Add support for Prometheus

.gitmodules

@@ -7,3 +7,6 @@
 [submodule "3rdparty/xxHash"]
 	path = 3rdparty/xxHash
 	url = https://github.com/Cyan4973/xxHash.git
+[submodule "3rdparty/nodejs-http-parser"]
+	path = 3rdparty/nodejs-http-parser
+	url = https://github.com/nodejs/http-parser

3rdparty/CMakeLists.txt

@@ -14,6 +14,7 @@ include("libcyaml")
 include("sentry")
 include("t1ha")
 include("xxhash")
+include("nodejs-http-parser")
 
 set(DEPS_LIST_LIBRARIES "${DEPS_LIST_LIBRARIES}" PARENT_SCOPE)
 set(DEPS_LIST_INCLUDE_DIRS "${DEPS_LIST_INCLUDE_DIRS}" PARENT_SCOPE)

3rdparty/nodejs-http-parser.cmake

@@ -0,0 +1,25 @@
+include(ExternalProject)
+file(GLOB SRC_FILES_NODEJS_HTTP_PARSER "nodejs-http-parser/http_parser.c")
+
+set(NODEJS_HTTP_PARSER_BUILD_PATH "${CMAKE_BINARY_DIR}/_deps/src/nodejs-http-parser-install-build")
+
+add_library(
+        nodejs_http_parser
+        ${SRC_FILES_NODEJS_HTTP_PARSER})
+
+target_compile_options(
+        nodejs_http_parser
+        PRIVATE
+        -Wall -Wextra -Werror -O2 -fPIC)
+
+target_compile_definitions(
+        nodejs_http_parser
+        PRIVATE
+        HTTP_PARSER_STRICT=0)
+
+target_include_directories(
+        nodejs_http_parser
+        PUBLIC
+        nodejs-http-parser/)
+
+list(APPEND DEPS_LIST_LIBRARIES "nodejs_http_parser")

README.md

@@ -23,6 +23,7 @@ system to process the network data
 Key features:
 - [Modular](https://github.com/danielealbano/cachegrand/blob/main/docs/architecture/modules.md) architecture to support widely used protocols, e.g.
   [Redis](https://github.com/danielealbano/cachegrand/blob/main/docs/architecture/modules/redis.md),
+  [Prometheus](https://github.com/danielealbano/cachegrand/blob/main/docs/architecture/modules/prometheus.md), etc.
 - [Time-series database](https://github.com/danielealbano/cachegrand/blob/main/docs/architecture/timeseries-db.md) for fast data writes and retrieval with
   primitives built to handle different data types (e.g. small strings, large blobs, jsons, etc.);
 - [Hashtable](https://github.com/danielealbano/cachegrand/blob/main/docs/architecture/hashtable.md) GET Lock-free and Wait-free operations, SET and DELETE use
@@ -33,9 +34,9 @@ Key features:
   [benches](https://github.com/danielealbano/cachegrand/blob/main/docs/benchmarks/linear-vertical-scalability.md));
 
 Planned Key Features:
-- More modules for additional platforms compatibility, e.g. [Memcache](https://github.com/danielealbano/cachegrand/blob/main/docs/architecture/modules/memcache.md),
-  etc., or to add support for monitoring, e.g. [Prometheus](https://github.com/danielealbano/cachegrand/blob/main/docs/architecture/modules/prometheus.md),
-  [DataDog](https://github.com/danielealbano/cachegrand/blob/main/docs/architecture/modules/datadog.md), etc.;
+- More modules for additional platforms compatibility, e.g. [Memcache](https://github.com/danielealbano/cachegrand/blob/main/docs/architecture/modules/memcache.md), AWS S3,
+  etc., or to add support for monitoring, e.g. [DataDog](https://github.com/danielealbano/cachegrand/blob/main/docs/architecture/modules/datadog.md),
+- etc.;
 - Ad ad-hoc network stack based on DPDK / Linux XDP (eXpress Data Path) and the FreeBSD network stack;
 - [Built for flash memories](https://github.com/danielealbano/cachegrand/blob/main/docs/architecture/timeseries-db.md#flash-memories) to be able to efficiently saturate the
   available IOPS in modern DC NVMEs and SSDs;

docs/architecture/modules.md

@@ -4,7 +4,8 @@ Modules
 cachegrand is a modular platform that provide the necessary infrastructure to implement high level business logic to
 provide end-user functionalities.
 
-Currently, the only module implemented is the one to expose Redis-like functionalities
+Currently, there are two modules implemented, one that exposes Redis-like functionalities and one to provide an endpoint
+for Prometheus to be able to collect metrics and setup monitoring & alerting.
 
 ## Architecture
 

docs/architecture/modules/prometheus.md

@@ -1,4 +1,61 @@
 Module - Prometheus
 ===================
 
-Planned to be implemented
+[Prometheus](https://prometheus.io/) is a well known and widely used monitoring and time-series database, often used in
+combination with [Grafana](https://grafana.com/).
+
+cachegrand provides a module to expose metrics for Prometheus via an HTTP endpoint that can be configured using the
+config file.
+
+## Configuration
+
+To enable the module is enough to add the protocol configuration to the `protocols` section in the config file:
+```yaml
+    - type: prometheus
+      timeout:
+        read_ms: 10000
+        write_ms: 10000
+      bindings:
+        - host: 127.0.0.1
+          port: 9090
+```
+
+The module currently doesn't provide any sort of authentication or https support.
+
+### Metrics
+
+The metrics are exposed on the `/metrics` endpoint.
+
+The available metrics are:
+
+| Name                                               | Description                                   | Type    |
+|----------------------------------------------------|-----------------------------------------------|---------|
+| cachegrand_network_total_received_packets          | Total amount of received packets              | Counter |
+| cachegrand_network_total_received_data             | Total amount of received data (in bytes)      | Counter |
+| cachegrand_network_total_sent_packets              | Total amount of sent packets                  | Counter |
+| cachegrand_network_total_sent_data                 | Total amount of sent data                     | Counter |
+| cachegrand_network_total_accepted_connections      | Total amount of accepted connections          | Counter |
+| cachegrand_network_total_active_connections        | Total amount of active connections            | Counter |
+| cachegrand_storage_total_written_data              | Total amount of data written to disk          | Counter |
+| cachegrand_storage_total_write_iops                | Total amount of write IOPS                    | Counter |
+| cachegrand_storage_total_read_data                 | Total amount of read data from disk           | Counter |
+| cachegrand_storage_total_read_iops                 | Total amount of read IOPS                     | Counter |
+| cachegrand_storage_total_open_files                | Total amount of open files                    | Counter |
+| cachegrand_network_per_minute_received_packets     | Per minute amount of received packets         | Counter |
+| cachegrand_network_per_minute_received_data        | Per minute amount of received data (in bytes) | Counter |
+| cachegrand_network_per_minute_sent_packets         | Per minute amount of sent packets             | Counter |
+| cachegrand_network_per_minute_sent_data            | Per minute amount of sent data                | Counter |
+| cachegrand_network_per_minute_accepted_connections | Per minute amount of accepted connections     | Counter |
+| cachegrand_storage_per_minute_written_data         | Per minute amount of data written to disk     | Counter |
+| cachegrand_storage_per_minute_write_iops           | Per minute amount of write IOPS               | Counter |
+| cachegrand_storage_per_minute_read_data            | Per minute amount of read data from disk      | Counter |
+| cachegrand_storage_per_minute_read_iops            | Per minute amount of read IOPS                | Counter |
+| cachegrand_uptime                                  | Uptime in seconds                             | Counter |
+
+### Metrics labels
+
+It is possible to tag the metrics with any amount of arbitrary labels using environment variables.
+
+If one or more environment variables prefixed with `CACHEGRAND_METRIC_ENV_` exist then these will be parsed by
+cachegrand and the suffix will be converted to lowercase and used as name of the label and the value of the environment
+variable will be used as value.

docs/benchmarks.md

@@ -0,0 +1,4 @@
+Benchmarks
+==========
+
+TODO

etc/cachegrand.yaml.skel

@@ -41,6 +41,16 @@ network:
         - host: "::"
           port: 6379
 
+# Uncomment to enable prometheus support, more information available at
+# https://github.com/danielealbano/cachegrand/blob/main/docs/architecture/modules/prometheus.md
+#    - type: prometheus
+#      timeout:
+#        read_ms: -1
+#        write_ms: 10000
+#      bindings:
+#        - host: 127.0.0.1
+#          port: 9090
+
 database:
   # Currently the internal cachegrand hashtable doesn't support the automatic resizing, if there isn't enough room the
   # SET command will just fail with a generic error.

src/clock.c

@@ -20,6 +20,12 @@ void clock_monotonic(timespec_t *timespec) {
     }
 }
 
+void clock_realtime(timespec_t *timespec) {
+    if (clock_gettime(CLOCK_REALTIME, timespec) < 0) {
+        FATAL(TAG, "Unable to fetch the time");
+    }
+}
+
 void clock_diff(timespec_t *result, timespec_t *a, timespec_t *b) {
     result->tv_sec = a->tv_sec - b->tv_sec;
     result->tv_nsec = a->tv_nsec - b->tv_nsec;

src/clock.h

@@ -8,6 +8,7 @@ extern "C" {
 typedef struct timespec timespec_t;
 
 void clock_monotonic(timespec_t *timespec);
+void clock_realtime(timespec_t *timespec);
 void clock_diff(timespec_t *result, timespec_t *a, timespec_t *b);
 
 #ifdef __cplusplus

src/config.h

@@ -7,6 +7,7 @@ extern "C" {
 
 enum config_network_protocol_type {
     CONFIG_PROTOCOL_TYPE_REDIS,
+    CONFIG_PROTOCOL_TYPE_PROMETHEUS
 };
 typedef enum config_network_protocol_type config_network_protocol_type_t;
 

src/config_cyaml_schema.c

@@ -71,6 +71,7 @@ const cyaml_schema_field_t config_network_protocol_redis_schema[] = {
 // Allowed strings for for config -> network -> protocols -> protocol-> type (config_network_protocol_type_t)
 const cyaml_strval_t config_network_protocol_type_schema_strings[] = {
         { "redis", CONFIG_PROTOCOL_TYPE_REDIS },
+        { "prometheus", CONFIG_PROTOCOL_TYPE_PROMETHEUS },
 };
 
 typedef enum config_network_protocol_type config_network_protocol_type_t;
@@ -94,7 +95,7 @@ const cyaml_schema_field_t config_network_protocol_fields_schema[] = {
                 "keepalive", CYAML_FLAG_POINTER | CYAML_FLAG_OPTIONAL,
                 config_network_protocol_t, keepalive, config_network_protocol_keepalive_schema),
         CYAML_FIELD_MAPPING_PTR(
-                "redis", CYAML_FLAG_POINTER,
+                "redis", CYAML_FLAG_POINTER | CYAML_FLAG_OPTIONAL,
                 config_network_protocol_t, redis, config_network_protocol_redis_schema),
         CYAML_FIELD_SEQUENCE(
                 "bindings", CYAML_FLAG_POINTER,

src/network/network.c

@@ -143,8 +143,10 @@ network_op_result_t network_receive(
             channel->address.str);
 
     worker_stats_t *stats = worker_stats_get();
-    stats->network.per_second.received_packets++;
+    stats->network.per_minute.received_packets++;
     stats->network.total.received_packets++;
+    stats->network.per_minute.received_data += receive_length;
+    stats->network.total.received_data += receive_length;
 
     return NETWORK_OP_RESULT_OK;
 }
@@ -194,8 +196,10 @@ network_op_result_t network_send(
             channel->address.str);
 
     worker_stats_t *stats = worker_stats_get();
-    stats->network.per_second.sent_packets++;
+    stats->network.per_minute.sent_packets++;
     stats->network.total.sent_packets++;
+    stats->network.per_minute.sent_data += send_length;
+    stats->network.total.sent_data += send_length;
 
     return NETWORK_OP_RESULT_OK;
 }

src/network/protocol/network_protocol.h

@@ -8,6 +8,7 @@ extern "C" {
 enum network_protocols {
     NETWORK_PROTOCOLS_UNKNOWN = 0,
     NETWORK_PROTOCOLS_REDIS,
+    NETWORK_PROTOCOLS_PROMETHEUS
 };
 typedef enum network_protocols network_protocols_t;