diff options
| author | Alexander_Kabui <alexanderkabui@proton.me> | 2026-10-13 23:36:39 +0300 |
|---|---|---|
| committer | Alexander_Kabui <alexanderkabui@proton.me> | 2026-10-14 07:28:51 +0300 |
| commit | 22ff53e2d65e47df27d76aa7420a146074c00282 (patch) | |
| tree | 07f550107ebb9501b1fa79e8e8d6b2d48210572d /src | |
| parent | a188b75157516c5e00dd130d9ca8dfbd6cf36679 (diff) | |
| download | genenetwork4-22ff53e2d65e47df27d76aa7420a146074c00282.tar.gz | |
config: GN4_DB_MODE (mariadb optional on dev) + no-DB hardening
* src/config.rs (DbMode, Config.db_mode, parse_db_mode): add
GN4_DB_MODE=dev|production in the layered config; dev runs with no
configured DB; fail-closed parse (a typo refuses).
* src/db.rs (DbError, pool, probe), src/error.rs (ApiError::from_db):
mint a typed 503 db_unavailable at the db::pool() choke point,
distinguishing Unavailable from Query.
* src/main.rs: loud no-DB startup line; production refuses to start
(exit non-zero) before axum::serve.
* src/routes/{accessions,datasets,groups,mapping_methods,publications,species,traits}.rs:
route every DB error through ApiError::from_db.
* src/bin/gn-indexer.rs: refuse to build an index when the DB is unreachable.
* tests/no_db.rs: pin no-DB degradation (search 200 partial, DB routes 503).
His ask (DM 2026-10-12): "Make it an option to switch it off for
development machines? I mean mariadb." The no-DB hardening is ours
(review-driven). Default GN4_DB_MODE left unflipped (=dev); the flip needs
owner sign-off.
Diffstat (limited to 'src')
| -rw-r--r-- | src/bin/gn-indexer.rs | 14 | ||||
| -rw-r--r-- | src/config.rs | 58 | ||||
| -rw-r--r-- | src/db.rs | 126 | ||||
| -rw-r--r-- | src/error.rs | 26 | ||||
| -rw-r--r-- | src/lib.rs | 2 | ||||
| -rw-r--r-- | src/main.rs | 25 | ||||
| -rw-r--r-- | src/routes/accessions.rs | 11 | ||||
| -rw-r--r-- | src/routes/datasets.rs | 50 | ||||
| -rw-r--r-- | src/routes/groups.rs | 92 | ||||
| -rw-r--r-- | src/routes/mapping_methods.rs | 11 | ||||
| -rw-r--r-- | src/routes/publications.rs | 17 | ||||
| -rw-r--r-- | src/routes/species.rs | 33 | ||||
| -rw-r--r-- | src/routes/traits.rs | 35 |
13 files changed, 270 insertions, 230 deletions
diff --git a/src/bin/gn-indexer.rs b/src/bin/gn-indexer.rs index 3b70945..d437699 100644 --- a/src/bin/gn-indexer.rs +++ b/src/bin/gn-indexer.rs @@ -67,6 +67,20 @@ fn main() -> ExitCode { publish_filter, }; + // Fail loudly BEFORE touching the index when the database is + // unreachable: the crawl needs it, and a run must never build an + // empty index. The message names SQL_URI/GN4_DB_MODE so the cause is + // unambiguous. (The crawl itself also fails, but this makes the + // reason explicit and keeps the index untouched by construction.) + if let Err(error) = genenetwork4::db::probe() { + eprintln!( + "gn-indexer: database unreachable (SQL_URI; GN4_DB_MODE={:?}); refusing to build an \ + index: {error}", + config.db_mode + ); + return ExitCode::from(1); + } + let result = match mode.as_str() { "full" => run_full(&run_config), "incremental" => run_incremental(&run_config), diff --git a/src/config.rs b/src/config.rs index 5030576..5ae2913 100644 --- a/src/config.rs +++ b/src/config.rs @@ -17,6 +17,12 @@ pub struct Config { /// mysql://user:password@host/database. Same SQL_URI config key /// GeneNetwork3 uses, so the two services can share configuration. pub sql_uri: String, + /// Whether the database is required (`production`) or optional + /// (`dev`). Set via the GN4_DB_MODE key. `production` (the default) + /// refuses to start when the database is unreachable; `dev` starts + /// loudly without it, with DB-backed routes disabled. See + /// `DbMode` and src/main.rs. + pub db_mode: DbMode, /// Base URL of the GeneNetwork3 API, used by proxy endpoints. /// Must end with a slash, e.g. http://127.0.0.1:5003/api/. pub gn3_url: String, @@ -89,6 +95,12 @@ impl Config { "SQL_URI".to_string(), "mysql://webqtlout:webqtlout@127.0.0.1/db_webqtl".to_string(), ); + // Preserve today's behaviour: with no GN4_DB_MODE set, gn4 still + // starts when the database is unreachable (it just logs loudly + // and answers DB-backed routes as 503). Flipping this default to + // `production` (refuse to start) is a consumer-facing behaviour + // change and needs owner sign-off -- see REPORT.md. + values.insert("GN4_DB_MODE".to_string(), "dev".to_string()); values.insert("GENENETWORK4_PORT".to_string(), "3000".to_string()); values.insert( "AUTH_SERVER_URL".to_string(), @@ -151,6 +163,7 @@ impl Config { // Layer 4: environment wins over everything. for key in [ "SQL_URI", + "GN4_DB_MODE", "GENENETWORK4_PORT", "AUTH_SERVER_URL", "GENENETWORK4_BIND", @@ -177,6 +190,7 @@ impl Config { Config { sql_uri: values["SQL_URI"].clone(), + db_mode: parse_db_mode(&values["GN4_DB_MODE"]), gn3_url, gn3_timeout_secs: values["GN3_TIMEOUT_SECS"] .parse() @@ -201,6 +215,31 @@ impl Config { } } +/// Reads the GN4_DB_MODE setting, failing closed: only the exact +/// values `dev` and `production` (any case, trimmed) are accepted; +/// anything else -- including a typo like "prodution" -- refuses to +/// start rather than silently picking a mode. The empty string (an +/// explicitly blank key) is likewise refused. +fn parse_db_mode(value: &str) -> DbMode { + match value.trim().to_ascii_lowercase().as_str() { + "dev" => DbMode::Dev, + "production" => DbMode::Production, + other => panic!("GN4_DB_MODE must be \"dev\" or \"production\", not {other:?}"), + } +} + +/// Whether the database is required for the process to serve. +#[derive(Clone, Copy, PartialEq, Eq, Debug)] +pub enum DbMode { + /// The database is optional: gn4 starts without a reachable + /// database, logging loudly and serving DB-backed routes as 503. + Dev, + /// The database is required: an unreachable/unconfigured database + /// at startup is a misconfiguration and the process refuses to + /// start (exit non-zero). + Production, +} + /// Reads an on/off setting, failing closed: only a clearly affirmative /// value ("true", "1", "yes", "on", any case) is on; "false", "0", "no", /// "off", the empty string and anything unrecognised (a typo like @@ -253,6 +292,7 @@ mod tests { fn defaults_apply_with_no_files_or_env() { for key in [ "SQL_URI", + "GN4_DB_MODE", "GN4_CONF", "GN4_SECRETS", "GENENETWORK4_PORT", @@ -279,6 +319,11 @@ mod tests { } let config = Config::load(); assert!(config.sql_uri.contains("db_webqtl")); + assert_eq!( + config.db_mode, + DbMode::Dev, + "no-DB start preserved; the production flip needs sign-off" + ); assert_eq!(config.port, 3000); assert_eq!(config.gemma_wrapper_cmd, "gemma-wrapper"); assert!(config.gemma_cache_dir.ends_with("gn4-gemma")); @@ -295,6 +340,19 @@ mod tests { } #[test] + fn db_mode_accepts_only_dev_and_production() { + assert_eq!(parse_db_mode("dev"), DbMode::Dev); + assert_eq!(parse_db_mode(" DEV "), DbMode::Dev); + assert_eq!(parse_db_mode("production"), DbMode::Production); + assert_eq!(parse_db_mode("Production"), DbMode::Production); + // Fail-closed: a typo or a blank value must refuse, not default. + for bad in ["prodution", "prod", "", "none", "devmode"] { + let refused = std::panic::catch_unwind(|| parse_db_mode(bad)).is_err(); + assert!(refused, "{bad:?} must be refused"); + } + } + + #[test] fn a_switch_is_on_only_for_a_clearly_affirmative_value() { for on in ["true", "TRUE", "True", "1", "yes", "ON", " on "] { assert!(parse_switch(on), "{on:?}"); diff --git a/src/db.rs b/src/db.rs index ba22811..2675875 100644 --- a/src/db.rs +++ b/src/db.rs @@ -12,6 +12,42 @@ fn sql_uri() -> String { crate::config::Config::load().sql_uri } +/// Why a database-backed operation failed. +/// +/// The distinction is minted at the `pool()` choke point, the single +/// place a connection is (re)established: `Unavailable` means the pool +/// could not be built at all, so no configured/reachable database +/// exists and DB-backed routes are meaningless (they map to a typed +/// 503 `db_unavailable`). `Query` is any failure on an already-built +/// pool -- a bad query, or a connection that died after the pool was +/// built; those keep the existing 500 `database_error` shape. See the +/// lane REPORT's probe-window limitation. +#[derive(Debug)] +pub enum DbError { + /// The pool could not be built: `SQL_URI` is unset/unparseable or + /// the database is unreachable. Minted only here. + Unavailable(String), + /// A query (or pool checkout) failed on an established pool. + Query(mysql::Error), +} + +impl std::fmt::Display for DbError { + fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result { + match self { + DbError::Unavailable(detail) => write!(f, "{detail}"), + DbError::Query(error) => write!(f, "{error}"), + } + } +} + +impl std::error::Error for DbError {} + +impl From<mysql::Error> for DbError { + fn from(error: mysql::Error) -> Self { + DbError::Query(error) + } +} + /// The shared connection pool, built once and reused by every query /// function below. mysql::Pool::new eagerly opens PoolConstraints:: /// DEFAULT's 10 connections before returning (about 7s against this @@ -37,22 +73,39 @@ fn sql_uri() -> String { /// and the ping only protects against a connection the server closed /// while idle -- which would now surface as a database error on the /// next query using it (the pool then drops that connection). -fn pool() -> Result<Pool, mysql::Error> { +fn pool() -> Result<Pool, DbError> { static POOL: OnceLock<Mutex<Option<Pool>>> = OnceLock::new(); let mut guard = POOL.get_or_init(|| Mutex::new(None)).lock().unwrap(); if let Some(pool) = guard.as_ref() { return Ok(pool.clone()); } - let opts = OptsBuilder::from_opts(Opts::from_url(&sql_uri())?).pool_opts( + let opts = OptsBuilder::from_opts( + Opts::from_url(&sql_uri()).map_err(|e| DbError::Unavailable(e.to_string()))?, + ) + .pool_opts( PoolOpts::default() .with_reset_connection(false) .with_check_health(false), ); - let new_pool = Pool::new(opts)?; + let new_pool = Pool::new(opts).map_err(|e| DbError::Unavailable(e.to_string()))?; *guard = Some(new_pool.clone()); Ok(new_pool) } +/// Startup reachability probe: builds the shared pool (connecting to +/// `SQL_URI`) and runs `SELECT 1`. `Ok(())` means the database is +/// reachable; an `Err` is exactly what makes DB-backed routes answer +/// 503. Used by src/main.rs to refuse (`production`) or warn loudly +/// (`dev`) before serving, and by gn-indexer to refuse before touching +/// an index. The built pool is cached, so a successful probe warms the +/// pool the routes then reuse. +pub fn probe() -> Result<(), DbError> { + let pool = pool()?; + let mut conn = pool.get_conn()?; + conn.query_drop("SELECT 1")?; + Ok(()) +} + /// GeneNetwork stores LRS (likelihood ratio statistic); LOD = LRS / 4.604 /// (the divisor GeneNetwork's RDF data uses; 2 ln 10 is 4.605). Every /// `lod` the API reports is derived with this one constant -- in SQL for rows read from the database, via `lod` for @@ -75,7 +128,7 @@ pub struct Species { } /// Mirrors GeneNetwork3's gn3.db.species.get_all_species. -pub fn get_all_species() -> Result<Vec<Species>, mysql::Error> { +pub fn get_all_species() -> Result<Vec<Species>, DbError> { let pool = pool()?; let mut conn = pool.get_conn()?; conn.query_map( @@ -88,6 +141,7 @@ pub fn get_all_species() -> Result<Vec<Species>, mysql::Error> { family, }, ) + .map_err(DbError::from) } /// One full row of the Species table. @@ -122,7 +176,7 @@ type SpeciesRow = ( /// Fetch one species by Name, FullName or SpeciesName, matching /// GeneNetwork2's wqflask.api.router.get_species_info, but returning /// every column of the Species table rather than a fixed subset. -pub fn get_species_by_name(name: &str) -> Result<Option<SpeciesDetail>, mysql::Error> { +pub fn get_species_by_name(name: &str) -> Result<Option<SpeciesDetail>, DbError> { let pool = pool()?; let mut conn = pool.get_conn()?; conn.exec_first( @@ -161,6 +215,7 @@ pub fn get_species_by_name(name: &str) -> Result<Option<SpeciesDetail>, mysql::E }, ) }) + .map_err(DbError::from) } /// Summary of one InbredSet row, matching the shape of GeneNetwork2's @@ -189,7 +244,7 @@ pub struct Group { /// Mirrors GeneNetwork2's wqflask.api.router.get_groups_list, plus the /// family and species name (filtering is done by the handler). -pub fn get_all_groups() -> Result<Vec<Group>, mysql::Error> { +pub fn get_all_groups() -> Result<Vec<Group>, DbError> { let pool = pool()?; let mut conn = pool.get_conn()?; conn.query_map( @@ -229,6 +284,7 @@ pub fn get_all_groups() -> Result<Vec<Group>, mysql::Error> { } }, ) + .map_err(DbError::from) } /// One mapping tool GeneNetwork offers (a MappingMethod row), e.g. @@ -241,13 +297,14 @@ pub struct MappingMethod { pub name: String, } -pub fn get_mapping_methods() -> Result<Vec<MappingMethod>, mysql::Error> { +pub fn get_mapping_methods() -> Result<Vec<MappingMethod>, DbError> { let pool = pool()?; let mut conn = pool.get_conn()?; conn.query_map( "SELECT Id, Name FROM MappingMethod ORDER BY Id", |(id, name)| MappingMethod { id, name }, ) + .map_err(DbError::from) } /// A GeneNetwork accession number (the "GN602" of a dataset's page): an @@ -270,7 +327,7 @@ pub struct Accession { /// Resolve an accession number: its InfoFiles row, then the dataset of /// that name (PublishFreeze, else ProbeSetFreeze) for its category and /// group. None when no InfoFiles row has the accession. -pub fn get_accession(accession: u32) -> Result<Option<Accession>, mysql::Error> { +pub fn get_accession(accession: u32) -> Result<Option<Accession>, DbError> { let pool = pool()?; let mut conn = pool.get_conn()?; let row: Option<(String, Option<String>)> = conn.exec_first( @@ -351,7 +408,7 @@ pub struct GroupDetail { /// doesn't match by Id) plus gn-guile's gn.data.population:population-info /// (which only matches by Id), returning every column of the InbredSet /// table rather than a fixed subset. -pub fn get_group_by_name(name: &str) -> Result<Option<GroupDetail>, mysql::Error> { +pub fn get_group_by_name(name: &str) -> Result<Option<GroupDetail>, DbError> { let pool = pool()?; let mut conn = pool.get_conn()?; conn.exec_first( @@ -366,6 +423,7 @@ pub fn get_group_by_name(name: &str) -> Result<Option<GroupDetail>, mysql::Error OR InbredSet.FullName = :name OR CAST(InbredSet.Id AS CHAR) = :name", mysql::params! { "name" => name }, ) + .map_err(DbError::from) } /// One strain/sample belonging to a group, matching gn-guile's @@ -385,7 +443,7 @@ pub struct Strain { pub fn get_group_strains( group_id: u16, used_for_mapping: Option<bool>, -) -> Result<Vec<Strain>, mysql::Error> { +) -> Result<Vec<Strain>, DbError> { let pool = pool()?; let mut conn = pool.get_conn()?; conn.exec_map( @@ -402,6 +460,7 @@ pub fn get_group_strains( used_for_mapping: used_for_mapping == "Y", }, ) + .map_err(DbError::from) } /// One strain's value for a published (phenotype) trait, matching @@ -424,7 +483,7 @@ pub struct TraitValue { pub fn get_group_trait_values( group_id: u16, trait_id: u32, -) -> Result<Vec<TraitValue>, mysql::Error> { +) -> Result<Vec<TraitValue>, DbError> { let pool = pool()?; let mut conn = pool.get_conn()?; conn.exec_map( @@ -436,6 +495,7 @@ pub fn get_group_trait_values( mysql::params! { "group_id" => group_id, "trait_id" => trait_id }, |(name, value)| TraitValue { name, value }, ) + .map_err(DbError::from) } /// One published (phenotype) trait's metadata: its PublishXRef mapping @@ -534,7 +594,7 @@ fn trait_info_sql(filter: &str, order_by: &str) -> String { pub fn get_group_trait_info( group_id: u16, trait_id: u32, -) -> Result<Option<TraitInfo>, mysql::Error> { +) -> Result<Option<TraitInfo>, DbError> { let pool = pool()?; let mut conn = pool.get_conn()?; conn.exec_first( @@ -544,17 +604,19 @@ pub fn get_group_trait_info( ), mysql::params! { "group_id" => group_id, "trait_id" => trait_id }, ) + .map_err(DbError::from) } /// Fetch every published trait's metadata (see TraitInfo) within a /// group, the listing equivalent of get_group_trait_info. -pub fn get_group_traits(group_id: u16) -> Result<Vec<TraitInfo>, mysql::Error> { +pub fn get_group_traits(group_id: u16) -> Result<Vec<TraitInfo>, DbError> { let pool = pool()?; let mut conn = pool.get_conn()?; conn.exec( trait_info_sql("PublishXRef.InbredSetId = :group_id", "PublishXRef.Id"), mysql::params! { "group_id" => group_id }, ) + .map_err(DbError::from) } /// The traits of a group whose short name (either the post- or the @@ -562,7 +624,7 @@ pub fn get_group_traits(group_id: u16) -> Result<Vec<TraitInfo>, mysql::Error> { pub fn get_group_traits_by_abbreviation( group_id: u16, abbreviation: &str, -) -> Result<Vec<TraitInfo>, mysql::Error> { +) -> Result<Vec<TraitInfo>, DbError> { let pool = pool()?; let mut conn = pool.get_conn()?; conn.exec( @@ -574,11 +636,12 @@ pub fn get_group_traits_by_abbreviation( ), mysql::params! { "group_id" => group_id, "abbreviation" => abbreviation }, ) + .map_err(DbError::from) } /// Every published trait (across groups) reported in the publication /// with this PubMed id, or None when no such publication exists. -pub fn get_publication_traits(pubmed_id: u32) -> Result<Option<Vec<TraitInfo>>, mysql::Error> { +pub fn get_publication_traits(pubmed_id: u32) -> Result<Option<Vec<TraitInfo>>, DbError> { let pool = pool()?; let mut conn = pool.get_conn()?; let exists: Option<u32> = conn.exec_first( @@ -596,6 +659,7 @@ pub fn get_publication_traits(pubmed_id: u32) -> Result<Option<Vec<TraitInfo>>, mysql::params! { "pubmed_id" => pubmed_id }, ) .map(Some) + .map_err(DbError::from) } /// What search hits are annotated with from PublishXRef/Phenotype @@ -611,7 +675,7 @@ pub struct PublishAnnotation { /// exist are absent from the result. pub fn get_publish_annotations( keys: &[(String, u32)], -) -> Result<std::collections::HashMap<(String, u32), PublishAnnotation>, mysql::Error> { +) -> Result<std::collections::HashMap<(String, u32), PublishAnnotation>, DbError> { let pool = pool()?; let mut conn = pool.get_conn()?; let pairs = vec!["(?, ?)"; keys.len()].join(", "); @@ -669,7 +733,7 @@ pub struct Dataset { /// GeneNetwork2's wqflask.api.router.get_datasets_for_group (without /// its species-scoped variant, which this API doesn't need: a group /// name alone already picks one species). -pub fn get_group_datasets(group_id: u16) -> Result<Vec<Dataset>, mysql::Error> { +pub fn get_group_datasets(group_id: u16) -> Result<Vec<Dataset>, DbError> { let pool = pool()?; let mut conn = pool.get_conn()?; conn.exec_map( @@ -711,6 +775,7 @@ pub fn get_group_datasets(group_id: u16) -> Result<Vec<Dataset>, mysql::Error> { } }, ) + .map_err(DbError::from) } /// One strain's case-attribute values within a group: extra per-strain @@ -743,13 +808,14 @@ pub struct CaseAttributes { /// (trimmed to just the name -- Description isn't used by genenetwork4 /// yet, and CaseAttributeId/InbredSetId are already implied by this /// being scoped to one group). -fn get_group_case_attribute_names(group_id: u16) -> Result<Vec<String>, mysql::Error> { +fn get_group_case_attribute_names(group_id: u16) -> Result<Vec<String>, DbError> { let pool = pool()?; let mut conn = pool.get_conn()?; conn.exec( "SELECT Name FROM CaseAttribute WHERE InbredSetId = :group_id ORDER BY Name", mysql::params! { "group_id" => group_id }, ) + .map_err(DbError::from) } /// Fetch every strain's case-attribute values for a group, based on @@ -769,7 +835,7 @@ fn get_group_case_attribute_names(group_id: u16) -> Result<Vec<String>, mysql::E /// ordered by strain name, so adjacent rows for the same strain are /// folded into one `StrainCaseAttributes` as they're read rather than /// grouped afterwards. -fn get_group_case_attribute_values(group_id: u16) -> Result<Vec<StrainCaseAttributes>, mysql::Error> { +fn get_group_case_attribute_values(group_id: u16) -> Result<Vec<StrainCaseAttributes>, DbError> { let pool = pool()?; let mut conn = pool.get_conn()?; let rows: Vec<(String, String, String)> = conn.exec( @@ -804,7 +870,7 @@ fn get_group_case_attribute_values(group_id: u16) -> Result<Vec<StrainCaseAttrib /// Fetch a group's case attributes: every attribute name defined for /// it, and every strain's values for them. See `CaseAttributes`. -pub fn get_group_case_attributes(group_id: u16) -> Result<CaseAttributes, mysql::Error> { +pub fn get_group_case_attributes(group_id: u16) -> Result<CaseAttributes, DbError> { Ok(CaseAttributes { names: get_group_case_attribute_names(group_id)?, values: get_group_case_attribute_values(group_id)?, @@ -816,7 +882,7 @@ pub fn get_group_case_attributes(group_id: u16) -> Result<CaseAttributes, mysql: /// GeneNetwork2's own species-scoped variant of /// wqflask.api.router.get_datasets_for_group still requires a group /// name (species only disambiguates it); this doesn't need one. -pub fn get_species_datasets(species_id: u16) -> Result<Vec<Dataset>, mysql::Error> { +pub fn get_species_datasets(species_id: u16) -> Result<Vec<Dataset>, DbError> { let pool = pool()?; let mut conn = pool.get_conn()?; conn.exec_map( @@ -859,6 +925,7 @@ pub fn get_species_datasets(species_id: u16) -> Result<Vec<Dataset>, mysql::Erro } }, ) + .map_err(DbError::from) } /// Columns of the Dataset row, in SELECT order. @@ -879,7 +946,7 @@ type DatasetRow = ( /// Fetch one dataset by Name, Name2, FullName, ShortName or numeric /// Id, matching GeneNetwork2's wqflask.api.router.get_dataset_info's /// matching logic for its ProbeSet (mRNA expression) branch. -pub fn get_dataset_by_name(name: &str) -> Result<Option<Dataset>, mysql::Error> { +pub fn get_dataset_by_name(name: &str) -> Result<Option<Dataset>, DbError> { let pool = pool()?; let mut conn = pool.get_conn()?; conn.exec_first( @@ -926,6 +993,7 @@ pub fn get_dataset_by_name(name: &str) -> Result<Option<Dataset>, mysql::Error> }, ) }) + .map_err(DbError::from) } /// One row of the PublishFreeze table -- the Publish/phenotype @@ -954,7 +1022,7 @@ pub struct PublishDataset { /// owning group's Id/Name/InbredSetName/FullName -- mirroring /// GeneNetwork2's wqflask.api.router.trait_sample_data's Publish /// branch, which accepts either (e.g. "BXDPublish" or plain "BXD"). -pub fn get_publish_freeze_by_name(name: &str) -> Result<Option<PublishDataset>, mysql::Error> { +pub fn get_publish_freeze_by_name(name: &str) -> Result<Option<PublishDataset>, DbError> { let pool = pool()?; let mut conn = pool.get_conn()?; conn.exec_first( @@ -970,6 +1038,7 @@ pub fn get_publish_freeze_by_name(name: &str) -> Result<Option<PublishDataset>, OR InbredSet.FullName = :name OR CAST(InbredSet.Id AS CHAR) = :name", mysql::params! { "name" => name }, ) + .map_err(DbError::from) } /// Which category a `/traits/{dataset}/...` request's dataset resolved @@ -984,7 +1053,7 @@ pub enum TraitDataset { /// then PublishFreeze (a dataset name is expected to match at most one /// of these). Geno/GenoFreeze isn't supported yet, matching the rest /// of this API. -pub fn resolve_trait_dataset(name: &str) -> Result<Option<TraitDataset>, mysql::Error> { +pub fn resolve_trait_dataset(name: &str) -> Result<Option<TraitDataset>, DbError> { if let Some(dataset) = get_dataset_by_name(name)? { return Ok(Some(TraitDataset::ProbeSet(dataset))); } @@ -1041,7 +1110,7 @@ type ProbeSetInfoRow = ( pub fn get_dataset_probeset_info( dataset_id: u16, probeset_name: &str, -) -> Result<Option<ProbeSetInfo>, mysql::Error> { +) -> Result<Option<ProbeSetInfo>, DbError> { let pool = pool()?; let mut conn = pool.get_conn()?; conn.exec_first( @@ -1089,12 +1158,13 @@ pub fn get_dataset_probeset_info( }, ) }) + .map_err(DbError::from) } /// Fetch every expression (ProbeSet) trait's metadata (see /// ProbeSetInfo) within a dataset, the listing equivalent of /// get_dataset_probeset_info. -pub fn get_dataset_probesets(dataset_id: u16) -> Result<Vec<ProbeSetInfo>, mysql::Error> { +pub fn get_dataset_probesets(dataset_id: u16) -> Result<Vec<ProbeSetInfo>, DbError> { let pool = pool()?; let mut conn = pool.get_conn()?; conn.exec_map( @@ -1139,6 +1209,7 @@ pub fn get_dataset_probesets(dataset_id: u16) -> Result<Vec<ProbeSetInfo>, mysql } }, ) + .map_err(DbError::from) } /// One strain's value for an expression (ProbeSet) trait, matching @@ -1156,7 +1227,7 @@ pub struct ProbeSetValue { pub fn get_dataset_probeset_values( dataset_id: u16, probeset_name: &str, -) -> Result<Vec<ProbeSetValue>, mysql::Error> { +) -> Result<Vec<ProbeSetValue>, DbError> { let pool = pool()?; let mut conn = pool.get_conn()?; conn.exec_map( @@ -1169,4 +1240,5 @@ pub fn get_dataset_probeset_values( mysql::params! { "dataset_id" => dataset_id, "probeset_name" => probeset_name }, |(name, value)| ProbeSetValue { name, value }, ) + .map_err(DbError::from) } diff --git a/src/error.rs b/src/error.rs index 3f9479a..7daa103 100644 --- a/src/error.rs +++ b/src/error.rs @@ -111,6 +111,32 @@ impl ApiError { self.next = Some(next.to_string()); self } + + /// Map a database failure onto the shared error contract. An + /// unreachable/unconfigured database (minted at `db::pool()`) is a + /// typed 503 `db_unavailable` -- the database itself is down, not + /// this request; a query failure on an established pool keeps the + /// existing 500 `database_error`. Every DB-touching route maps + /// through this one place, so all four `From`-adapter shapes and + /// the inline `map_err` sites agree. + pub fn from_db(error: crate::db::DbError, operation_id: &str) -> ApiError { + match error { + crate::db::DbError::Unavailable(detail) => ApiError::new( + StatusCode::SERVICE_UNAVAILABLE, + "db_unavailable", + "Database unavailable", + detail, + operation_id, + ), + crate::db::DbError::Query(error) => ApiError::new( + StatusCode::INTERNAL_SERVER_ERROR, + "database_error", + "Database error", + error.to_string(), + operation_id, + ), + } + } } impl IntoResponse for ApiError { diff --git a/src/lib.rs b/src/lib.rs index f011219..9297d87 100644 --- a/src/lib.rs +++ b/src/lib.rs @@ -8,7 +8,7 @@ use utoipa_axum::routes; pub mod auth; pub mod config; -mod db; +pub mod db; pub mod embedded; mod error; pub mod indexer; diff --git a/src/main.rs b/src/main.rs index 21a0e55..39c9c1d 100644 --- a/src/main.rs +++ b/src/main.rs @@ -1,6 +1,31 @@ +use genenetwork4::config::DbMode; + #[tokio::main] async fn main() { let config = genenetwork4::config::Config::load(); + // Startup database probe (`SELECT 1` through the shared pool), + // BEFORE the listener is bound. `production` treats an unreachable + // database as a misconfiguration and refuses to start; `dev` starts + // anyway, loudly, with DB-backed routes answering the typed 503 + // (src/error.rs). The message names both config keys so the refusal + // is unambiguous. + if let Err(error) = genenetwork4::db::probe() { + match config.db_mode { + DbMode::Production => { + eprintln!( + "FATAL: GN4_DB_MODE=production requires a reachable database, but SQL_URI \ + is unreachable or unconfigured ({error}); refusing to start." + ); + std::process::exit(1); + } + DbMode::Dev => { + eprintln!( + "database not configured/unreachable — DB-backed routes disabled \ + (GN4_DB_MODE=dev, SQL_URI unreachable or unconfigured): {error}" + ); + } + } + } let bind_ip: std::net::IpAddr = config .bind .parse() diff --git a/src/routes/accessions.rs b/src/routes/accessions.rs index c55bfd7..66801e0 100644 --- a/src/routes/accessions.rs +++ b/src/routes/accessions.rs @@ -58,16 +58,7 @@ pub async fn accession( ) .with_request_id(rid.clone()) })? - .map_err(|e| { - ApiError::new( - StatusCode::INTERNAL_SERVER_ERROR, - "database_error", - "Database error", - e.to_string(), - "accession", - ) - .with_request_id(rid.clone()) - })? + .map_err(|e| ApiError::from_db(e, "accession").with_request_id(rid.clone()))? .map(|found| { let group = found.group.as_deref().map(path_segment); let name = path_segment(&found.name); diff --git a/src/routes/datasets.rs b/src/routes/datasets.rs index b011099..8c903ce 100644 --- a/src/routes/datasets.rs +++ b/src/routes/datasets.rs @@ -41,16 +41,7 @@ pub async fn dataset_by_name( ) .with_request_id(rid.clone()) })? - .map_err(|e| { - ApiError::new( - StatusCode::INTERNAL_SERVER_ERROR, - "database_error", - "Database error", - e.to_string(), - "dataset_by_name", - ) - .with_request_id(rid.clone()) - })? + .map_err(|e| ApiError::from_db(e, "dataset_by_name").with_request_id(rid.clone()))? .map(|dataset| { Json(ctx.envelope(dataset).with_next( &format!("/datasets/{}/probesets", path_segment(¬_found_name)), @@ -140,14 +131,9 @@ pub async fn dataset_probesets( .with_request_id(rid.clone()) })? .map_err(|e| match e { - DatasetAuthError::Db(e) => ApiError::new( - StatusCode::INTERNAL_SERVER_ERROR, - "database_error", - "Database error", - e.to_string(), - "dataset_probesets", - ) - .with_request_id(rid.clone()), + DatasetAuthError::Db(e) => { + ApiError::from_db(e, "dataset_probesets").with_request_id(rid.clone()) + } DatasetAuthError::Auth(e) => ApiError::new( StatusCode::BAD_GATEWAY, "auth_server_error", @@ -202,12 +188,12 @@ enum DatasetProbesetLookup<T> { /// Either a database error, or a failure to reach/parse gn-auth's /// answer while checking authorization for a non-public dataset. enum DatasetAuthError { - Db(mysql::Error), + Db(db::DbError), Auth(String), } -impl From<mysql::Error> for DatasetAuthError { - fn from(e: mysql::Error) -> Self { +impl From<db::DbError> for DatasetAuthError { + fn from(e: db::DbError) -> Self { DatasetAuthError::Db(e) } } @@ -277,14 +263,9 @@ pub async fn dataset_probeset_info( .with_request_id(rid.clone()) })? .map_err(|e| match e { - DatasetAuthError::Db(e) => ApiError::new( - StatusCode::INTERNAL_SERVER_ERROR, - "database_error", - "Database error", - e.to_string(), - "dataset_probeset_info", - ) - .with_request_id(rid.clone()), + DatasetAuthError::Db(e) => { + ApiError::from_db(e, "dataset_probeset_info").with_request_id(rid.clone()) + } DatasetAuthError::Auth(e) => ApiError::new( StatusCode::BAD_GATEWAY, "auth_server_error", @@ -404,14 +385,9 @@ pub async fn dataset_probeset_values( .with_request_id(rid.clone()) })? .map_err(|e| match e { - DatasetAuthError::Db(e) => ApiError::new( - StatusCode::INTERNAL_SERVER_ERROR, - "database_error", - "Database error", - e.to_string(), - "dataset_probeset_values", - ) - .with_request_id(rid.clone()), + DatasetAuthError::Db(e) => { + ApiError::from_db(e, "dataset_probeset_values").with_request_id(rid.clone()) + } DatasetAuthError::Auth(e) => ApiError::new( StatusCode::BAD_GATEWAY, "auth_server_error", diff --git a/src/routes/groups.rs b/src/routes/groups.rs index 1c87cdd..b43fb31 100644 --- a/src/routes/groups.rs +++ b/src/routes/groups.rs @@ -81,16 +81,7 @@ pub async fn groups( "The species the groups belong to (filter groups with ?species=<name>)", )) }) - .map_err(|e| { - ApiError::new( - StatusCode::INTERNAL_SERVER_ERROR, - "database_error", - "Database error", - e.to_string(), - "groups", - ) - .with_request_id(rid) - }) + .map_err(|e| ApiError::from_db(e, "groups").with_request_id(rid)) } #[utoipa::path(get, path = "/groups/{group}", summary = "Get one group's full record by name or numeric id", params( @@ -121,16 +112,7 @@ pub async fn group_by_name( ) .with_request_id(rid.clone()) })? - .map_err(|e| { - ApiError::new( - StatusCode::INTERNAL_SERVER_ERROR, - "database_error", - "Database error", - e.to_string(), - "group_by_name", - ) - .with_request_id(rid.clone()) - })? + .map_err(|e| ApiError::from_db(e, "group_by_name").with_request_id(rid.clone()))? .map(|detail| { let name = path_segment(&detail.name); Json( @@ -204,16 +186,7 @@ pub async fn group_strains( ) .with_request_id(rid.clone()) })? - .map_err(|e| { - ApiError::new( - StatusCode::INTERNAL_SERVER_ERROR, - "database_error", - "Database error", - e.to_string(), - "group_strains", - ) - .with_request_id(rid.clone()) - })? + .map_err(|e| ApiError::from_db(e, "group_strains").with_request_id(rid.clone()))? .map(|strains| { let group = path_segment(¬_found_name); Json( @@ -255,12 +228,12 @@ enum GroupTraitLookup<T> { /// Either a database error, or a failure to reach/parse gn-auth's /// answer while checking authorization for a non-public group. enum GroupAuthError { - Db(mysql::Error), + Db(db::DbError), Auth(String), } -impl From<mysql::Error> for GroupAuthError { - fn from(e: mysql::Error) -> Self { +impl From<db::DbError> for GroupAuthError { + fn from(e: db::DbError) -> Self { GroupAuthError::Db(e) } } @@ -350,14 +323,9 @@ pub async fn group_trait_values( .with_request_id(rid.clone()) })? .map_err(|e| match e { - GroupAuthError::Db(e) => ApiError::new( - StatusCode::INTERNAL_SERVER_ERROR, - "database_error", - "Database error", - e.to_string(), - "group_trait_values", - ) - .with_request_id(rid.clone()), + GroupAuthError::Db(e) => { + ApiError::from_db(e, "group_trait_values").with_request_id(rid.clone()) + } GroupAuthError::Auth(e) => ApiError::new( StatusCode::BAD_GATEWAY, "auth_server_error", @@ -462,14 +430,9 @@ pub async fn group_trait_info( .with_request_id(rid.clone()) })? .map_err(|e| match e { - GroupAuthError::Db(e) => ApiError::new( - StatusCode::INTERNAL_SERVER_ERROR, - "database_error", - "Database error", - e.to_string(), - "group_trait_info", - ) - .with_request_id(rid.clone()), + GroupAuthError::Db(e) => { + ApiError::from_db(e, "group_trait_info").with_request_id(rid.clone()) + } GroupAuthError::Auth(e) => ApiError::new( StatusCode::BAD_GATEWAY, "auth_server_error", @@ -604,14 +567,7 @@ pub async fn group_traits( .with_request_id(rid.clone()) })? .map_err(|e| match e { - GroupAuthError::Db(e) => ApiError::new( - StatusCode::INTERNAL_SERVER_ERROR, - "database_error", - "Database error", - e.to_string(), - "group_traits", - ) - .with_request_id(rid.clone()), + GroupAuthError::Db(e) => ApiError::from_db(e, "group_traits").with_request_id(rid.clone()), GroupAuthError::Auth(e) => ApiError::new( StatusCode::BAD_GATEWAY, "auth_server_error", @@ -686,16 +642,7 @@ pub async fn group_case_attributes( ) .with_request_id(rid.clone()) })? - .map_err(|e| { - ApiError::new( - StatusCode::INTERNAL_SERVER_ERROR, - "database_error", - "Database error", - e.to_string(), - "group_case_attributes", - ) - .with_request_id(rid.clone()) - })? + .map_err(|e| ApiError::from_db(e, "group_case_attributes").with_request_id(rid.clone()))? .map(|case_attributes| { Json(ctx.envelope(case_attributes).with_next( &format!("/groups/{}/strains", path_segment(¬_found_name)), @@ -748,16 +695,7 @@ pub async fn group_datasets( ) .with_request_id(rid.clone()) })? - .map_err(|e| { - ApiError::new( - StatusCode::INTERNAL_SERVER_ERROR, - "database_error", - "Database error", - e.to_string(), - "group_datasets", - ) - .with_request_id(rid.clone()) - })? + .map_err(|e| ApiError::from_db(e, "group_datasets").with_request_id(rid.clone()))? .map(|datasets| { Json(ctx.envelope(datasets).with_next( &format!("/groups/{}", path_segment(¬_found_name)), diff --git a/src/routes/mapping_methods.rs b/src/routes/mapping_methods.rs index 224d90f..96bd49d 100644 --- a/src/routes/mapping_methods.rs +++ b/src/routes/mapping_methods.rs @@ -39,14 +39,5 @@ pub async fn mapping_methods( "Groups, each with the mapping tool offered for it (`mapping_method`)", )) }) - .map_err(|e| { - ApiError::new( - StatusCode::INTERNAL_SERVER_ERROR, - "database_error", - "Database error", - e.to_string(), - "mapping_methods", - ) - .with_request_id(rid) - }) + .map_err(|e| ApiError::from_db(e, "mapping_methods").with_request_id(rid)) } diff --git a/src/routes/publications.rs b/src/routes/publications.rs index 850f934..635486a 100644 --- a/src/routes/publications.rs +++ b/src/routes/publications.rs @@ -18,12 +18,12 @@ enum PublicationTraits { } enum LookupError { - Db(mysql::Error), + Db(db::DbError), Auth(String), } -impl From<mysql::Error> for LookupError { - fn from(e: mysql::Error) -> Self { +impl From<db::DbError> for LookupError { + fn from(e: db::DbError) -> Self { LookupError::Db(e) } } @@ -102,14 +102,9 @@ pub async fn publication_traits( .with_request_id(rid.clone()) })? .map_err(|e| match e { - LookupError::Db(e) => ApiError::new( - StatusCode::INTERNAL_SERVER_ERROR, - "database_error", - "Database error", - e.to_string(), - "publication_traits", - ) - .with_request_id(rid.clone()), + LookupError::Db(e) => { + ApiError::from_db(e, "publication_traits").with_request_id(rid.clone()) + } LookupError::Auth(e) => ApiError::new( StatusCode::BAD_GATEWAY, "auth_server_error", diff --git a/src/routes/species.rs b/src/routes/species.rs index 94f8736..4214f02 100644 --- a/src/routes/species.rs +++ b/src/routes/species.rs @@ -42,16 +42,7 @@ pub async fn species( "Groups (populations); add ?species=<name> to list one species' groups", )) }) - .map_err(|e| { - ApiError::new( - StatusCode::INTERNAL_SERVER_ERROR, - "database_error", - "Database error", - e.to_string(), - "species", - ) - .with_request_id(rid) - }) + .map_err(|e| ApiError::from_db(e, "species").with_request_id(rid)) } #[utoipa::path(get, path = "/species/{species}", summary = "Get one species' full record by name", params( @@ -82,16 +73,7 @@ pub async fn species_by_name( ) .with_request_id(rid.clone()) })? - .map_err(|e| { - ApiError::new( - StatusCode::INTERNAL_SERVER_ERROR, - "database_error", - "Database error", - e.to_string(), - "species_by_name", - ) - .with_request_id(rid.clone()) - })? + .map_err(|e| ApiError::from_db(e, "species_by_name").with_request_id(rid.clone()))? .map(|detail| { let name = path_segment(&detail.name); Json( @@ -149,16 +131,7 @@ pub async fn species_datasets( ) .with_request_id(rid.clone()) })? - .map_err(|e| { - ApiError::new( - StatusCode::INTERNAL_SERVER_ERROR, - "database_error", - "Database error", - e.to_string(), - "species_datasets", - ) - .with_request_id(rid.clone()) - })? + .map_err(|e| ApiError::from_db(e, "species_datasets").with_request_id(rid.clone()))? .map(|datasets| { Json(ctx.envelope(datasets).with_next( &format!("/species/{}", path_segment(¬_found_name)), diff --git a/src/routes/traits.rs b/src/routes/traits.rs index 0b7e77c..1ceb752 100644 --- a/src/routes/traits.rs +++ b/src/routes/traits.rs @@ -43,12 +43,12 @@ pub enum TraitInfoResponse { /// matching this codebase's existing per-module lookup/error types /// (see e.g. GroupTraitLookup vs DatasetProbesetLookup). enum TraitAuthError { - Db(mysql::Error), + Db(db::DbError), Auth(String), } -impl From<mysql::Error> for TraitAuthError { - fn from(e: mysql::Error) -> Self { +impl From<db::DbError> for TraitAuthError { + fn from(e: db::DbError) -> Self { TraitAuthError::Db(e) } } @@ -147,14 +147,9 @@ pub async fn traits( .with_request_id(rid.clone()) })? .map_err(|e| match e { - TraitAuthError::Db(e) => ApiError::new( - StatusCode::INTERNAL_SERVER_ERROR, - "database_error", - "Database error", - e.to_string(), - "traits", - ) - .with_request_id(rid.clone()), + TraitAuthError::Db(e) => { + ApiError::from_db(e, "traits").with_request_id(rid.clone()) + } TraitAuthError::Auth(e) => ApiError::new( StatusCode::BAD_GATEWAY, "auth_server_error", @@ -294,14 +289,7 @@ pub async fn trait_info( .with_request_id(rid.clone()) })? .map_err(|e| match e { - TraitAuthError::Db(e) => ApiError::new( - StatusCode::INTERNAL_SERVER_ERROR, - "database_error", - "Database error", - e.to_string(), - "trait_info", - ) - .with_request_id(rid.clone()), + TraitAuthError::Db(e) => ApiError::from_db(e, "trait_info").with_request_id(rid.clone()), TraitAuthError::Auth(e) => ApiError::new( StatusCode::BAD_GATEWAY, "auth_server_error", @@ -473,14 +461,7 @@ pub async fn trait_values( .with_request_id(rid.clone()) })? .map_err(|e| match e { - TraitAuthError::Db(e) => ApiError::new( - StatusCode::INTERNAL_SERVER_ERROR, - "database_error", - "Database error", - e.to_string(), - "trait_values", - ) - .with_request_id(rid.clone()), + TraitAuthError::Db(e) => ApiError::from_db(e, "trait_values").with_request_id(rid.clone()), TraitAuthError::Auth(e) => ApiError::new( StatusCode::BAD_GATEWAY, "auth_server_error", |
