Gather .health instance stats

Reworks instance checks to use the same timestamp and be transactional. Also prevents long DB lock times during list updates.

.env.example

@@ -52,3 +52,5 @@ CLEANUP_INTERVAL_S=86400
 ERROR_RETENTION_PER_HOST=100
 # session database URI
 SESSION_DB_URI="sqlite:./sessions.db?mode=rwc"
+# interval in seconds for gathering .health endpoint data
+STATS_INTERVAL_S=900

.gitignore

@@ -4,3 +4,4 @@
 /.env
 /sqlite.db-journal
 /out.txt
+/sessions.db

Cargo.lock

@@ -2244,7 +2244,7 @@ checksum = "e4a24736216ec316047a1fc4252e27dabb04218aa4a3f37c6e7ddbf1f9782b54"
 
 [[package]]
 name = "nitter-status"
-version = "0.4.0"
+version = "0.5.0"
 dependencies = [
  "chrono",
  "clap",
@@ -3229,6 +3229,7 @@ dependencies = [
  "sea-orm",
  "sea-query",
  "serde",
+ "serde_json",
  "thiserror",
  "tokio",
  "tracing",

Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "nitter-status"
-version = "0.4.0"
+version = "0.5.0"
 edition = "2021"
 publish = false
 license = "AGPL-3.0-only"
@@ -15,7 +15,7 @@ chrono = "0.4.22"
 serde = "1.0.144"
 sea-orm-migration = "0.12.0"
 sea-orm = "0.12.0"
-sea-query = "0.30" # keep in step sea-orm
+sea-query = "0.30" # keep in step with sea-orm
 futures = "0.3"
 tracing = "0.1"
 tracing-subscriber = "0.3"

entities/src/health_check.rs

@@ -46,6 +46,7 @@ pub struct HealthyAmount {
 }
 
 impl HealthyAmount {
+    /// Fetch health check graph data for all or selected hosts in the selected time range. 
     pub async fn fetch(db: &DatabaseConnection, from: DateTimeUtc, to: DateTimeUtc, hosts: Option<&[i32]>) -> Result<Vec<HealthyAmount>,DbErr> {
         let builder = db.get_database_backend();
         let mut stmt: sea_query::SelectStatement = Query::select();
@@ -90,6 +91,6 @@ mod test {
             .checked_sub_signed(chrono::Duration::days(14))
             .unwrap();
         let hosts = vec![1,2,3];
-        dbg!(HealthyAmount::fetch(&db, time_3h, time_now,Some(&hosts)).await.unwrap());
+        HealthyAmount::fetch(&db, time_3h, time_now,Some(&hosts)).await.unwrap();
     }
 }

entities/src/instance_stats.rs

@@ -0,0 +1,98 @@
+//! `SeaORM` Entity. Generated by sea-orm-codegen 0.11.3
+
+use sea_orm::{entity::prelude::*, FromQueryResult};
+use sea_query::{Query, SimpleExpr, Order, Alias};
+use serde::Serialize;
+
+#[derive(Clone, Debug, PartialEq, DeriveEntityModel, Eq, Serialize)]
+#[sea_orm(table_name = "instance_stats")]
+pub struct Model {
+    #[sea_orm(primary_key, auto_increment = false)]
+    pub time: i64,
+    #[sea_orm(primary_key, auto_increment = false)]
+    pub host: i32,
+    pub limited_accs: i32,
+    pub total_accs: i32,
+    pub total_requests: i64,
+    pub req_photo_rail: i32,
+    pub req_user_screen_name: i32,
+    pub req_search: i32,
+    pub req_list_tweets: i32,
+    pub req_user_media: i32,
+    pub req_tweet_detail: i32,
+    pub req_list: i32,
+    pub req_user_tweets: i32,
+    pub req_user_tweets_and_replies: i32,
+}
+
+#[derive(Copy, Clone, Debug, EnumIter, DeriveRelation)]
+pub enum Relation {
+    #[sea_orm(
+        belongs_to = "super::host::Entity",
+        from = "Column::Host",
+        to = "super::host::Column::Id",
+        on_update = "Cascade",
+        on_delete = "Cascade"
+    )]
+    Host,
+}
+
+impl Related<super::host::Entity> for Entity {
+    fn to() -> RelationDef {
+        Relation::Host.def()
+    }
+}
+
+impl ActiveModelBehavior for ActiveModel {}
+
+#[derive(Debug, FromQueryResult, Serialize)]
+pub struct StatsAmount {
+    pub time: i64,
+    pub limited_accs_max: i32,
+    pub limited_accs_avg: i32,
+    pub total_accs_max: i32,
+    pub total_accs_avg: i32,
+    pub total_requests_max: i64,
+    pub total_requests_avg: i64,
+    pub req_photo_rail_max: i32,
+    pub req_photo_rail_avg: i32,
+    pub req_user_screen_name_max: i32,
+    pub req_user_screen_name_avg: i32,
+    pub req_search_max: i32,
+    pub req_search_avg: i32,
+    pub req_list_tweets_max: i32,
+    pub req_list_tweets_avg: i32,
+    pub req_user_media_max: i32,
+    pub req_user_media_avg: i32,
+    pub req_tweet_detail_max: i32,
+    pub req_tweet_detail_avg: i32,
+    pub req_list_max: i32,
+    pub req_list_avg: i32,
+    pub req_user_tweets_max: i32,
+    pub req_user_tweets_avg: i32,
+    pub req_user_tweets_and_replies_max: i32,
+    pub req_user_tweets_and_replies_avg: i32,
+}
+
+impl StatsAmount {
+    /// Fetch health check graph data for all or selected hosts in the selected time range. 
+    pub async fn fetch(db: &DatabaseConnection, from: DateTimeUtc, to: DateTimeUtc, hosts: Option<&[i32]>) -> Result<Vec<StatsAmount>,DbErr> {
+        let builder = db.get_database_backend();
+        let columns = ["limited_accs","total_accs","total_requests","req_photo_rail","req_user_screen_name","req_search","req_list_tweets","req_user_media","req_tweet_detail","req_list","req_user_tweets","req_user_tweets_and_replies"];
+        let mut stmt: sea_query::SelectStatement = Query::select();
+        let col_stmt = stmt.column(self::Column::Time);
+        for col in columns {
+            col_stmt.expr_as(SimpleExpr::Custom(format!("MAX({col})")), Alias::new(format!("{col}_max")))
+            .expr_as(SimpleExpr::Custom(format!("CAST(ifnull(AVG({col}),0) as int)")), Alias::new(format!("{col}_avg")));
+        }
+        col_stmt.group_by_col(self::Column::Time)
+            .from(self::Entity)
+            .and_where(self::Column::Time.between(from.timestamp(), to.timestamp()));
+        if let Some(hosts)  = hosts {
+            stmt.and_where(self::Column::Host.is_in(hosts.iter().map(|v| *v)));
+        }
+        stmt.group_by_col(self::Column::Time)
+            .order_by(self::Column::Time, Order::Asc);
+        StatsAmount::find_by_statement(builder.build(&stmt)).all(db).await
+    }
+}

entities/src/mod.rs

@@ -5,6 +5,7 @@ pub mod prelude;
 pub mod check_errors;
 pub mod health_check;
 pub mod host;
+pub mod instance_stats;
 
 // has to be re-added on entity regeneration
 pub mod state;

entities/src/prelude.rs

@@ -3,3 +3,4 @@
 pub use super::check_errors::Entity as CheckErrors;
 pub use super::health_check::Entity as HealthCheck;
 pub use super::host::Entity as Host;
+pub use super::instance_stats::Entity as InstanceStats;

entities/src/state/scanner.rs

@@ -7,6 +7,8 @@ pub struct Config {
     pub list_fetch_interval: Duration,
     /// time until next instance ping check
     pub instance_check_interval: Duration,
+    /// time until next instance statistics check
+    pub instance_stats_interval: Duration,
     /// instances list URL
     pub instance_list_url: String,
     /// profile path for health check
@@ -48,8 +50,9 @@ pub struct Config {
 impl Config {
     pub fn test_defaults() -> ScannerConfig {
         Arc::new(Config {
-            list_fetch_interval: Duration::from_secs(60 * 5),
+            instance_stats_interval: Duration::from_secs(15 * 60),
-            instance_check_interval: Duration::from_secs(60 * 5),
+            list_fetch_interval: Duration::from_secs(15 * 60),
+            instance_check_interval: Duration::from_secs(15 * 60),
             instance_list_url: String::from("https://github.com/zedeus/nitter/wiki/Instances"),
             profile_path: String::from("/jack"),
             rss_path: String::from("/jack/rss"),

migration/src/lib.rs

@@ -7,6 +7,7 @@ mod m20230803_154714_version_url;
 mod m20230829_201916_country;
 mod m20230914_231514_connectivity;
 mod m20231011_231223_errors;
+mod m20231112_142206_stats;
 
 pub struct Migrator;
 
@@ -21,6 +22,7 @@ impl MigratorTrait for Migrator {
             Box::new(m20230829_201916_country::Migration),
             Box::new(m20230914_231514_connectivity::Migration),
             Box::new(m20231011_231223_errors::Migration),
+            Box::new(m20231112_142206_stats::Migration),
         ]
     }
 }

migration/src/m20231112_142206_stats.rs

@@ -0,0 +1,39 @@
+use sea_orm_migration::prelude::*;
+
+#[derive(DeriveMigrationName)]
+pub struct Migration;
+
+#[async_trait::async_trait]
+impl MigrationTrait for Migration {
+    async fn up(&self, manager: &SchemaManager) -> Result<(), DbErr> {
+        let cmd = r#"CREATE TABLE "instance_stats" (
+            "time" integer NOT NULL,
+            "host" integer NOT NULL,
+            "limited_accs" integer NOT NULL,
+            "total_accs" integer NOT NULL,
+            "total_requests" integer NOT NULL,
+            "req_photo_rail" integer NOT NULL,
+            "req_user_screen_name" integer NOT NULL,
+            "req_search" integer NOT NULL,
+            "req_list_tweets" integer NOT NULL,
+            "req_user_media" integer NOT NULL,
+            "req_tweet_detail" integer NOT NULL,
+            "req_list" integer NOT NULL,
+            "req_user_tweets" integer NOT NULL,
+            "req_user_tweets_and_replies" integer NOT NULL,
+            CONSTRAINT "pk_instance_stats" PRIMARY KEY ("host", "time"),
+            FOREIGN KEY ("host") REFERENCES "host" ("id") ON DELETE CASCADE ON UPDATE CASCADE
+        ) WITHOUT ROWID, STRICT;"#;
+        let db = manager.get_connection();
+        db.execute_unprepared("BEGIN EXCLUSIVE").await?;
+        tracing::info!("adding instance_stats table..");
+        db.execute_unprepared(cmd).await?;
+        db.execute_unprepared("COMMIT TRANSACTION").await?;
+        db.execute_unprepared("VACUUM").await?;
+        Ok(())
+    }
+
+    async fn down(&self, _manager: &SchemaManager) -> Result<(), DbErr> {
+        panic!("Can't migrate down");
+    }
+}

scanner/Cargo.toml

@@ -19,6 +19,8 @@ regex = { workspace = true }
 git2 = "0.17.2"
 # testing
 serde = { workspace = true, features = ["derive"] }
+# health parsing
+serde_json = "1"
 
 [dev-dependencies]
 tracing-test = { workspace = true }

scanner/src/instance_check.rs

@@ -40,6 +40,7 @@ impl Scanner {
                 .map_or(false, |check| !check.healthy);
             join_set.spawn(async move {
                 scanner.health_check_host(model, muted_host).await;
+                
             });
         }
         // wait till all of them are finished, preventing DoS

scanner/src/lib.rs

@@ -35,6 +35,7 @@ mod instance_parser;
 mod list_update;
 mod profile_parser;
 mod version_check;
+mod update_stats;
 
 const CAPTCHA_TEXT: &'static str = "Enable JavaScript and cookies to continue";
 const CAPTCHA_CODE: u16 = 403;
@@ -62,8 +63,18 @@ pub enum ScannerError {
     FetchError(#[from] FetchError),
     #[error("Failed fetching git: {0}")]
     GitFetch(#[from] git2::Error),
-    #[error("Couldn't find git branch")]
+    /// Configured git branch can't be found
-    GitBranch,
+    #[error("Couldn't find configured git branch")]
+    GitBranchNotFound,
+    /// Failed to parse .health endpoint
+    #[error("Failed to parse .health endpoint. {0} Body: {1}")]
+    StatsParsing(serde_json::Error, String),
+    /// Generic error for parsing an instance URL to [reqwest::Url]
+    #[error("Can't parse instance URL")]
+    InstanceUrlParse,
+    /// Failed to join a tokio thread in a JoinSet
+    #[error("Failed to join worker task {0}")]
+    JoinError(#[from] tokio::task::JoinError)
 }
 
 #[derive(Error, Debug)]
@@ -154,6 +165,7 @@ struct InnerScanner {
     about_parser: AboutParser,
     profile_parser: ProfileParser,
     last_list_fetch: Mutex<DateTime<Utc>>,
+    last_stats_fetch: Mutex<DateTime<Utc>>,
     last_uptime_check: Mutex<DateTime<Utc>>,
     rss_check_regex: Regex,
     client_ipv4: Client,
@@ -197,11 +209,16 @@ impl Scanner {
             .build()
             .into_diagnostic()?;
 
-        let last_uptime_check = Self::query_last_fetch(&db)
+        let last_uptime_check = Self::query_last_entry_for_table(&db,"health_check")
             .await
             .into_diagnostic()
             .wrap_err("Fetching last uptime check failed!")?;
-        tracing::info!(?last_uptime_check);
+
+        let last_stats_check = Self::query_last_entry_for_table(&db,"instance_stats")
+            .await
+            .into_diagnostic()
+            .wrap_err("Fetching last stats check failed!")?;
+        tracing::info!(?last_uptime_check, ?last_stats_check);// TODO: wrong timestamp
         let scanner = Self {
             inner: Arc::new(InnerScanner {
                 db,
@@ -215,6 +232,7 @@ impl Scanner {
                 profile_parser: ProfileParser::new(),
                 last_list_fetch: Mutex::new(last_uptime_check),
                 last_uptime_check: Mutex::new(last_uptime_check),
+                last_stats_fetch: Mutex::new(last_stats_check),
                 rss_check_regex: builder_regex_rss
                     .build()
                     .into_diagnostic()
@@ -229,8 +247,8 @@ impl Scanner {
         Ok(scanner)
     }
 
-    /// Retrieves the last uptime fetch that happened
+    /// Retrieves the last stats fetching that happened
-    pub async fn query_last_fetch(db: &DatabaseConnection) -> Result<DateTime<Utc>> {
+    pub async fn query_last_entry_for_table(db: &DatabaseConnection, table: &str) -> Result<DateTime<Utc>> {
         #[derive(Debug, FromQueryResult, Default)]
         pub(crate) struct TimeMax {
             pub max: Option<i64>,
@@ -239,7 +257,7 @@ impl Scanner {
         // will crash on an empty DB
         let time_max = TimeMax::find_by_statement(Statement::from_sql_and_values(
             DbBackend::Sqlite,
-            r#"SELECT MAX(time) FROM health_check"#,
+            format!("SELECT MAX(time) as max FROM {table}"),
             [],
         ))
         .one(db)
@@ -282,23 +300,28 @@ impl Scanner {
             }
             if self.is_instance_check_outdated() {
                 if let Err(e) = self.check_uptime().await {
-                    tracing::error!(error=?e,"Failed checking instance");
+                    tracing::error!(error=?e,"Failed checking instances for health");
+                }
+            }
+            if self.is_instance_stats_outdated() {
+                if let Err(e) = self.check_health().await {
+                    tracing::error!(error=?e,"Failed checking instances for stats");
                 }
             }
             if let Err(e) = self.update_cache().await {
-                tracing::error!(error=?e,"Failed updating cache!");
+                tracing::error!(error=?e,"Failed to update cache!");
             }
             self.sleep_till_deadline().await;
         }
     }
 
     async fn sleep_till_deadline(&self) {
-        let delay_instance_check =
+        let delay_instance_check = self.last_uptime_check() + self.inner.config.instance_check_interval;
-            self.last_uptime_check() + self.inner.config.instance_check_interval;
-
         let delay_list_update = self.last_list_fetch() + self.inner.config.list_fetch_interval;
-        tracing::debug!(?delay_list_update, ?delay_instance_check);
+        let delay_stats_update = self.last_stats_fetch() + self.inner.config.instance_stats_interval;
-        let next_deadline = delay_instance_check.min(delay_list_update);
+        tracing::debug!(?delay_list_update, ?delay_instance_check, ?delay_stats_update);
+
+        let next_deadline = delay_instance_check.min(delay_list_update).min(delay_stats_update);
         let now = Utc::now();
         let sleep_time = next_deadline.signed_duration_since(now);
         if sleep_time <= Duration::zero() {
@@ -318,6 +341,10 @@ impl Scanner {
         *self.inner.last_list_fetch.lock().unwrap()
     }
 
+    fn last_stats_fetch(&self) -> DateTime<Utc> {
+        *self.inner.last_stats_fetch.lock().unwrap()
+    }
+
     fn is_instance_check_outdated(&self) -> bool {
         let val = self.last_uptime_check();
         Utc::now().signed_duration_since(val).to_std().unwrap()
@@ -330,6 +357,12 @@ impl Scanner {
             >= self.inner.config.list_fetch_interval
     }
 
+    fn is_instance_stats_outdated(&self) -> bool {
+        let val = self.last_stats_fetch();
+        Utc::now().signed_duration_since(val).to_std().unwrap()
+            >= self.inner.config.instance_stats_interval
+    }
+
     async fn fetch_instance_list(&self) -> Result<String> {
         let (_, body) = self.fetch_url(&self.inner.config.instance_list_url).await?;
         Ok(body)

scanner/src/list_update.rs

@@ -59,6 +59,10 @@ impl Scanner {
         let found_instances: usize = parsed_instances.len();
         // find last update checks to detect spam
         let last_status = self.query_latest_check(&transaction).await?;
+        // prevent blocking the DB forever during host updates
+        // worst case we disable hosts that should be disabled and add no new ones
+        transaction.commit().await?;
+
         let mut join_set = JoinSet::new();
         for (_, instance) in parsed_instances {
             // TODO: parallelize this!
@@ -112,25 +116,30 @@ impl Scanner {
                 }
             });
         }
+
+        let mut update_models = Vec::with_capacity(join_set.len());
         while let Some(update_model) = join_set.join_next().await.map(|v| v.unwrap()) {
-            Host::insert(update_model)
+            update_models.push(update_model);
-                .on_conflict(
-                    OnConflict::column(host::Column::Domain)
-                        .update_columns([
-                            host::Column::Enabled,
-                            host::Column::Updated,
-                            host::Column::Url,
-                            host::Column::Rss,
-                            host::Column::Version,
-                            host::Column::VersionUrl,
-                            host::Column::Country,
-                            host::Column::Connectivity,
-                        ])
-                        .to_owned(),
-                )
-                .exec(&transaction)
-                .await?;
         }
+        // insert all at once
+        let transaction = self.inner.db.begin().await?;
+        Host::insert_many(update_models)
+        .on_conflict(
+            OnConflict::column(host::Column::Domain)
+                .update_columns([
+                    host::Column::Enabled,
+                    host::Column::Updated,
+                    host::Column::Url,
+                    host::Column::Rss,
+                    host::Column::Version,
+                    host::Column::VersionUrl,
+                    host::Column::Country,
+                    host::Column::Connectivity,
+                ])
+                .to_owned(),
+        )
+        .exec(&transaction)
+        .await?;
 
         transaction.commit().await?;
         let end = Instant::now();

scanner/src/update_stats.rs

@@ -0,0 +1,126 @@
+// SPDX-License-Identifier: AGPL-3.0-only
+
+//! for .health endpoint statistics gathering
+
+use std::time::Instant;
+
+use chrono::Utc;
+use entities::{host, instance_stats};
+use entities::prelude::Host;
+use reqwest::Url;
+use sea_orm::prelude::DateTimeUtc;
+use sea_orm::{QueryFilter, ColumnTrait, EntityTrait, ActiveValue};
+use serde::Deserialize;
+use tokio::task::JoinSet;
+
+use crate::{Result, Scanner, ScannerError};
+
+/// Instance stats reported by .health
+#[derive(Debug, Deserialize)]
+struct InstanceStats {
+    accounts: InstanceStatsAccs,
+    requests: RequestStats,
+}
+
+#[derive(Debug, Deserialize)]
+struct InstanceStatsAccs {
+    total: i32,
+    limited: i32,
+    oldest: DateTimeUtc,
+    newest: DateTimeUtc,
+    average: DateTimeUtc,
+}
+
+#[derive(Debug, Deserialize)]
+struct RequestStats {
+    total: i64,
+    apis: APIStats,
+}
+
+/// Instance api stats reported by .health
+#[derive(Debug, Deserialize)]
+#[allow(non_snake_case)]
+struct APIStats {
+    pub photoRail: i32,
+    pub userScreenName: i32,
+    pub search: i32,
+    pub listTweets: i32,
+    pub userMedia: i32,
+    pub tweetDetail: i32,
+    pub list: i32,
+    pub userTweets: i32,
+    pub userTweetsAndReplies: i32,
+}
+
+impl Scanner {
+    pub(crate) async fn check_health(&self) -> Result<()> {
+        let hosts = Host::find()
+            .filter(host::Column::Enabled.eq(true))
+            .all(&self.inner.db)
+            .await?;
+        let start = Instant::now();
+
+        let mut join_set = JoinSet::new();
+        let time = Utc::now();
+        for model in hosts.into_iter() {
+            let scanner = self.clone();
+            join_set.spawn(async move {
+                let res = scanner.fetch_instance_stats(time, &model).await;
+                if let Err(e) = &res {
+                    tracing::debug!(host=model.id, error=?e,"Failed to fetch instance stats");
+                }
+                res.ok()
+            });
+        }
+
+        let mut stat_data = Vec::with_capacity(join_set.len());
+        while let Some(join_res) = join_set.join_next().await {
+            if let Some(data) = join_res? {
+                stat_data.push(data);
+            }
+        }
+        tracing::debug!(db_stats_entries=stat_data.len());
+        if !stat_data.is_empty() {
+            instance_stats::Entity::insert_many(stat_data)
+                .exec(&self.inner.db).await?;
+        }
+
+        let end = Instant::now();
+        let duration = end - start;
+        {
+            *self.inner.last_stats_fetch.lock().unwrap() = Utc::now();
+        }
+        tracing::debug!(duration=?duration,"stats check finished");
+        Ok(())
+    }
+
+    async fn fetch_instance_stats(&self, time: DateTimeUtc, host: &host::Model) -> Result<instance_stats::ActiveModel> {
+        let mut url = Url::parse(&host.url)
+            .map_err(|e|ScannerError::InstanceUrlParse)?;
+        url.set_path(".health");
+        let (_code, body) = self.fetch_url(url.as_str()).await?;
+
+        let stats_data: InstanceStats = 
+        serde_json::from_str(&body)
+            .map_err(|e|ScannerError::StatsParsing(e,body))?;
+
+        let stats_model = instance_stats::ActiveModel {
+            time: ActiveValue::Set(time.timestamp()),
+            host: ActiveValue::Set(host.id),
+            limited_accs: ActiveValue::Set(stats_data.accounts.limited),
+            total_accs: ActiveValue::Set(stats_data.accounts.total),
+            total_requests: ActiveValue::Set(stats_data.requests.total),
+            req_photo_rail: ActiveValue::Set(stats_data.requests.apis.photoRail),
+            req_user_screen_name: ActiveValue::Set(stats_data.requests.apis.userScreenName),
+            req_search: ActiveValue::Set(stats_data.requests.apis.search),
+            req_list_tweets: ActiveValue::Set(stats_data.requests.apis.listTweets),
+            req_user_media: ActiveValue::Set(stats_data.requests.apis.userMedia),
+            req_tweet_detail: ActiveValue::Set(stats_data.requests.apis.tweetDetail),
+            req_list: ActiveValue::Set(stats_data.requests.apis.list),
+            req_user_tweets: ActiveValue::Set(stats_data.requests.apis.userTweets),
+            req_user_tweets_and_replies: ActiveValue::Set(stats_data.requests.apis.userTweetsAndReplies),
+        };
+
+        Ok(stats_model)
+    }
+}

scanner/src/version_check.rs

@@ -56,7 +56,7 @@ pub(crate) fn fetch_git_state(config: ScannerConfig) -> Result<CurrentVersion> {
             version: commit,
             config,
         })
-        .ok_or(ScannerError::GitBranch)?)
+        .ok_or(ScannerError::GitBranchNotFound)?)
 }
 
 #[cfg(test)]

server/src/admin.rs

@@ -14,6 +14,7 @@ use constant_time_eq::constant_time_eq;
 use entities::check_errors;
 use entities::health_check;
 use entities::host;
+use entities::instance_stats;
 use entities::state::AppState;
 use hyper::header::REFERER;
 use hyper::HeaderMap;
@@ -319,15 +320,18 @@ pub async fn history_json(
     struct ReturnData {
         pub global: Vec<health_check::HealthyAmount>,
         pub user: Vec<health_check::HealthyAmount>,
+        pub stats: Vec<instance_stats::StatsAmount>,
     }
     let (_login, hosts) = get_all_login_hosts(&session, db, false).await?;
     let host_ids: Vec<_> = hosts.iter().map(|host|host.id).collect();
     let data_owned = health_check::HealthyAmount::fetch(db, input.start, input.end,Some(&host_ids)).await?;
     let data_global = health_check::HealthyAmount::fetch(db, input.start, input.end,None).await?;
+    let data_stats = instance_stats::StatsAmount::fetch(db, input.start, input.end,None).await?;
 
     Ok(Json(ReturnData {
         global: data_global,
         user: data_owned,
+        stats: data_stats,
     }).into_response())
 }
 
@@ -338,15 +342,29 @@ pub async fn history_json_specific(
     Json(input): Json<DateRangeInput>,
 ) -> Result<axum::response::Response> {
     let host = get_specific_login_host(host, &session, db).await?;
+    #[derive(Debug, Serialize)]
+    struct ReturnData {
+        pub stats: Vec<instance_stats::Model>,
+        pub health: Vec<health_check::Model>,
+    }
 
-    let history: Vec<health_check::Model> = health_check::Entity::find()
+    let history_health: Vec<health_check::Model> = health_check::Entity::find()
         .filter(health_check::Column::Host.eq(host.id))
         .order_by_asc(health_check::Column::Time)
         .filter(health_check::Column::Time.between(input.start.timestamp(), input.end.timestamp()))
         .all(db)
         .await?;
+    let history_stats: Vec<instance_stats::Model> = instance_stats::Entity::find()
+        .filter(instance_stats::Column::Host.eq(host.id))
+        .order_by_asc(instance_stats::Column::Time)
+        .filter(instance_stats::Column::Time.between(input.start.timestamp(), input.end.timestamp()))
+        .all(db)
+        .await?;
 
-    Ok(Json(history).into_response())
+    Ok(Json(ReturnData{
+        health: history_health,
+        stats: history_stats
+    }).into_response())
 }
 
 pub async fn history_view(

server/static/admin.js

@@ -39,73 +39,209 @@ async function fetchDataAndCreateChart(startDate,endDate) {
     } catch (error) {
         console.error('Failed to fetch data:', error);
     }
+    // Create arrays for y-values for both datasets
+    const healthyData = data.global.map(entry => {
+        return {x: moment.unix(entry.time), y: entry.alive};
+    });
+
+    const unhealthyData = data.global.map(entry => {
+        return {x: moment.unix(entry.time), y: entry.dead};
+    });
+
+    let statsData = [];
+    let statsKeys = [];
+    if (data.stats.length > 0) {
+        
+        statsKeys = Object.keys(data.stats[0]).filter(key => key !== 'time');
+        statsData = statsKeys.map(key => ({"key": key, data: data.stats.map(entry => ({x: moment.unix(entry.time), y: entry[key]}))}));
+    }
+    
 
-    // Extracting data for chart
-    const timestamps = data.global.map(entry => entry.time * 1000); // Converting UNIX timestamps to milliseconds
-    const healthyData = data.global.map(entry => entry.alive);
-    const unhealthyData = data.global.map(entry => entry.dead);
     const userUnhealthyData = data.user.filter(entry => entry.dead > 0).map(entry => ({"x": entry.time * 1000, "y": entry.dead}));
-    const paramData = {"global": {"healthyData":  healthyData, "unhealthyData": unhealthyData}, "user": userUnhealthyData};
+    const paramData = {"healthyData":  healthyData, "unhealthyData": unhealthyData, "user": userUnhealthyData, "stats": statsData, "statsKeys": statsKeys};
-    console.log(paramData);
-    createChart(timestamps, paramData);
     
+    createChartOverview(paramData);    
 }
 
-function createChart(timestamps, data) {
+function mapKey(key) {
-    const ctx = document.getElementById('graph').getContext('2d');
+    const keyMappings = {
-    let high_data = timestamps.length > 1000;
+        "limited_accs_max": "Max Lim. Accs",
+        "limited_accs_avg": "Avg Lim. Accs",
+        "total_accs_max": "Max Accs",
+        "total_accs_avg": "Avg Accs",
+        "total_requests_max": "Max Requests Req.",
+        "total_requests_avg": "Avg Requests Req.",
+        "req_photo_rail_max": "Max PhotoRail Req.",
+        "req_photo_rail_avg": "Avg PhotoRail Req.",
+        "req_user_screen_name_max": "Max UserScreenName Req.",
+        "req_user_screen_name_avg": "Avg UserScreenName Req.",
+        "req_search_max": "Max Search Req.",
+        "req_search_avg": "Avg Search Req.",
+        "req_list_tweets_max": "Max ListTweets Req.",
+        "req_list_tweets_avg": "Avg ListTweets Req.",
+        "req_user_media_max": "Max UserMedia Req.",
+        "req_user_media_avg": "Avg UserMedia Req.",
+        "req_tweet_detail_max": "Max TweetDetail Req.",
+        "req_tweet_detail_avg": "Avg TweetDetail Req.",
+        "req_list_max": "Max List Req.",
+        "req_list_avg": "Avg List Req.",
+        "req_user_tweets_max": "Max UserTweets Req.",
+        "req_user_tweets_avg": "Avg UserTweets Req.",
+        "req_user_tweets_and_replies_max": "Max UserTweetsAndReplies Req.",
+        "req_user_tweets_and_replies_avg": "Avg UserTweetsAndReplies Req.",
+    }
+    return keyMappings[key];
+}
+
+function nameToColor(name) {
+    const colors = ["#fafa6e"
+    ,"#eafb71"
+    ,"#dafb75"
+    ,"#cafb7b"
+    ,"#b9fa81"
+    ,"#a8f989"
+    ,"#97f890"
+    ,"#85f799"
+    ,"#71f5a1"
+    ,"#5cf3aa"
+    ,"#42f1b3"
+    ,"#15efbc"
+    ,"#00ecc4"
+    ,"#00e9cd"
+    ,"#00e6d5"
+    ,"#00e3dd"
+    ,"#00e0e4"
+    ,"#00dcea"
+    ,"#00d8f0"
+    ,"#00d5f5"
+    ,"#00d0f9"
+    ,"#00ccfd"
+    ,"#00c8ff"
+    ,"#00c3ff"
+    ,"#00beff"
+    ,"#00b9ff"
+    ,"#00b4ff"];
+    var hash = hashStr(name);
+    var index = hash % colors.length;
+    return colors[index];
+}
+
+function hashStr(str) {
+    var hash = 0;
+    for (var i = 0; i < str.length; i++) {
+        var charCode = str.charCodeAt(i);
+        hash += charCode;
+    }
+    return hash;
+}
+
+function createChartOverview(data) {
+    const ctx = document.getElementById('graph-health').getContext('2d');
+    let high_data = data.stats.length > 1000;
     if (chart) {
         chart.destroy();
     }
+    let datasets = [{
+        label: 'Healthy',
+        data: data.healthyData,
+        borderColor: 'green',
+        backgroundColor: 'rgba(0, 255, 0, 0.2)',
+        fill: true
+    },
+    {
+        label: 'Unhealthy',
+        data: data.unhealthyData,
+        borderColor: 'orange',
+        backgroundColor: 'rgba(241, 90, 34, 0.2)',
+        fill: true
+    },
+    {
+        label: 'Own Instances Unhealthy',
+        data: data.user,
+        pointRadius: 5,
+        pointBackgroundColor: 'rgba(255, 0, 0, 1)',
+        fill: false,
+        datasetIndex: "x",
+    }];
+    datasets = datasets.concat(data.stats.map(entry => ({
+        yAxisID: 'yStats',
+        label: mapKey(entry.key),
+        data: entry.data,
+        borderColor: nameToColor(entry.key),
+        fill: false,
+        hidden: (entry.key !== 'total_requests_avg' && entry.key !== 'limited_accs_avg'),
+    })));
+    console.log(datasets);
     chart = new Chart(ctx, {
         type: 'line',
         data: {
-            labels: timestamps,
+            datasets: datasets,
-            datasets: [
-                {
-                    label: 'Healthy',
-                    data: data.global.healthyData,
-                    borderColor: 'green',
-                    backgroundColor: 'rgba(0, 255, 0, 0.2)',
-                    fill: true
-                },
-                {
-                    label: 'Unhealthy',
-                    data: data.global.unhealthyData,
-                    borderColor: 'orange',
-                    backgroundColor: 'rgba(241, 90, 34, 0.2)',
-                    fill: true
-                },
-                {
-                    label: 'Own Instances Unhealthy',
-                    data: data.user,
-                    pointRadius: 5,
-                    pointBackgroundColor: 'rgba(255, 0, 0, 1)',
-                    fill: false,
-                    datasetIndex: "x",
-                }
-            ]
         },
         options: {
+            responsive: false,
+            maintainAspectRatio: false,
             plugins: {
+                colors: {
+                    enabled: true
+                  },
                 decimation: {
                     enabled: high_data,
                     algorithm: 'min-max',
                 },
+                tooltip: {
+                    // callbacks: {
+                    //     label: function(context) {
+                    //         console.log(context);
+                    //         if (context.dataset) {
+                    //             if (context.dataset.yAxisID == 'yStats') {
+                    //                 let dataset = context.dataset.data[context.dataIndex];
+                    //                 if (!dataset) {
+                    //                     console.warn(context.dataset);
+                    //                     console.warn(context.dataset.data[context.dataIndex]);
+                    //                 }
+                    //                 let mainValue = dataset.y;
+                    //                 let detailValue = dataset.additional.total_requests_max;
+                    //                 return [`${context.dataset.label}: ${mainValue}`,`Max Total Requests: ${detailValue}`];
+                    //             } else {
+                    //                 return `${context.dataset.label}: ${context.formattedValue}`;
+                    //             }
+                    //         } else {
+                    //             console.log(context);
+                    //             return null;
+                    //         }
+                    //         // let label = context.dataset.label || '';
+    
+                    //         // if (label) {
+                    //         //     label += ': ';
+                    //         // }
+                    //         // if (context.parsed.y !== null) {
+                    //         //     label += new Intl.NumberFormat('en-US', { style: 'currency', currency: 'USD' }).format(context.parsed.y);
+                    //         // }
+                    //         return 'asdf';
+                    //     }
+                    // }
+                }
             },
             animation: !high_data,
             scales: {
                 x: {
                     type: 'time',
                     time: {
-                        unit: 'day' // You might want to adjust this based on your data resolution
+                        unit: 'day'
                     },
                     autoSkip: true,
                 },
                 y: {
                     beginAtZero: true
+                },
+                yStats: {
+                    beginAtZero: false
                 }
-            }
+            },
+            interaction: {
+                mode: 'index',
+                intersect: false
+            },
         }
     });
 }

server/static/instance.js

@@ -25,6 +25,24 @@ document.addEventListener("DOMContentLoaded", function() {
     fetchDataAndCreateChart(initialStartDate,initialEndDate);
 });
 
+function mapKey(key) {
+    const keyMappings = {
+        "limited_accs": "Limited Accs",
+        "total_accs": "Total Accs",
+        "total_requests": "Total Requests",
+        "req_photo_rail": "PhotoRail Requests",
+        "req_user_screen_name": "UserScreeName Req.",
+        "req_search": "Search Requests",
+        "req_list_tweets": "ListTeweets Req.",
+        "req_user_media": "UserMedia Req.",
+        "req_tweet_detail": "TweetDetail Req.",
+        "req_list": "List Requests",
+        "req_user_tweets": "UserTweets Req.",
+        "req_user_tweets_and_replies": "UserTweetsAndReplies Req.",
+    }
+    return keyMappings[key];
+}
+
 async function fetchDataAndCreateChart(startDate,endDate) {
     let jsonData;
     try {
@@ -41,15 +59,14 @@ async function fetchDataAndCreateChart(startDate,endDate) {
         console.error('Failed to fetch data:', error);
     }
     if (jsonData) {
-        createChart(jsonData);
+        createChartOverview(jsonData);
     }
 }
 
-function createChart(jsonData) {
+function createChartOverview(jsonData) {
-    const timeLabels = jsonData.map(dataPoint => new Date(dataPoint.time * 1000));
+    const healthyData = jsonData.health.filter(entry => entry.healthy).map(entry => ({x: moment.unix(entry.time), y: entry.resp_time}));
-    console.log(timeLabels);
+    const unhealthyData = jsonData.health.filter(entry => !entry.healthy).map(entry => ({x: moment.unix(entry.time), y: entry.resp_time}));
-    const healthyData = jsonData.map(dataPoint => dataPoint.healthy ? (dataPoint.resp_time ? dataPoint.resp_time : 0) : null);
+    const statsData = jsonData.stats.map(entry => ({x: moment.unix(entry.time), y: entry.total_requests}));
-    const unhealthyData = jsonData.map(dataPoint => dataPoint.healthy ? null : (dataPoint.resp_time ? dataPoint.resp_time : 0));
 
     const ctx = document.getElementById('graph').getContext('2d');
     if (chart) {
@@ -58,23 +75,33 @@ function createChart(jsonData) {
     chart = new Chart(ctx, {
         type: 'line',
         data: {
-            labels: timeLabels,
             datasets: [
                 {
-                    label: 'Healthy',
+                    label: 'Healthy Response Time (ms)',
                     data: healthyData,
                     borderColor: 'green',
                     fill: false,
                 },
                 {
-                    label: 'Unhealthy',
+                    label: 'Unhealthy Response Time (ms)',
                     data: unhealthyData,
                     borderColor: 'red',
                     fill: false,
                 },
+                {
+                    label: 'Total API Requests',
+                    data: statsData,
+                    borderColor: 'blue',
+                    fill: false,
+                    yAxisID: 'yStats',
+                },
             ]
         },
         options: {
+            interaction: {
+                mode: 'index',
+                intersect: false
+            },
             scales: {
                 x: {
                     type: 'time',
@@ -91,6 +118,12 @@ function createChart(jsonData) {
                         display: true,
                         labelString: 'Response Time'
                     }
+                },
+                yStats: {
+                    scaleLabel: {
+                        display: true,
+                        labelString: 'Requests'
+                    }
                 }
             }
         }

server/templates/admin.html.j2

@@ -16,7 +16,7 @@
     
     <h4>Instances</h4>
     {% if is_admin %}<p class="text-danger-emphasis">root mode</p>{% endif %}
-    <div><canvas id="graph" width="600" height="200"></canvas><noscript>History feature works only with javascript enabled.</noscript></div>
+    <div class="table-responsive"><canvas id="graph-health" class="table" height="400"></canvas><noscript>History feature works only with javascript enabled.</noscript></div>
     <div class="container mt-5">
       <form>
           <div class="row">

src/main.rs

@@ -135,8 +135,12 @@ fn read_scanner_cfg() -> miette::Result<ScannerConfig> {
     let error_retention_per_host: usize = require_env_str("ERROR_RETENTION_PER_HOST")?
         .parse()
         .expect("CLEANUP_INTERVAL_S must be a number");
+    let instance_stats_interval: u64 = require_env_str("STATS_INTERVAL_S")?
+        .parse()
+        .expect("STATS_INTERVAL_S must be a positive number");
 
     Ok(Arc::new(entities::state::scanner::Config {
+        instance_stats_interval: Duration::from_secs(instance_stats_interval),
         list_fetch_interval: Duration::from_secs(instance_list_interval),
         instance_check_interval: Duration::from_secs(instance_ping_interval),
         instance_list_url: nitter_instancelist,