(:@__railwatch_reported, true)
+      rescue StandardError
+        nil
+      end
+
       # The group hash `capture` would assign this error. Public so
       # Railwatch.attach can file an attachment against the same issue without
       # having to re-derive the bucketing rule (source snippets are skipped:

data/lib/railwatch/subscribers/process_info.rb

@@ -84,9 +84,12 @@ module Railwatch
 
       # Puma is loaded in every process of an app that bundles it, so a Solid
       # Queue worker is recognised first, by how it was started (bin/jobs or
-      # `rake solid_queue:start`).
+      # `rake solid_queue:start`) or by the procline Solid Queue gives every
+      # process it forks ("solid-queue-worker(1.7.0): ..."), which replaces
+      # $PROGRAM_NAME after boot and would otherwise turn the supervisor,
+      # dispatcher, and scheduler into "web" on every health sample.
       def role
-        if defined?(::SolidQueue) && ($PROGRAM_NAME.include?("jobs") || ARGV.first.to_s.start_with?("solid_queue:")) then "worker"
+        if defined?(::SolidQueue) && ($PROGRAM_NAME.include?("jobs") || $PROGRAM_NAME.start_with?("solid-queue-") || ARGV.first.to_s.start_with?("solid_queue:")) then "worker"
         elsif defined?(::Rails::Console) then "console"
         elsif $PROGRAM_NAME.end_with?("rake") then "command"
         elsif defined?(::Puma) then "web"

data/lib/railwatch/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module Railwatch
-  VERSION = "0.1.1"
+  VERSION = "0.1.2"
 end

metadata

@@ -1,13 +1,13 @@
 --- !ruby/object:Gem::Specification
 name: railwatch
 version: !ruby/object:Gem::Version
-  version: 0.1.1
+  version: 0.1.2
 platform: ruby
 authors:
 - Cole Robertson
 bindir: bin
 cert_chain: []
-date: 2026-09-14 00:00:00.000000000 Z
+date: 2026-09-15 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: rails