bug-1906113: document metrics

This documents all metrics in a format similar to how Glean defines metrics. This adds a filter when running tests that raises an error if a metric is emitted that's not documented. This adds a Sphinx extension that automatically documents all metrics defined in `tecken/tecken_metrics.yaml`.
mozilla-services · Aug 6, 2024 · 725b8e3 · 725b8e3
1 parent 6df25dd
commit 725b8e3
Show file tree

Hide file tree

Showing 11 changed files with 343 additions and 41 deletions.
diff --git a/.editorconfig b/.editorconfig
@@ -17,5 +17,5 @@ indent_style = tab
 [*.py]
 max_line_length = 88
 
-[*.{css,js,json,html,yml}]
+[*.{css,js,json,html,yml,yaml}]
 indent_size = 2
diff --git a/docs/conf.py b/docs/conf.py
@@ -57,6 +57,7 @@
     "sphinxcontrib.httpdomain",
     "everett.sphinxext",
     "exts.adr_log",
+    "exts.document_metrics",
 ]
 
 # Add any paths that contain templates here, relative to this directory.

diff --git a/docs/exts/eliot_metrics.py → docs/exts/document_metrics.py b/docs/exts/eliot_metrics.py → docs/exts/document_metrics.py
@@ -24,13 +24,13 @@ def build_table(table):
     yield "  ".join("=" * width for width in col_size)
     yield "  ".join(
         header + (" " * (width - len(header)))
-        for header, width in zip(table[0], col_size)
+        for header, width in zip(table[0], col_size, strict=True)
     )
     yield "  ".join("=" * width for width in col_size)
     for row in table[1:]:
         yield "  ".join(
             col + (" " * (width - len(col)))
-            for col, width in zip(row, col_size)
+            for col, width in zip(row, col_size, strict=True)
         )
     yield "  ".join("=" * width for width in col_size)
 
@@ -53,7 +53,7 @@ def generate_docs(self, clspath):
         module = sys.modules[modpath]
         metrics = getattr(module, name)
 
-        sourcename = "metrics of %s" % clspath
+        sourcename = f"metrics of {clspath}"
 
         # First build a table of metric items
         self.add_line("Table of metrics:", sourcename)
@@ -62,7 +62,7 @@ def generate_docs(self, clspath):
         table = []
         table.append(("Key", "Type"))
         for key, metric in metrics.items():
-            table.append((":py:data:`%s`" % key, metric.stat_type))
+            table.append((f":py:data:`{key}`", metric["type"]))
 
         for line in build_table(table):
             self.add_line(line, sourcename)
@@ -72,14 +72,14 @@ def generate_docs(self, clspath):
         self.add_line("", sourcename)
 
         for key, metric in metrics.items():
-            self.add_line(".. py:data:: %s" % key, sourcename)
+            self.add_line(f".. py:data:: {key}", sourcename)
             self.add_line("", sourcename)
             self.add_line("", sourcename)
-            self.add_line("   Type: %s" % metric.stat_type, sourcename)
+            self.add_line(f"   **Type**: ``{metric['type']}``", sourcename)
             self.add_line("", sourcename)
             self.add_line("", sourcename)
-            for line in textwrap.dedent(metric.description).splitlines():
-                self.add_line("   " + line, sourcename)
+            for line in textwrap.dedent(metric["description"]).splitlines():
+                self.add_line(f"   {line}", sourcename)
             self.add_line("", sourcename)
             self.add_line("", sourcename)
 

diff --git a/docs/index.rst b/docs/index.rst
@@ -17,6 +17,7 @@
    dev
    contributing
    configuration
+   metrics
    frontend
    redis
    adr_log

diff --git a/docs/metrics.rst b/docs/metrics.rst
@@ -0,0 +1,10 @@
+.. _metrics-chapter:
+
+=======
+Metrics
+=======
+
+Metrics in Tecken
+=================
+
+.. autometrics:: tecken.libmarkus.TECKEN_METRICS
diff --git a/requirements.in b/requirements.in
@@ -19,6 +19,7 @@ psycopg2==2.9.9
 pytest==8.3.2
 pytest-django==4.8.0
 python-dateutil==2.9.0.post0
+PyYAML==6.0.1
 redis==5.0.8
 requests-mock==1.12.1
 requests==2.32.3

diff --git a/requirements.txt b/requirements.txt
@@ -626,6 +626,59 @@ python-dateutil==2.9.0.post0 \
     # via
     #   -r requirements.in
     #   botocore
+pyyaml==6.0.1 \
+    --hash=sha256:04ac92ad1925b2cff1db0cfebffb6ffc43457495c9b3c39d3fcae417d7125dc5 \
+    --hash=sha256:062582fca9fabdd2c8b54a3ef1c978d786e0f6b3a1510e0ac93ef59e0ddae2bc \
+    --hash=sha256:0d3304d8c0adc42be59c5f8a4d9e3d7379e6955ad754aa9d6ab7a398b59dd1df \
+    --hash=sha256:1635fd110e8d85d55237ab316b5b011de701ea0f29d07611174a1b42f1444741 \
+    --hash=sha256:184c5108a2aca3c5b3d3bf9395d50893a7ab82a38004c8f61c258d4428e80206 \
+    --hash=sha256:18aeb1bf9a78867dc38b259769503436b7c72f7a1f1f4c93ff9a17de54319b27 \
+    --hash=sha256:1d4c7e777c441b20e32f52bd377e0c409713e8bb1386e1099c2415f26e479595 \
+    --hash=sha256:1e2722cc9fbb45d9b87631ac70924c11d3a401b2d7f410cc0e3bbf249f2dca62 \
+    --hash=sha256:1fe35611261b29bd1de0070f0b2f47cb6ff71fa6595c077e42bd0c419fa27b98 \
+    --hash=sha256:28c119d996beec18c05208a8bd78cbe4007878c6dd15091efb73a30e90539696 \
+    --hash=sha256:326c013efe8048858a6d312ddd31d56e468118ad4cdeda36c719bf5bb6192290 \
+    --hash=sha256:40df9b996c2b73138957fe23a16a4f0ba614f4c0efce1e9406a184b6d07fa3a9 \
+    --hash=sha256:42f8152b8dbc4fe7d96729ec2b99c7097d656dc1213a3229ca5383f973a5ed6d \
+    --hash=sha256:49a183be227561de579b4a36efbb21b3eab9651dd81b1858589f796549873dd6 \
+    --hash=sha256:4fb147e7a67ef577a588a0e2c17b6db51dda102c71de36f8549b6816a96e1867 \
+    --hash=sha256:50550eb667afee136e9a77d6dc71ae76a44df8b3e51e41b77f6de2932bfe0f47 \
+    --hash=sha256:510c9deebc5c0225e8c96813043e62b680ba2f9c50a08d3724c7f28a747d1486 \
+    --hash=sha256:5773183b6446b2c99bb77e77595dd486303b4faab2b086e7b17bc6bef28865f6 \
+    --hash=sha256:596106435fa6ad000c2991a98fa58eeb8656ef2325d7e158344fb33864ed87e3 \
+    --hash=sha256:6965a7bc3cf88e5a1c3bd2e0b5c22f8d677dc88a455344035f03399034eb3007 \
+    --hash=sha256:69b023b2b4daa7548bcfbd4aa3da05b3a74b772db9e23b982788168117739938 \
+    --hash=sha256:6c22bec3fbe2524cde73d7ada88f6566758a8f7227bfbf93a408a9d86bcc12a0 \
+    --hash=sha256:704219a11b772aea0d8ecd7058d0082713c3562b4e271b849ad7dc4a5c90c13c \
+    --hash=sha256:7e07cbde391ba96ab58e532ff4803f79c4129397514e1413a7dc761ccd755735 \
+    --hash=sha256:81e0b275a9ecc9c0c0c07b4b90ba548307583c125f54d5b6946cfee6360c733d \
+    --hash=sha256:855fb52b0dc35af121542a76b9a84f8d1cd886ea97c84703eaa6d88e37a2ad28 \
+    --hash=sha256:8d4e9c88387b0f5c7d5f281e55304de64cf7f9c0021a3525bd3b1c542da3b0e4 \
+    --hash=sha256:9046c58c4395dff28dd494285c82ba00b546adfc7ef001486fbf0324bc174fba \
+    --hash=sha256:9eb6caa9a297fc2c2fb8862bc5370d0303ddba53ba97e71f08023b6cd73d16a8 \
+    --hash=sha256:a08c6f0fe150303c1c6b71ebcd7213c2858041a7e01975da3a99aed1e7a378ef \
+    --hash=sha256:a0cd17c15d3bb3fa06978b4e8958dcdc6e0174ccea823003a106c7d4d7899ac5 \
+    --hash=sha256:afd7e57eddb1a54f0f1a974bc4391af8bcce0b444685d936840f125cf046d5bd \
+    --hash=sha256:b1275ad35a5d18c62a7220633c913e1b42d44b46ee12554e5fd39c70a243d6a3 \
+    --hash=sha256:b786eecbdf8499b9ca1d697215862083bd6d2a99965554781d0d8d1ad31e13a0 \
+    --hash=sha256:ba336e390cd8e4d1739f42dfe9bb83a3cc2e80f567d8805e11b46f4a943f5515 \
+    --hash=sha256:baa90d3f661d43131ca170712d903e6295d1f7a0f595074f151c0aed377c9b9c \
+    --hash=sha256:bc1bf2925a1ecd43da378f4db9e4f799775d6367bdb94671027b73b393a7c42c \
+    --hash=sha256:bd4af7373a854424dabd882decdc5579653d7868b8fb26dc7d0e99f823aa5924 \
+    --hash=sha256:bf07ee2fef7014951eeb99f56f39c9bb4af143d8aa3c21b1677805985307da34 \
+    --hash=sha256:bfdf460b1736c775f2ba9f6a92bca30bc2095067b8a9d77876d1fad6cc3b4a43 \
+    --hash=sha256:c8098ddcc2a85b61647b2590f825f3db38891662cfc2fc776415143f599bb859 \
+    --hash=sha256:d2b04aac4d386b172d5b9692e2d2da8de7bfb6c387fa4f801fbf6fb2e6ba4673 \
+    --hash=sha256:d483d2cdf104e7c9fa60c544d92981f12ad66a457afae824d146093b8c294c54 \
+    --hash=sha256:d858aa552c999bc8a8d57426ed01e40bef403cd8ccdd0fc5f6f04a00414cac2a \
+    --hash=sha256:e7d73685e87afe9f3b36c799222440d6cf362062f78be1013661b00c5c6f678b \
+    --hash=sha256:f003ed9ad21d6a4713f0a9b5a7a0a79e08dd0f221aff4525a2be4c346ee60aab \
+    --hash=sha256:f22ac1c3cac4dbc50079e965eba2c1058622631e526bd9afd45fedd49ba781fa \
+    --hash=sha256:faca3bdcf85b2fc05d06ff3fbc1f83e1391b3e724afa3feba7d13eeab355484c \
+    --hash=sha256:fca0e3a251908a499833aa292323f32437106001d436eca0e6e7833256674585 \
+    --hash=sha256:fd1592b3fdf65fff2ad0004b5e363300ef59ced41c2e6b3a99d4089fa8c5435d \
+    --hash=sha256:fd66fc5d0da6d9815ba2cebeb4205f95818ff4b79c3ebe268e75d961704af52f
+    # via -r requirements.in
 redis==5.0.8 \
     --hash=sha256:0c5b10d387568dfe0698c6fad6615750c24170e548ca2deac10c649d463e9870 \
     --hash=sha256:56134ee08ea909106090934adc36f65c9bcbbaecea5b21ba704ba6fb561f8eb4

diff --git a/tecken/apps.py b/tecken/apps.py
@@ -21,7 +21,7 @@
 from django.apps import AppConfig
 
 from tecken.libdockerflow import get_release_name
-from tecken.libmarkus import METRICS, setup_markus
+from tecken.libmarkus import METRICS, set_up_markus
 
 
 logger = logging.getLogger("django")
@@ -104,7 +104,11 @@ def ready(self):
 
     @staticmethod
     def _configure_markus():
-        setup_markus(settings.MARKUS_BACKENDS, settings.HOSTNAME)
+        set_up_markus(
+            backends=settings.MARKUS_BACKENDS,
+            hostname=settings.HOSTNAME,
+            debug=settings.LOCAL_DEV_ENV or settings.TEST_ENV,
+        )
 
     @staticmethod
     def _fix_default_redis_connection():

diff --git a/tecken/libmarkus.py b/tecken/libmarkus.py
@@ -3,25 +3,50 @@
 # file, You can obtain one at https://mozilla.org/MPL/2.0/.
 
 import json
+from pathlib import Path
 import time
 from functools import partialmethod
 
 import markus
 from markus import INCR, GAUGE, HISTOGRAM, TIMING
 from markus.backends import BackendBase
-from markus.filters import AddTagFilter
+from markus.filters import AddTagFilter, RegisteredMetricsFilter
+import yaml
+
 
 _IS_MARKUS_SETUP = False
+
 METRICS = markus.get_metrics("tecken")
 
 
-def setup_markus(backends, hostname):
+# Complete index of all Eliot metrics. This is used in documentation and to filter
+# outgoing metrics.
+def _load_registered_metrics():
+    # Load the eliot_metrics.yaml file in this directory
+    path = Path(__file__).parent / "tecken_metrics.yaml"
+    with open(path) as fp:
+        data = yaml.safe_load(fp)
+    return data
+
+
+TECKEN_METRICS = _load_registered_metrics()
+
+
+def set_up_markus(backends, hostname, debug=False):
     global _IS_MARKUS_SETUP, METRICS
     if _IS_MARKUS_SETUP:
         return
 
     markus.configure(backends)
 
+    if debug:
+        # In local dev and test environments, we want the RegisteredMetricsFilter to
+        # raise exceptions when metrics are used incorrectly.
+        metrics_filter = RegisteredMetricsFilter(
+            registered_metrics=TECKEN_METRICS, raise_error=True
+        )
+        METRICS.filters.append(metrics_filter)
+
     if hostname:
         # Define host tag here instead of in the backend so it shows up in tests
         METRICS.filters.append(AddTagFilter(f"host:{hostname}"))

diff --git a/tecken/tecken_metrics.yaml b/tecken/tecken_metrics.yaml
@@ -0,0 +1,159 @@
+# statsd metrics emitted using Markus.
+#
+# When adding a new metric, make sure to add it here first.
+---
+
+tecken.api:
+  type: "timing"
+  description: |
+    Timer for how long it takes to handle an API request.
+
+    Tags:
+
+    * ``endpoint``: the API endpoint name
+
+tecken.api_stats:
+  type: "timing"
+  description: |
+    Timer for how long it takes to compute upload API stats.
+
+tecken.download_symbol:
+  type: "timing"
+  description: |
+    Timer for how long it takes to handle a download API request.
+
+tecken.download_symbol_code_id_lookup:
+  type: "incr"
+  description: |
+    Counter for successful symbol code id lookups in download API.
+
+tecken.gunicorn_worker_abort:
+  type: "incr"
+  description: |
+    Counter for number of gunicorn workers were terminated.
+
+tecken.remove_orphaned_files.delete_file:
+  type: "incr"
+  description: |
+    Counter for successfully deleted orphaned files.
+
+tecken.remove_orphaned_files.delete_file_error:
+  type: "incr"
+  description: |
+    Counter for errors encountered when deleting orphaned files.
+
+tecken.remove_orphaned_files.timing:
+  type: "timing"
+  description: |
+    Timer for how long it took to run the ``remove_orphaned_files`` Django
+    command.
+
+tecken.symboldownloader_exists:
+  type: "timing"
+  description: |
+    Timer for retrieving object metadata indicating the symbols file exists
+    in storage.
+
+tecken.symboldownloader.file_age_days:
+  type: "histogram"
+  description: |
+    Histogram for how long the file has been in storage. Value is in days.
+
+tecken.syminfo.lookup.cached:
+  type: "incr"
+  description: |
+    Counter for symbol information lookup.
+
+    Tags:
+
+    * ``result``: true or false as to whether symbol information came from the
+      cache
+
+tecken.syminfo.lookup.timing:
+  type: "timing"
+  description: |
+    Timer for how long it takes to look up symbol information.
+
+tecken.upload_archive:
+  type: "timing"
+  description: |
+    Timer for how long it takes to handle an upload API request.
+
+tecken.upload_download_by_url:
+  type: "timing"
+  description: |
+    Timer for how long it takes to download the symbols zip archive from the
+    download url indicated in the upload API payload.
+
+tecken.upload_dump_and_extract:
+  type: "timing"
+  description: |
+    Timer for how long it takes to unzip the symbols zip archive and extract
+    files to a temporary directory on disk.
+
+tecken.upload_file_exists:
+  type: "timing"
+  description: |
+    Timer for retrieving object metadata for a file.
+
+tecken.upload_file_upload:
+  type: "timing"
+  description: |
+    Timer for how long it takes to handle uploading a file to storage. This
+    includes determining whether the file already exists,
+    compressing/decompressing it, saving a record to the database, and any
+    other processing required.
+
+tecken.upload_file_upload_skip:
+  type: "incr"
+  description: |
+    Counter for files to be uploaded to storage but were skipped because
+    they're already there.
+
+tecken.upload_file_upload_upload:
+  type: "incr"
+  description: |
+    Counter for each file successfully uploaded to storage.
+
+tecken.upload_gzip_payload:
+  type: "timing"
+  description: |
+    Timer for how long it takes to gzip a file before uploading to storage.
+
+tecken.upload_put_object:
+  type: "timing"
+  description: |
+    Timer for uploading a file to storage.
+
+tecken.upload_skip_early_compressed:
+  type: "incr"
+  description: |
+    Counter for each time we skip uploading a *compressed* symbol file because
+    it exists in storage already.
+
+tecken.upload_skip_early_uncompressed:
+  type: "incr"
+  description: |
+    Counter for each time we skip uploading an *uncompressed* symbol file
+    because it exists in storage already.
+
+tecken.upload_uploads:
+  type: "incr"
+  description: |
+    Counter for upload API requests that were successfully completed.
+
+    Tags:
+
+    * ``try``: whether or not it was a try upload
+    * ``bucket``: the storage bucket name files were uploaded to
+
+tecken.useradmin_is_blocked_in_auth0:
+  type: "timing"
+  description: |
+    Timer for how long it takes to determine if the user is blocked in Auth0.
+
+tecken.sentry_scrub_error:
+  type: "incr"
+  description: |
+    Emitted when there are errors scrubbing Sentry events. Monitor these
+    because it means we're missing Sentry event data.
-Original file line number
+Diff line change
@@ Expand Up / @@ -17,6 +17,7 @@ @@
        dev
        contributing
        configuration
+       metrics
        frontend
        redis
        adr_log
@@ Expand Down @@