Line data Source code
1 : //!
2 : //! Management HTTP API
3 : //!
4 : use std::cmp::Reverse;
5 : use std::collections::BinaryHeap;
6 : use std::collections::HashMap;
7 : use std::str::FromStr;
8 : use std::sync::Arc;
9 : use std::time::Duration;
10 :
11 : use anyhow::{anyhow, Context, Result};
12 : use enumset::EnumSet;
13 : use futures::TryFutureExt;
14 : use humantime::format_rfc3339;
15 : use hyper::header;
16 : use hyper::StatusCode;
17 : use hyper::{Body, Request, Response, Uri};
18 : use metrics::launch_timestamp::LaunchTimestamp;
19 : use pageserver_api::models::AuxFilePolicy;
20 : use pageserver_api::models::IngestAuxFilesRequest;
21 : use pageserver_api::models::ListAuxFilesRequest;
22 : use pageserver_api::models::LocationConfig;
23 : use pageserver_api::models::LocationConfigListResponse;
24 : use pageserver_api::models::ShardParameters;
25 : use pageserver_api::models::TenantDetails;
26 : use pageserver_api::models::TenantLocationConfigResponse;
27 : use pageserver_api::models::TenantScanRemoteStorageResponse;
28 : use pageserver_api::models::TenantScanRemoteStorageShard;
29 : use pageserver_api::models::TenantShardLocation;
30 : use pageserver_api::models::TenantShardSplitRequest;
31 : use pageserver_api::models::TenantShardSplitResponse;
32 : use pageserver_api::models::TenantSorting;
33 : use pageserver_api::models::TenantState;
34 : use pageserver_api::models::TopTenantShardItem;
35 : use pageserver_api::models::TopTenantShardsRequest;
36 : use pageserver_api::models::TopTenantShardsResponse;
37 : use pageserver_api::models::{
38 : DownloadRemoteLayersTaskSpawnRequest, LocationConfigMode, TenantAttachRequest,
39 : TenantLocationConfigRequest,
40 : };
41 : use pageserver_api::shard::ShardCount;
42 : use pageserver_api::shard::TenantShardId;
43 : use remote_storage::DownloadError;
44 : use remote_storage::GenericRemoteStorage;
45 : use remote_storage::TimeTravelError;
46 : use tenant_size_model::{SizeResult, StorageModel};
47 : use tokio_util::sync::CancellationToken;
48 : use tracing::*;
49 : use utils::auth::JwtAuth;
50 : use utils::failpoint_support::failpoints_handler;
51 : use utils::http::endpoint::prometheus_metrics_handler;
52 : use utils::http::endpoint::request_span;
53 : use utils::http::json::json_request_or_empty_body;
54 : use utils::http::request::{get_request_param, must_get_query_param, parse_query_param};
55 :
56 : use crate::context::{DownloadBehavior, RequestContext};
57 : use crate::deletion_queue::DeletionQueueClient;
58 : use crate::metrics::{StorageTimeOperation, STORAGE_TIME_GLOBAL};
59 : use crate::pgdatadir_mapping::LsnForTimestamp;
60 : use crate::task_mgr::TaskKind;
61 : use crate::tenant::config::{LocationConf, TenantConfOpt};
62 : use crate::tenant::mgr::GetActiveTenantError;
63 : use crate::tenant::mgr::{
64 : GetTenantError, TenantManager, TenantMapError, TenantMapInsertError, TenantSlotError,
65 : TenantSlotUpsertError, TenantStateError,
66 : };
67 : use crate::tenant::mgr::{TenantSlot, UpsertLocationError};
68 : use crate::tenant::remote_timeline_client;
69 : use crate::tenant::remote_timeline_client::download_index_part;
70 : use crate::tenant::remote_timeline_client::list_remote_tenant_shards;
71 : use crate::tenant::remote_timeline_client::list_remote_timelines;
72 : use crate::tenant::secondary::SecondaryController;
73 : use crate::tenant::size::ModelInputs;
74 : use crate::tenant::storage_layer::LayerAccessStatsReset;
75 : use crate::tenant::storage_layer::LayerName;
76 : use crate::tenant::timeline::CompactFlags;
77 : use crate::tenant::timeline::CompactionError;
78 : use crate::tenant::timeline::Timeline;
79 : use crate::tenant::GetTimelineError;
80 : use crate::tenant::SpawnMode;
81 : use crate::tenant::{LogicalSizeCalculationCause, PageReconstructError};
82 : use crate::{config::PageServerConf, tenant::mgr};
83 : use crate::{disk_usage_eviction_task, tenant};
84 : use pageserver_api::models::{
85 : StatusResponse, TenantConfigRequest, TenantCreateRequest, TenantCreateResponse, TenantInfo,
86 : TimelineCreateRequest, TimelineGcRequest, TimelineInfo,
87 : };
88 : use utils::{
89 : auth::SwappableJwtAuth,
90 : generation::Generation,
91 : http::{
92 : endpoint::{self, attach_openapi_ui, auth_middleware, check_permission_with},
93 : error::{ApiError, HttpErrorBody},
94 : json::{json_request, json_response},
95 : request::parse_request_param,
96 : RequestExt, RouterBuilder,
97 : },
98 : id::{TenantId, TimelineId},
99 : lsn::Lsn,
100 : };
101 :
102 : // For APIs that require an Active tenant, how long should we block waiting for that state?
103 : // This is not functionally necessary (clients will retry), but avoids generating a lot of
104 : // failed API calls while tenants are activating.
105 : #[cfg(not(feature = "testing"))]
106 : pub(crate) const ACTIVE_TENANT_TIMEOUT: Duration = Duration::from_millis(5000);
107 :
108 : // Tests run on slow/oversubscribed nodes, and may need to wait much longer for tenants to
109 : // finish attaching, if calls to remote storage are slow.
110 : #[cfg(feature = "testing")]
111 : pub(crate) const ACTIVE_TENANT_TIMEOUT: Duration = Duration::from_millis(30000);
112 :
113 : pub struct State {
114 : conf: &'static PageServerConf,
115 : tenant_manager: Arc<TenantManager>,
116 : auth: Option<Arc<SwappableJwtAuth>>,
117 : allowlist_routes: Vec<Uri>,
118 : remote_storage: GenericRemoteStorage,
119 : broker_client: storage_broker::BrokerClientChannel,
120 : disk_usage_eviction_state: Arc<disk_usage_eviction_task::State>,
121 : deletion_queue_client: DeletionQueueClient,
122 : secondary_controller: SecondaryController,
123 : latest_utilization: tokio::sync::Mutex<Option<(std::time::Instant, bytes::Bytes)>>,
124 : }
125 :
126 : impl State {
127 : #[allow(clippy::too_many_arguments)]
128 0 : pub fn new(
129 0 : conf: &'static PageServerConf,
130 0 : tenant_manager: Arc<TenantManager>,
131 0 : auth: Option<Arc<SwappableJwtAuth>>,
132 0 : remote_storage: GenericRemoteStorage,
133 0 : broker_client: storage_broker::BrokerClientChannel,
134 0 : disk_usage_eviction_state: Arc<disk_usage_eviction_task::State>,
135 0 : deletion_queue_client: DeletionQueueClient,
136 0 : secondary_controller: SecondaryController,
137 0 : ) -> anyhow::Result<Self> {
138 0 : let allowlist_routes = ["/v1/status", "/v1/doc", "/swagger.yml", "/metrics"]
139 0 : .iter()
140 0 : .map(|v| v.parse().unwrap())
141 0 : .collect::<Vec<_>>();
142 0 : Ok(Self {
143 0 : conf,
144 0 : tenant_manager,
145 0 : auth,
146 0 : allowlist_routes,
147 0 : remote_storage,
148 0 : broker_client,
149 0 : disk_usage_eviction_state,
150 0 : deletion_queue_client,
151 0 : secondary_controller,
152 0 : latest_utilization: Default::default(),
153 0 : })
154 0 : }
155 : }
156 :
157 : #[inline(always)]
158 0 : fn get_state(request: &Request<Body>) -> &State {
159 0 : request
160 0 : .data::<Arc<State>>()
161 0 : .expect("unknown state type")
162 0 : .as_ref()
163 0 : }
164 :
165 : #[inline(always)]
166 0 : fn get_config(request: &Request<Body>) -> &'static PageServerConf {
167 0 : get_state(request).conf
168 0 : }
169 :
170 : /// Check that the requester is authorized to operate on given tenant
171 0 : fn check_permission(request: &Request<Body>, tenant_id: Option<TenantId>) -> Result<(), ApiError> {
172 0 : check_permission_with(request, |claims| {
173 0 : crate::auth::check_permission(claims, tenant_id)
174 0 : })
175 0 : }
176 :
177 : impl From<PageReconstructError> for ApiError {
178 0 : fn from(pre: PageReconstructError) -> ApiError {
179 0 : match pre {
180 0 : PageReconstructError::Other(pre) => ApiError::InternalServerError(pre),
181 0 : PageReconstructError::MissingKey(e) => {
182 0 : ApiError::InternalServerError(anyhow::anyhow!("{e}"))
183 : }
184 0 : PageReconstructError::Cancelled => ApiError::Cancelled,
185 0 : PageReconstructError::AncestorLsnTimeout(e) => ApiError::Timeout(format!("{e}").into()),
186 0 : PageReconstructError::WalRedo(pre) => ApiError::InternalServerError(pre),
187 : }
188 0 : }
189 : }
190 :
191 : impl From<TenantMapInsertError> for ApiError {
192 0 : fn from(tmie: TenantMapInsertError) -> ApiError {
193 0 : match tmie {
194 0 : TenantMapInsertError::SlotError(e) => e.into(),
195 0 : TenantMapInsertError::SlotUpsertError(e) => e.into(),
196 0 : TenantMapInsertError::Other(e) => ApiError::InternalServerError(e),
197 : }
198 0 : }
199 : }
200 :
201 : impl From<TenantSlotError> for ApiError {
202 0 : fn from(e: TenantSlotError) -> ApiError {
203 0 : use TenantSlotError::*;
204 0 : match e {
205 0 : NotFound(tenant_id) => {
206 0 : ApiError::NotFound(anyhow::anyhow!("NotFound: tenant {tenant_id}").into())
207 : }
208 : InProgress => {
209 0 : ApiError::ResourceUnavailable("Tenant is being modified concurrently".into())
210 : }
211 0 : MapState(e) => e.into(),
212 : }
213 0 : }
214 : }
215 :
216 : impl From<TenantSlotUpsertError> for ApiError {
217 0 : fn from(e: TenantSlotUpsertError) -> ApiError {
218 0 : use TenantSlotUpsertError::*;
219 0 : match e {
220 0 : InternalError(e) => ApiError::InternalServerError(anyhow::anyhow!("{e}")),
221 0 : MapState(e) => e.into(),
222 0 : ShuttingDown(_) => ApiError::ShuttingDown,
223 : }
224 0 : }
225 : }
226 :
227 : impl From<UpsertLocationError> for ApiError {
228 0 : fn from(e: UpsertLocationError) -> ApiError {
229 0 : use UpsertLocationError::*;
230 0 : match e {
231 0 : BadRequest(e) => ApiError::BadRequest(e),
232 0 : Unavailable(_) => ApiError::ShuttingDown,
233 0 : e @ InProgress => ApiError::Conflict(format!("{e}")),
234 0 : Flush(e) | Other(e) => ApiError::InternalServerError(e),
235 : }
236 0 : }
237 : }
238 :
239 : impl From<TenantMapError> for ApiError {
240 0 : fn from(e: TenantMapError) -> ApiError {
241 0 : use TenantMapError::*;
242 0 : match e {
243 : StillInitializing | ShuttingDown => {
244 0 : ApiError::ResourceUnavailable(format!("{e}").into())
245 0 : }
246 0 : }
247 0 : }
248 : }
249 :
250 : impl From<TenantStateError> for ApiError {
251 0 : fn from(tse: TenantStateError) -> ApiError {
252 0 : match tse {
253 : TenantStateError::IsStopping(_) => {
254 0 : ApiError::ResourceUnavailable("Tenant is stopping".into())
255 : }
256 0 : TenantStateError::SlotError(e) => e.into(),
257 0 : TenantStateError::SlotUpsertError(e) => e.into(),
258 0 : TenantStateError::Other(e) => ApiError::InternalServerError(anyhow!(e)),
259 : }
260 0 : }
261 : }
262 :
263 : impl From<GetTenantError> for ApiError {
264 0 : fn from(tse: GetTenantError) -> ApiError {
265 0 : match tse {
266 0 : GetTenantError::NotFound(tid) => ApiError::NotFound(anyhow!("tenant {}", tid).into()),
267 : GetTenantError::NotActive(_) => {
268 : // Why is this not `ApiError::NotFound`?
269 : // Because we must be careful to never return 404 for a tenant if it does
270 : // in fact exist locally. If we did, the caller could draw the conclusion
271 : // that it can attach the tenant to another PS and we'd be in split-brain.
272 0 : ApiError::ResourceUnavailable("Tenant not yet active".into())
273 : }
274 0 : GetTenantError::MapState(e) => ApiError::ResourceUnavailable(format!("{e}").into()),
275 : }
276 0 : }
277 : }
278 :
279 : impl From<GetTimelineError> for ApiError {
280 0 : fn from(gte: GetTimelineError) -> Self {
281 0 : // Rationale: tenant is activated only after eligble timelines activate
282 0 : ApiError::NotFound(gte.into())
283 0 : }
284 : }
285 :
286 : impl From<GetActiveTenantError> for ApiError {
287 0 : fn from(e: GetActiveTenantError) -> ApiError {
288 0 : match e {
289 0 : GetActiveTenantError::Broken(reason) => {
290 0 : ApiError::InternalServerError(anyhow!("tenant is broken: {}", reason))
291 : }
292 0 : GetActiveTenantError::WillNotBecomeActive(_) => ApiError::Conflict(format!("{}", e)),
293 0 : GetActiveTenantError::Cancelled => ApiError::ShuttingDown,
294 0 : GetActiveTenantError::NotFound(gte) => gte.into(),
295 : GetActiveTenantError::WaitForActiveTimeout { .. } => {
296 0 : ApiError::ResourceUnavailable(format!("{}", e).into())
297 : }
298 : }
299 0 : }
300 : }
301 :
302 : impl From<crate::tenant::DeleteTimelineError> for ApiError {
303 0 : fn from(value: crate::tenant::DeleteTimelineError) -> Self {
304 0 : use crate::tenant::DeleteTimelineError::*;
305 0 : match value {
306 0 : NotFound => ApiError::NotFound(anyhow::anyhow!("timeline not found").into()),
307 0 : HasChildren(children) => ApiError::PreconditionFailed(
308 0 : format!("Cannot delete timeline which has child timelines: {children:?}")
309 0 : .into_boxed_str(),
310 0 : ),
311 0 : a @ AlreadyInProgress(_) => ApiError::Conflict(a.to_string()),
312 0 : Other(e) => ApiError::InternalServerError(e),
313 : }
314 0 : }
315 : }
316 :
317 : impl From<crate::tenant::mgr::DeleteTimelineError> for ApiError {
318 0 : fn from(value: crate::tenant::mgr::DeleteTimelineError) -> Self {
319 : use crate::tenant::mgr::DeleteTimelineError::*;
320 0 : match value {
321 : // Report Precondition failed so client can distinguish between
322 : // "tenant is missing" case from "timeline is missing"
323 0 : Tenant(GetTenantError::NotFound(..)) => ApiError::PreconditionFailed(
324 0 : "Requested tenant is missing".to_owned().into_boxed_str(),
325 0 : ),
326 0 : Tenant(t) => ApiError::from(t),
327 0 : Timeline(t) => ApiError::from(t),
328 : }
329 0 : }
330 : }
331 :
332 : impl From<crate::tenant::delete::DeleteTenantError> for ApiError {
333 0 : fn from(value: crate::tenant::delete::DeleteTenantError) -> Self {
334 0 : use crate::tenant::delete::DeleteTenantError::*;
335 0 : match value {
336 0 : Get(g) => ApiError::from(g),
337 0 : Timeline(t) => ApiError::from(t),
338 0 : SlotError(e) => e.into(),
339 0 : SlotUpsertError(e) => e.into(),
340 0 : Other(o) => ApiError::InternalServerError(o),
341 0 : Cancelled => ApiError::ShuttingDown,
342 : }
343 0 : }
344 : }
345 :
346 : // Helper function to construct a TimelineInfo struct for a timeline
347 0 : async fn build_timeline_info(
348 0 : timeline: &Arc<Timeline>,
349 0 : include_non_incremental_logical_size: bool,
350 0 : force_await_initial_logical_size: bool,
351 0 : ctx: &RequestContext,
352 0 : ) -> anyhow::Result<TimelineInfo> {
353 0 : crate::tenant::debug_assert_current_span_has_tenant_and_timeline_id();
354 0 :
355 0 : if force_await_initial_logical_size {
356 0 : timeline.clone().await_initial_logical_size().await
357 0 : }
358 :
359 0 : let mut info = build_timeline_info_common(
360 0 : timeline,
361 0 : ctx,
362 0 : tenant::timeline::GetLogicalSizePriority::Background,
363 0 : )
364 0 : .await?;
365 0 : if include_non_incremental_logical_size {
366 : // XXX we should be using spawn_ondemand_logical_size_calculation here.
367 : // Otherwise, if someone deletes the timeline / detaches the tenant while
368 : // we're executing this function, we will outlive the timeline on-disk state.
369 : info.current_logical_size_non_incremental = Some(
370 0 : timeline
371 0 : .get_current_logical_size_non_incremental(info.last_record_lsn, ctx)
372 0 : .await?,
373 : );
374 0 : }
375 0 : Ok(info)
376 0 : }
377 :
378 0 : async fn build_timeline_info_common(
379 0 : timeline: &Arc<Timeline>,
380 0 : ctx: &RequestContext,
381 0 : logical_size_task_priority: tenant::timeline::GetLogicalSizePriority,
382 0 : ) -> anyhow::Result<TimelineInfo> {
383 0 : crate::tenant::debug_assert_current_span_has_tenant_and_timeline_id();
384 0 : let initdb_lsn = timeline.initdb_lsn;
385 0 : let last_record_lsn = timeline.get_last_record_lsn();
386 0 : let (wal_source_connstr, last_received_msg_lsn, last_received_msg_ts) = {
387 0 : let guard = timeline.last_received_wal.lock().unwrap();
388 0 : if let Some(info) = guard.as_ref() {
389 0 : (
390 0 : Some(format!("{}", info.wal_source_connconf)), // Password is hidden, but it's for statistics only.
391 0 : Some(info.last_received_msg_lsn),
392 0 : Some(info.last_received_msg_ts),
393 0 : )
394 : } else {
395 0 : (None, None, None)
396 : }
397 : };
398 :
399 0 : let ancestor_timeline_id = timeline.get_ancestor_timeline_id();
400 0 : let ancestor_lsn = match timeline.get_ancestor_lsn() {
401 0 : Lsn(0) => None,
402 0 : lsn @ Lsn(_) => Some(lsn),
403 : };
404 0 : let current_logical_size = timeline.get_current_logical_size(logical_size_task_priority, ctx);
405 0 : let current_physical_size = Some(timeline.layer_size_sum().await);
406 0 : let state = timeline.current_state();
407 0 : let remote_consistent_lsn_projected = timeline
408 0 : .get_remote_consistent_lsn_projected()
409 0 : .unwrap_or(Lsn(0));
410 0 : let remote_consistent_lsn_visible = timeline
411 0 : .get_remote_consistent_lsn_visible()
412 0 : .unwrap_or(Lsn(0));
413 0 :
414 0 : let walreceiver_status = timeline.walreceiver_status();
415 :
416 0 : let info = TimelineInfo {
417 0 : tenant_id: timeline.tenant_shard_id,
418 0 : timeline_id: timeline.timeline_id,
419 0 : ancestor_timeline_id,
420 0 : ancestor_lsn,
421 0 : disk_consistent_lsn: timeline.get_disk_consistent_lsn(),
422 0 : remote_consistent_lsn: remote_consistent_lsn_projected,
423 0 : remote_consistent_lsn_visible,
424 0 : initdb_lsn,
425 0 : last_record_lsn,
426 0 : prev_record_lsn: Some(timeline.get_prev_record_lsn()),
427 0 : latest_gc_cutoff_lsn: *timeline.get_latest_gc_cutoff_lsn(),
428 0 : current_logical_size: current_logical_size.size_dont_care_about_accuracy(),
429 0 : current_logical_size_is_accurate: match current_logical_size.accuracy() {
430 0 : tenant::timeline::logical_size::Accuracy::Approximate => false,
431 0 : tenant::timeline::logical_size::Accuracy::Exact => true,
432 : },
433 0 : directory_entries_counts: timeline.get_directory_metrics().to_vec(),
434 0 : current_physical_size,
435 0 : current_logical_size_non_incremental: None,
436 0 : timeline_dir_layer_file_size_sum: None,
437 0 : wal_source_connstr,
438 0 : last_received_msg_lsn,
439 0 : last_received_msg_ts,
440 0 : pg_version: timeline.pg_version,
441 0 :
442 0 : state,
443 0 :
444 0 : walreceiver_status,
445 0 :
446 0 : last_aux_file_policy: timeline.last_aux_file_policy.load(),
447 0 : };
448 0 : Ok(info)
449 0 : }
450 :
451 : // healthcheck handler
452 0 : async fn status_handler(
453 0 : request: Request<Body>,
454 0 : _cancel: CancellationToken,
455 0 : ) -> Result<Response<Body>, ApiError> {
456 0 : check_permission(&request, None)?;
457 0 : let config = get_config(&request);
458 0 : json_response(StatusCode::OK, StatusResponse { id: config.id })
459 0 : }
460 :
461 0 : async fn reload_auth_validation_keys_handler(
462 0 : request: Request<Body>,
463 0 : _cancel: CancellationToken,
464 0 : ) -> Result<Response<Body>, ApiError> {
465 0 : check_permission(&request, None)?;
466 0 : let config = get_config(&request);
467 0 : let state = get_state(&request);
468 0 : let Some(shared_auth) = &state.auth else {
469 0 : return json_response(StatusCode::BAD_REQUEST, ());
470 : };
471 : // unwrap is ok because check is performed when creating config, so path is set and exists
472 0 : let key_path = config.auth_validation_public_key_path.as_ref().unwrap();
473 0 : info!("Reloading public key(s) for verifying JWT tokens from {key_path:?}");
474 :
475 0 : match JwtAuth::from_key_path(key_path) {
476 0 : Ok(new_auth) => {
477 0 : shared_auth.swap(new_auth);
478 0 : json_response(StatusCode::OK, ())
479 : }
480 0 : Err(e) => {
481 0 : let err_msg = "Error reloading public keys";
482 0 : warn!("Error reloading public keys from {key_path:?}: {e:}");
483 0 : json_response(
484 0 : StatusCode::INTERNAL_SERVER_ERROR,
485 0 : HttpErrorBody::from_msg(err_msg.to_string()),
486 0 : )
487 : }
488 : }
489 0 : }
490 :
491 0 : async fn timeline_create_handler(
492 0 : mut request: Request<Body>,
493 0 : _cancel: CancellationToken,
494 0 : ) -> Result<Response<Body>, ApiError> {
495 0 : let tenant_shard_id: TenantShardId = parse_request_param(&request, "tenant_shard_id")?;
496 0 : let request_data: TimelineCreateRequest = json_request(&mut request).await?;
497 0 : check_permission(&request, Some(tenant_shard_id.tenant_id))?;
498 :
499 0 : let new_timeline_id = request_data.new_timeline_id;
500 0 :
501 0 : let ctx = RequestContext::new(TaskKind::MgmtRequest, DownloadBehavior::Error);
502 0 :
503 0 : let state = get_state(&request);
504 :
505 0 : async {
506 0 : let tenant = state
507 0 : .tenant_manager
508 0 : .get_attached_tenant_shard(tenant_shard_id)?;
509 :
510 0 : tenant.wait_to_become_active(ACTIVE_TENANT_TIMEOUT).await?;
511 :
512 0 : if let Some(ancestor_id) = request_data.ancestor_timeline_id.as_ref() {
513 0 : tracing::info!(%ancestor_id, "starting to branch");
514 : } else {
515 0 : tracing::info!("bootstrapping");
516 : }
517 :
518 0 : match tenant
519 0 : .create_timeline(
520 0 : new_timeline_id,
521 0 : request_data.ancestor_timeline_id,
522 0 : request_data.ancestor_start_lsn,
523 0 : request_data.pg_version.unwrap_or(crate::DEFAULT_PG_VERSION),
524 0 : request_data.existing_initdb_timeline_id,
525 0 : state.broker_client.clone(),
526 0 : &ctx,
527 0 : )
528 0 : .await
529 : {
530 0 : Ok(new_timeline) => {
531 : // Created. Construct a TimelineInfo for it.
532 0 : let timeline_info = build_timeline_info_common(
533 0 : &new_timeline,
534 0 : &ctx,
535 0 : tenant::timeline::GetLogicalSizePriority::User,
536 0 : )
537 0 : .await
538 0 : .map_err(ApiError::InternalServerError)?;
539 0 : json_response(StatusCode::CREATED, timeline_info)
540 : }
541 0 : Err(_) if tenant.cancel.is_cancelled() => {
542 0 : // In case we get some ugly error type during shutdown, cast it into a clean 503.
543 0 : json_response(
544 0 : StatusCode::SERVICE_UNAVAILABLE,
545 0 : HttpErrorBody::from_msg("Tenant shutting down".to_string()),
546 0 : )
547 : }
548 0 : Err(e @ tenant::CreateTimelineError::Conflict) => {
549 0 : json_response(StatusCode::CONFLICT, HttpErrorBody::from_msg(e.to_string()))
550 : }
551 0 : Err(e @ tenant::CreateTimelineError::AlreadyCreating) => json_response(
552 0 : StatusCode::TOO_MANY_REQUESTS,
553 0 : HttpErrorBody::from_msg(e.to_string()),
554 0 : ),
555 0 : Err(tenant::CreateTimelineError::AncestorLsn(err)) => json_response(
556 0 : StatusCode::NOT_ACCEPTABLE,
557 0 : HttpErrorBody::from_msg(format!("{err:#}")),
558 0 : ),
559 0 : Err(e @ tenant::CreateTimelineError::AncestorNotActive) => json_response(
560 0 : StatusCode::SERVICE_UNAVAILABLE,
561 0 : HttpErrorBody::from_msg(e.to_string()),
562 0 : ),
563 0 : Err(tenant::CreateTimelineError::ShuttingDown) => json_response(
564 0 : StatusCode::SERVICE_UNAVAILABLE,
565 0 : HttpErrorBody::from_msg("tenant shutting down".to_string()),
566 0 : ),
567 0 : Err(tenant::CreateTimelineError::Other(err)) => Err(ApiError::InternalServerError(err)),
568 : }
569 0 : }
570 0 : .instrument(info_span!("timeline_create",
571 : tenant_id = %tenant_shard_id.tenant_id,
572 0 : shard_id = %tenant_shard_id.shard_slug(),
573 : timeline_id = %new_timeline_id,
574 : lsn=?request_data.ancestor_start_lsn,
575 : pg_version=?request_data.pg_version
576 : ))
577 0 : .await
578 0 : }
579 :
580 0 : async fn timeline_list_handler(
581 0 : request: Request<Body>,
582 0 : _cancel: CancellationToken,
583 0 : ) -> Result<Response<Body>, ApiError> {
584 0 : let tenant_shard_id: TenantShardId = parse_request_param(&request, "tenant_shard_id")?;
585 0 : let include_non_incremental_logical_size: Option<bool> =
586 0 : parse_query_param(&request, "include-non-incremental-logical-size")?;
587 0 : let force_await_initial_logical_size: Option<bool> =
588 0 : parse_query_param(&request, "force-await-initial-logical-size")?;
589 0 : check_permission(&request, Some(tenant_shard_id.tenant_id))?;
590 :
591 0 : let state = get_state(&request);
592 0 : let ctx = RequestContext::new(TaskKind::MgmtRequest, DownloadBehavior::Download);
593 :
594 0 : let response_data = async {
595 0 : let tenant = state
596 0 : .tenant_manager
597 0 : .get_attached_tenant_shard(tenant_shard_id)?;
598 :
599 0 : tenant.wait_to_become_active(ACTIVE_TENANT_TIMEOUT).await?;
600 :
601 0 : let timelines = tenant.list_timelines();
602 0 :
603 0 : let mut response_data = Vec::with_capacity(timelines.len());
604 0 : for timeline in timelines {
605 0 : let timeline_info = build_timeline_info(
606 0 : &timeline,
607 0 : include_non_incremental_logical_size.unwrap_or(false),
608 0 : force_await_initial_logical_size.unwrap_or(false),
609 0 : &ctx,
610 0 : )
611 0 : .instrument(info_span!("build_timeline_info", timeline_id = %timeline.timeline_id))
612 0 : .await
613 0 : .context("Failed to convert tenant timeline {timeline_id} into the local one: {e:?}")
614 0 : .map_err(ApiError::InternalServerError)?;
615 :
616 0 : response_data.push(timeline_info);
617 : }
618 0 : Ok::<Vec<TimelineInfo>, ApiError>(response_data)
619 0 : }
620 0 : .instrument(info_span!("timeline_list",
621 : tenant_id = %tenant_shard_id.tenant_id,
622 0 : shard_id = %tenant_shard_id.shard_slug()))
623 0 : .await?;
624 :
625 0 : json_response(StatusCode::OK, response_data)
626 0 : }
627 :
628 0 : async fn timeline_preserve_initdb_handler(
629 0 : request: Request<Body>,
630 0 : _cancel: CancellationToken,
631 0 : ) -> Result<Response<Body>, ApiError> {
632 0 : let tenant_shard_id: TenantShardId = parse_request_param(&request, "tenant_shard_id")?;
633 0 : let timeline_id: TimelineId = parse_request_param(&request, "timeline_id")?;
634 0 : check_permission(&request, Some(tenant_shard_id.tenant_id))?;
635 0 : let state = get_state(&request);
636 :
637 : // Part of the process for disaster recovery from safekeeper-stored WAL:
638 : // If we don't recover into a new timeline but want to keep the timeline ID,
639 : // then the initdb archive is deleted. This endpoint copies it to a different
640 : // location where timeline recreation cand find it.
641 :
642 0 : async {
643 0 : let tenant = state
644 0 : .tenant_manager
645 0 : .get_attached_tenant_shard(tenant_shard_id)?;
646 :
647 0 : let timeline = tenant.get_timeline(timeline_id, false)?;
648 :
649 0 : timeline
650 0 : .preserve_initdb_archive()
651 0 : .await
652 0 : .context("preserving initdb archive")
653 0 : .map_err(ApiError::InternalServerError)?;
654 :
655 0 : Ok::<_, ApiError>(())
656 0 : }
657 0 : .instrument(info_span!("timeline_preserve_initdb_archive",
658 : tenant_id = %tenant_shard_id.tenant_id,
659 0 : shard_id = %tenant_shard_id.shard_slug(),
660 : %timeline_id))
661 0 : .await?;
662 :
663 0 : json_response(StatusCode::OK, ())
664 0 : }
665 :
666 0 : async fn timeline_detail_handler(
667 0 : request: Request<Body>,
668 0 : _cancel: CancellationToken,
669 0 : ) -> Result<Response<Body>, ApiError> {
670 0 : let tenant_shard_id: TenantShardId = parse_request_param(&request, "tenant_shard_id")?;
671 0 : let timeline_id: TimelineId = parse_request_param(&request, "timeline_id")?;
672 0 : let include_non_incremental_logical_size: Option<bool> =
673 0 : parse_query_param(&request, "include-non-incremental-logical-size")?;
674 0 : let force_await_initial_logical_size: Option<bool> =
675 0 : parse_query_param(&request, "force-await-initial-logical-size")?;
676 0 : check_permission(&request, Some(tenant_shard_id.tenant_id))?;
677 :
678 : // Logical size calculation needs downloading.
679 0 : let ctx = RequestContext::new(TaskKind::MgmtRequest, DownloadBehavior::Download);
680 0 : let state = get_state(&request);
681 :
682 0 : let timeline_info = async {
683 0 : let tenant = state
684 0 : .tenant_manager
685 0 : .get_attached_tenant_shard(tenant_shard_id)?;
686 :
687 0 : tenant.wait_to_become_active(ACTIVE_TENANT_TIMEOUT).await?;
688 :
689 0 : let timeline = tenant.get_timeline(timeline_id, false)?;
690 :
691 0 : let timeline_info = build_timeline_info(
692 0 : &timeline,
693 0 : include_non_incremental_logical_size.unwrap_or(false),
694 0 : force_await_initial_logical_size.unwrap_or(false),
695 0 : &ctx,
696 0 : )
697 0 : .await
698 0 : .context("get local timeline info")
699 0 : .map_err(ApiError::InternalServerError)?;
700 :
701 0 : Ok::<_, ApiError>(timeline_info)
702 0 : }
703 0 : .instrument(info_span!("timeline_detail",
704 : tenant_id = %tenant_shard_id.tenant_id,
705 0 : shard_id = %tenant_shard_id.shard_slug(),
706 : %timeline_id))
707 0 : .await?;
708 :
709 0 : json_response(StatusCode::OK, timeline_info)
710 0 : }
711 :
712 0 : async fn get_lsn_by_timestamp_handler(
713 0 : request: Request<Body>,
714 0 : cancel: CancellationToken,
715 0 : ) -> Result<Response<Body>, ApiError> {
716 0 : let tenant_shard_id: TenantShardId = parse_request_param(&request, "tenant_shard_id")?;
717 0 : check_permission(&request, Some(tenant_shard_id.tenant_id))?;
718 0 : let state = get_state(&request);
719 0 :
720 0 : if !tenant_shard_id.is_shard_zero() {
721 : // Requires SLRU contents, which are only stored on shard zero
722 0 : return Err(ApiError::BadRequest(anyhow!(
723 0 : "Size calculations are only available on shard zero"
724 0 : )));
725 0 : }
726 :
727 0 : let timeline_id: TimelineId = parse_request_param(&request, "timeline_id")?;
728 0 : let timestamp_raw = must_get_query_param(&request, "timestamp")?;
729 0 : let timestamp = humantime::parse_rfc3339(×tamp_raw)
730 0 : .with_context(|| format!("Invalid time: {:?}", timestamp_raw))
731 0 : .map_err(ApiError::BadRequest)?;
732 0 : let timestamp_pg = postgres_ffi::to_pg_timestamp(timestamp);
733 0 :
734 0 : let ctx = RequestContext::new(TaskKind::MgmtRequest, DownloadBehavior::Download);
735 :
736 0 : let timeline =
737 0 : active_timeline_of_active_tenant(&state.tenant_manager, tenant_shard_id, timeline_id)
738 0 : .await?;
739 0 : let result = timeline
740 0 : .find_lsn_for_timestamp(timestamp_pg, &cancel, &ctx)
741 0 : .await?;
742 : #[derive(serde::Serialize, Debug)]
743 : struct Result {
744 : lsn: Lsn,
745 : kind: &'static str,
746 : }
747 0 : let (lsn, kind) = match result {
748 0 : LsnForTimestamp::Present(lsn) => (lsn, "present"),
749 0 : LsnForTimestamp::Future(lsn) => (lsn, "future"),
750 0 : LsnForTimestamp::Past(lsn) => (lsn, "past"),
751 0 : LsnForTimestamp::NoData(lsn) => (lsn, "nodata"),
752 : };
753 0 : let result = Result { lsn, kind };
754 0 : tracing::info!(
755 : lsn=?result.lsn,
756 : kind=%result.kind,
757 : timestamp=%timestamp_raw,
758 0 : "lsn_by_timestamp finished"
759 : );
760 0 : json_response(StatusCode::OK, result)
761 0 : }
762 :
763 0 : async fn get_timestamp_of_lsn_handler(
764 0 : request: Request<Body>,
765 0 : _cancel: CancellationToken,
766 0 : ) -> Result<Response<Body>, ApiError> {
767 0 : let tenant_shard_id: TenantShardId = parse_request_param(&request, "tenant_shard_id")?;
768 0 : check_permission(&request, Some(tenant_shard_id.tenant_id))?;
769 0 : let state = get_state(&request);
770 0 :
771 0 : if !tenant_shard_id.is_shard_zero() {
772 : // Requires SLRU contents, which are only stored on shard zero
773 0 : return Err(ApiError::BadRequest(anyhow!(
774 0 : "Size calculations are only available on shard zero"
775 0 : )));
776 0 : }
777 :
778 0 : let timeline_id: TimelineId = parse_request_param(&request, "timeline_id")?;
779 :
780 0 : let lsn_str = must_get_query_param(&request, "lsn")?;
781 0 : let lsn = Lsn::from_str(&lsn_str)
782 0 : .with_context(|| format!("Invalid LSN: {lsn_str:?}"))
783 0 : .map_err(ApiError::BadRequest)?;
784 :
785 0 : let ctx = RequestContext::new(TaskKind::MgmtRequest, DownloadBehavior::Download);
786 0 : let timeline =
787 0 : active_timeline_of_active_tenant(&state.tenant_manager, tenant_shard_id, timeline_id)
788 0 : .await?;
789 0 : let result = timeline.get_timestamp_for_lsn(lsn, &ctx).await?;
790 :
791 0 : match result {
792 0 : Some(time) => {
793 0 : let time = format_rfc3339(postgres_ffi::from_pg_timestamp(time)).to_string();
794 0 : json_response(StatusCode::OK, time)
795 : }
796 0 : None => Err(ApiError::NotFound(
797 0 : anyhow::anyhow!("Timestamp for lsn {} not found", lsn).into(),
798 0 : )),
799 : }
800 0 : }
801 :
802 0 : async fn tenant_attach_handler(
803 0 : mut request: Request<Body>,
804 0 : _cancel: CancellationToken,
805 0 : ) -> Result<Response<Body>, ApiError> {
806 0 : let tenant_id: TenantId = parse_request_param(&request, "tenant_id")?;
807 0 : check_permission(&request, Some(tenant_id))?;
808 :
809 0 : let maybe_body: Option<TenantAttachRequest> = json_request_or_empty_body(&mut request).await?;
810 0 : let tenant_conf = match &maybe_body {
811 0 : Some(request) => TenantConfOpt::try_from(&*request.config).map_err(ApiError::BadRequest)?,
812 0 : None => TenantConfOpt::default(),
813 : };
814 :
815 0 : let ctx = RequestContext::new(TaskKind::MgmtRequest, DownloadBehavior::Warn);
816 0 :
817 0 : info!("Handling tenant attach {tenant_id}");
818 :
819 0 : let state = get_state(&request);
820 :
821 0 : let generation = get_request_generation(state, maybe_body.as_ref().and_then(|r| r.generation))?;
822 :
823 0 : let tenant_shard_id = TenantShardId::unsharded(tenant_id);
824 0 : let shard_params = ShardParameters::default();
825 0 : let location_conf = LocationConf::attached_single(tenant_conf, generation, &shard_params);
826 :
827 0 : let tenant = state
828 0 : .tenant_manager
829 0 : .upsert_location(tenant_shard_id, location_conf, None, SpawnMode::Eager, &ctx)
830 0 : .await?;
831 :
832 0 : let Some(tenant) = tenant else {
833 : // This should never happen: indicates a bug in upsert_location
834 0 : return Err(ApiError::InternalServerError(anyhow::anyhow!(
835 0 : "Upsert succeeded but didn't return tenant!"
836 0 : )));
837 : };
838 :
839 : // We might have successfully constructed a Tenant, but it could still
840 : // end up in a broken state:
841 : if let TenantState::Broken {
842 0 : reason,
843 : backtrace: _,
844 0 : } = tenant.current_state()
845 : {
846 0 : return Err(ApiError::InternalServerError(anyhow::anyhow!(
847 0 : "Tenant state is Broken: {reason}"
848 0 : )));
849 0 : }
850 0 :
851 0 : json_response(StatusCode::ACCEPTED, ())
852 0 : }
853 :
854 0 : async fn timeline_delete_handler(
855 0 : request: Request<Body>,
856 0 : _cancel: CancellationToken,
857 0 : ) -> Result<Response<Body>, ApiError> {
858 0 : let tenant_shard_id: TenantShardId = parse_request_param(&request, "tenant_shard_id")?;
859 0 : let timeline_id: TimelineId = parse_request_param(&request, "timeline_id")?;
860 0 : check_permission(&request, Some(tenant_shard_id.tenant_id))?;
861 :
862 0 : let state = get_state(&request);
863 :
864 0 : let tenant = state
865 0 : .tenant_manager
866 0 : .get_attached_tenant_shard(tenant_shard_id)
867 0 : .map_err(|e| {
868 0 : match e {
869 : // GetTenantError has a built-in conversion to ApiError, but in this context we don't
870 : // want to treat missing tenants as 404, to avoid ambiguity with successful deletions.
871 0 : GetTenantError::NotFound(_) => ApiError::PreconditionFailed(
872 0 : "Requested tenant is missing".to_string().into_boxed_str(),
873 0 : ),
874 0 : e => e.into(),
875 : }
876 0 : })?;
877 0 : tenant.wait_to_become_active(ACTIVE_TENANT_TIMEOUT).await?;
878 0 : tenant.delete_timeline(timeline_id).instrument(info_span!("timeline_delete", tenant_id=%tenant_shard_id.tenant_id, shard_id=%tenant_shard_id.shard_slug(), %timeline_id))
879 0 : .await?;
880 :
881 0 : json_response(StatusCode::ACCEPTED, ())
882 0 : }
883 :
884 0 : async fn tenant_detach_handler(
885 0 : request: Request<Body>,
886 0 : _cancel: CancellationToken,
887 0 : ) -> Result<Response<Body>, ApiError> {
888 0 : let tenant_id: TenantId = parse_request_param(&request, "tenant_id")?;
889 0 : check_permission(&request, Some(tenant_id))?;
890 : // This is a legacy API (`/location_conf` is the replacement). It only supports unsharded tenants
891 0 : let tenant_shard_id = TenantShardId::unsharded(tenant_id);
892 0 :
893 0 : let state = get_state(&request);
894 0 : let conf = state.conf;
895 0 : state
896 0 : .tenant_manager
897 0 : .detach_tenant(conf, tenant_shard_id, &state.deletion_queue_client)
898 0 : .instrument(info_span!("tenant_detach", %tenant_id, shard_id=%tenant_shard_id.shard_slug()))
899 0 : .await?;
900 :
901 0 : json_response(StatusCode::OK, ())
902 0 : }
903 :
904 0 : async fn tenant_reset_handler(
905 0 : request: Request<Body>,
906 0 : _cancel: CancellationToken,
907 0 : ) -> Result<Response<Body>, ApiError> {
908 0 : let tenant_shard_id: TenantShardId = parse_request_param(&request, "tenant_shard_id")?;
909 0 : check_permission(&request, Some(tenant_shard_id.tenant_id))?;
910 :
911 0 : let drop_cache: Option<bool> = parse_query_param(&request, "drop_cache")?;
912 :
913 0 : let ctx = RequestContext::new(TaskKind::MgmtRequest, DownloadBehavior::Warn);
914 0 : let state = get_state(&request);
915 0 : state
916 0 : .tenant_manager
917 0 : .reset_tenant(tenant_shard_id, drop_cache.unwrap_or(false), &ctx)
918 0 : .await
919 0 : .map_err(ApiError::InternalServerError)?;
920 :
921 0 : json_response(StatusCode::OK, ())
922 0 : }
923 :
924 0 : async fn tenant_list_handler(
925 0 : request: Request<Body>,
926 0 : _cancel: CancellationToken,
927 0 : ) -> Result<Response<Body>, ApiError> {
928 0 : check_permission(&request, None)?;
929 0 : let state = get_state(&request);
930 :
931 0 : let response_data = state
932 0 : .tenant_manager
933 0 : .list_tenants()
934 0 : .map_err(|_| {
935 0 : ApiError::ResourceUnavailable("Tenant map is initializing or shutting down".into())
936 0 : })?
937 0 : .iter()
938 0 : .map(|(id, state, gen)| TenantInfo {
939 0 : id: *id,
940 0 : state: state.clone(),
941 0 : current_physical_size: None,
942 0 : attachment_status: state.attachment_status(),
943 0 : generation: (*gen).into(),
944 0 : })
945 0 : .collect::<Vec<TenantInfo>>();
946 0 :
947 0 : json_response(StatusCode::OK, response_data)
948 0 : }
949 :
950 0 : async fn tenant_status(
951 0 : request: Request<Body>,
952 0 : _cancel: CancellationToken,
953 0 : ) -> Result<Response<Body>, ApiError> {
954 0 : let tenant_shard_id: TenantShardId = parse_request_param(&request, "tenant_shard_id")?;
955 0 : check_permission(&request, Some(tenant_shard_id.tenant_id))?;
956 0 : let state = get_state(&request);
957 0 :
958 0 : // In tests, sometimes we want to query the state of a tenant without auto-activating it if it's currently waiting.
959 0 : let activate = true;
960 : #[cfg(feature = "testing")]
961 0 : let activate = parse_query_param(&request, "activate")?.unwrap_or(activate);
962 :
963 0 : let tenant_info = async {
964 0 : let tenant = state
965 0 : .tenant_manager
966 0 : .get_attached_tenant_shard(tenant_shard_id)?;
967 :
968 0 : if activate {
969 : // This is advisory: we prefer to let the tenant activate on-demand when this function is
970 : // called, but it is still valid to return 200 and describe the current state of the tenant
971 : // if it doesn't make it into an active state.
972 0 : tenant
973 0 : .wait_to_become_active(ACTIVE_TENANT_TIMEOUT)
974 0 : .await
975 0 : .ok();
976 0 : }
977 :
978 : // Calculate total physical size of all timelines
979 0 : let mut current_physical_size = 0;
980 0 : for timeline in tenant.list_timelines().iter() {
981 0 : current_physical_size += timeline.layer_size_sum().await;
982 : }
983 :
984 0 : let state = tenant.current_state();
985 0 : Result::<_, ApiError>::Ok(TenantDetails {
986 0 : tenant_info: TenantInfo {
987 0 : id: tenant_shard_id,
988 0 : state: state.clone(),
989 0 : current_physical_size: Some(current_physical_size),
990 0 : attachment_status: state.attachment_status(),
991 0 : generation: tenant.generation().into(),
992 0 : },
993 0 : walredo: tenant.wal_redo_manager_status(),
994 0 : timelines: tenant.list_timeline_ids(),
995 0 : })
996 0 : }
997 0 : .instrument(info_span!("tenant_status_handler",
998 : tenant_id = %tenant_shard_id.tenant_id,
999 0 : shard_id = %tenant_shard_id.shard_slug()))
1000 0 : .await?;
1001 :
1002 0 : json_response(StatusCode::OK, tenant_info)
1003 0 : }
1004 :
1005 0 : async fn tenant_delete_handler(
1006 0 : request: Request<Body>,
1007 0 : _cancel: CancellationToken,
1008 0 : ) -> Result<Response<Body>, ApiError> {
1009 : // TODO openapi spec
1010 0 : let tenant_shard_id: TenantShardId = parse_request_param(&request, "tenant_shard_id")?;
1011 0 : check_permission(&request, Some(tenant_shard_id.tenant_id))?;
1012 :
1013 0 : let state = get_state(&request);
1014 0 :
1015 0 : state
1016 0 : .tenant_manager
1017 0 : .delete_tenant(tenant_shard_id)
1018 0 : .instrument(info_span!("tenant_delete_handler",
1019 : tenant_id = %tenant_shard_id.tenant_id,
1020 0 : shard_id = %tenant_shard_id.shard_slug()
1021 : ))
1022 0 : .await?;
1023 :
1024 0 : json_response(StatusCode::OK, ())
1025 0 : }
1026 :
1027 : /// HTTP endpoint to query the current tenant_size of a tenant.
1028 : ///
1029 : /// This is not used by consumption metrics under [`crate::consumption_metrics`], but can be used
1030 : /// to debug any of the calculations. Requires `tenant_id` request parameter, supports
1031 : /// `inputs_only=true|false` (default false) which supports debugging failure to calculate model
1032 : /// values.
1033 : ///
1034 : /// 'retention_period' query parameter overrides the cutoff that is used to calculate the size
1035 : /// (only if it is shorter than the real cutoff).
1036 : ///
1037 : /// Note: we don't update the cached size and prometheus metric here.
1038 : /// The retention period might be different, and it's nice to have a method to just calculate it
1039 : /// without modifying anything anyway.
1040 0 : async fn tenant_size_handler(
1041 0 : request: Request<Body>,
1042 0 : cancel: CancellationToken,
1043 0 : ) -> Result<Response<Body>, ApiError> {
1044 0 : let tenant_shard_id: TenantShardId = parse_request_param(&request, "tenant_shard_id")?;
1045 0 : check_permission(&request, Some(tenant_shard_id.tenant_id))?;
1046 0 : let inputs_only: Option<bool> = parse_query_param(&request, "inputs_only")?;
1047 0 : let retention_period: Option<u64> = parse_query_param(&request, "retention_period")?;
1048 0 : let headers = request.headers();
1049 0 : let state = get_state(&request);
1050 0 :
1051 0 : if !tenant_shard_id.is_shard_zero() {
1052 0 : return Err(ApiError::BadRequest(anyhow!(
1053 0 : "Size calculations are only available on shard zero"
1054 0 : )));
1055 0 : }
1056 0 :
1057 0 : let ctx = RequestContext::new(TaskKind::MgmtRequest, DownloadBehavior::Download);
1058 0 : let tenant = state
1059 0 : .tenant_manager
1060 0 : .get_attached_tenant_shard(tenant_shard_id)?;
1061 0 : tenant.wait_to_become_active(ACTIVE_TENANT_TIMEOUT).await?;
1062 :
1063 : // this can be long operation
1064 0 : let inputs = tenant
1065 0 : .gather_size_inputs(
1066 0 : retention_period,
1067 0 : LogicalSizeCalculationCause::TenantSizeHandler,
1068 0 : &cancel,
1069 0 : &ctx,
1070 0 : )
1071 0 : .await
1072 0 : .map_err(|e| match e {
1073 0 : crate::tenant::size::CalculateSyntheticSizeError::Cancelled => ApiError::ShuttingDown,
1074 0 : other => ApiError::InternalServerError(anyhow::anyhow!(other)),
1075 0 : })?;
1076 :
1077 0 : let mut sizes = None;
1078 0 : let accepts_html = headers
1079 0 : .get(header::ACCEPT)
1080 0 : .map(|v| v == "text/html")
1081 0 : .unwrap_or_default();
1082 0 : if !inputs_only.unwrap_or(false) {
1083 0 : let storage_model = inputs.calculate_model();
1084 0 : let size = storage_model.calculate();
1085 0 :
1086 0 : // If request header expects html, return html
1087 0 : if accepts_html {
1088 0 : return synthetic_size_html_response(inputs, storage_model, size);
1089 0 : }
1090 0 : sizes = Some(size);
1091 0 : } else if accepts_html {
1092 0 : return Err(ApiError::BadRequest(anyhow!(
1093 0 : "inputs_only parameter is incompatible with html output request"
1094 0 : )));
1095 0 : }
1096 :
1097 : /// The type resides in the pageserver not to expose `ModelInputs`.
1098 : #[derive(serde::Serialize)]
1099 : struct TenantHistorySize {
1100 : id: TenantId,
1101 : /// Size is a mixture of WAL and logical size, so the unit is bytes.
1102 : ///
1103 : /// Will be none if `?inputs_only=true` was given.
1104 : size: Option<u64>,
1105 : /// Size of each segment used in the model.
1106 : /// Will be null if `?inputs_only=true` was given.
1107 : segment_sizes: Option<Vec<tenant_size_model::SegmentSizeResult>>,
1108 : inputs: crate::tenant::size::ModelInputs,
1109 : }
1110 :
1111 0 : json_response(
1112 0 : StatusCode::OK,
1113 0 : TenantHistorySize {
1114 0 : id: tenant_shard_id.tenant_id,
1115 0 : size: sizes.as_ref().map(|x| x.total_size),
1116 0 : segment_sizes: sizes.map(|x| x.segments),
1117 0 : inputs,
1118 0 : },
1119 0 : )
1120 0 : }
1121 :
1122 0 : async fn tenant_shard_split_handler(
1123 0 : mut request: Request<Body>,
1124 0 : _cancel: CancellationToken,
1125 0 : ) -> Result<Response<Body>, ApiError> {
1126 0 : let req: TenantShardSplitRequest = json_request(&mut request).await?;
1127 :
1128 0 : let tenant_shard_id: TenantShardId = parse_request_param(&request, "tenant_shard_id")?;
1129 0 : let state = get_state(&request);
1130 0 : let ctx = RequestContext::new(TaskKind::MgmtRequest, DownloadBehavior::Warn);
1131 :
1132 0 : let tenant = state
1133 0 : .tenant_manager
1134 0 : .get_attached_tenant_shard(tenant_shard_id)?;
1135 0 : tenant.wait_to_become_active(ACTIVE_TENANT_TIMEOUT).await?;
1136 :
1137 0 : let new_shards = state
1138 0 : .tenant_manager
1139 0 : .shard_split(
1140 0 : tenant,
1141 0 : ShardCount::new(req.new_shard_count),
1142 0 : req.new_stripe_size,
1143 0 : &ctx,
1144 0 : )
1145 0 : .await
1146 0 : .map_err(ApiError::InternalServerError)?;
1147 :
1148 0 : json_response(StatusCode::OK, TenantShardSplitResponse { new_shards })
1149 0 : }
1150 :
1151 0 : async fn layer_map_info_handler(
1152 0 : request: Request<Body>,
1153 0 : _cancel: CancellationToken,
1154 0 : ) -> Result<Response<Body>, ApiError> {
1155 0 : let tenant_shard_id: TenantShardId = parse_request_param(&request, "tenant_shard_id")?;
1156 0 : let timeline_id: TimelineId = parse_request_param(&request, "timeline_id")?;
1157 0 : let reset: LayerAccessStatsReset =
1158 0 : parse_query_param(&request, "reset")?.unwrap_or(LayerAccessStatsReset::NoReset);
1159 0 : let state = get_state(&request);
1160 0 :
1161 0 : check_permission(&request, Some(tenant_shard_id.tenant_id))?;
1162 :
1163 0 : let timeline =
1164 0 : active_timeline_of_active_tenant(&state.tenant_manager, tenant_shard_id, timeline_id)
1165 0 : .await?;
1166 0 : let layer_map_info = timeline.layer_map_info(reset).await;
1167 :
1168 0 : json_response(StatusCode::OK, layer_map_info)
1169 0 : }
1170 :
1171 0 : async fn layer_download_handler(
1172 0 : request: Request<Body>,
1173 0 : _cancel: CancellationToken,
1174 0 : ) -> Result<Response<Body>, ApiError> {
1175 0 : let tenant_shard_id: TenantShardId = parse_request_param(&request, "tenant_shard_id")?;
1176 0 : let timeline_id: TimelineId = parse_request_param(&request, "timeline_id")?;
1177 0 : let layer_file_name = get_request_param(&request, "layer_file_name")?;
1178 0 : check_permission(&request, Some(tenant_shard_id.tenant_id))?;
1179 0 : let layer_name = LayerName::from_str(layer_file_name)
1180 0 : .map_err(|s| ApiError::BadRequest(anyhow::anyhow!(s)))?;
1181 0 : let state = get_state(&request);
1182 :
1183 0 : let timeline =
1184 0 : active_timeline_of_active_tenant(&state.tenant_manager, tenant_shard_id, timeline_id)
1185 0 : .await?;
1186 0 : let downloaded = timeline
1187 0 : .download_layer(&layer_name)
1188 0 : .await
1189 0 : .map_err(ApiError::InternalServerError)?;
1190 :
1191 0 : match downloaded {
1192 0 : Some(true) => json_response(StatusCode::OK, ()),
1193 0 : Some(false) => json_response(StatusCode::NOT_MODIFIED, ()),
1194 0 : None => json_response(
1195 0 : StatusCode::BAD_REQUEST,
1196 0 : format!("Layer {tenant_shard_id}/{timeline_id}/{layer_file_name} not found"),
1197 0 : ),
1198 : }
1199 0 : }
1200 :
1201 0 : async fn evict_timeline_layer_handler(
1202 0 : request: Request<Body>,
1203 0 : _cancel: CancellationToken,
1204 0 : ) -> Result<Response<Body>, ApiError> {
1205 0 : let tenant_shard_id: TenantShardId = parse_request_param(&request, "tenant_shard_id")?;
1206 0 : check_permission(&request, Some(tenant_shard_id.tenant_id))?;
1207 0 : let timeline_id: TimelineId = parse_request_param(&request, "timeline_id")?;
1208 0 : let layer_file_name = get_request_param(&request, "layer_file_name")?;
1209 0 : let state = get_state(&request);
1210 :
1211 0 : let layer_name = LayerName::from_str(layer_file_name)
1212 0 : .map_err(|s| ApiError::BadRequest(anyhow::anyhow!(s)))?;
1213 :
1214 0 : let timeline =
1215 0 : active_timeline_of_active_tenant(&state.tenant_manager, tenant_shard_id, timeline_id)
1216 0 : .await?;
1217 0 : let evicted = timeline
1218 0 : .evict_layer(&layer_name)
1219 0 : .await
1220 0 : .map_err(ApiError::InternalServerError)?;
1221 :
1222 0 : match evicted {
1223 0 : Some(true) => json_response(StatusCode::OK, ()),
1224 0 : Some(false) => json_response(StatusCode::NOT_MODIFIED, ()),
1225 0 : None => json_response(
1226 0 : StatusCode::BAD_REQUEST,
1227 0 : format!("Layer {tenant_shard_id}/{timeline_id}/{layer_file_name} not found"),
1228 0 : ),
1229 : }
1230 0 : }
1231 :
1232 : /// Get tenant_size SVG graph along with the JSON data.
1233 0 : fn synthetic_size_html_response(
1234 0 : inputs: ModelInputs,
1235 0 : storage_model: StorageModel,
1236 0 : sizes: SizeResult,
1237 0 : ) -> Result<Response<Body>, ApiError> {
1238 0 : let mut timeline_ids: Vec<String> = Vec::new();
1239 0 : let mut timeline_map: HashMap<TimelineId, usize> = HashMap::new();
1240 0 : for (index, ti) in inputs.timeline_inputs.iter().enumerate() {
1241 0 : timeline_map.insert(ti.timeline_id, index);
1242 0 : timeline_ids.push(ti.timeline_id.to_string());
1243 0 : }
1244 0 : let seg_to_branch: Vec<usize> = inputs
1245 0 : .segments
1246 0 : .iter()
1247 0 : .map(|seg| *timeline_map.get(&seg.timeline_id).unwrap())
1248 0 : .collect();
1249 :
1250 0 : let svg =
1251 0 : tenant_size_model::svg::draw_svg(&storage_model, &timeline_ids, &seg_to_branch, &sizes)
1252 0 : .map_err(ApiError::InternalServerError)?;
1253 :
1254 0 : let mut response = String::new();
1255 0 :
1256 0 : use std::fmt::Write;
1257 0 : write!(response, "<html>\n<body>\n").unwrap();
1258 0 : write!(response, "<div>\n{svg}\n</div>").unwrap();
1259 0 : writeln!(response, "Project size: {}", sizes.total_size).unwrap();
1260 0 : writeln!(response, "<pre>").unwrap();
1261 0 : writeln!(
1262 0 : response,
1263 0 : "{}",
1264 0 : serde_json::to_string_pretty(&inputs).unwrap()
1265 0 : )
1266 0 : .unwrap();
1267 0 : writeln!(
1268 0 : response,
1269 0 : "{}",
1270 0 : serde_json::to_string_pretty(&sizes.segments).unwrap()
1271 0 : )
1272 0 : .unwrap();
1273 0 : writeln!(response, "</pre>").unwrap();
1274 0 : write!(response, "</body>\n</html>\n").unwrap();
1275 0 :
1276 0 : html_response(StatusCode::OK, response)
1277 0 : }
1278 :
1279 0 : pub fn html_response(status: StatusCode, data: String) -> Result<Response<Body>, ApiError> {
1280 0 : let response = Response::builder()
1281 0 : .status(status)
1282 0 : .header(header::CONTENT_TYPE, "text/html")
1283 0 : .body(Body::from(data.as_bytes().to_vec()))
1284 0 : .map_err(|e| ApiError::InternalServerError(e.into()))?;
1285 0 : Ok(response)
1286 0 : }
1287 :
1288 : /// Helper for requests that may take a generation, which is mandatory
1289 : /// when control_plane_api is set, but otherwise defaults to Generation::none()
1290 0 : fn get_request_generation(state: &State, req_gen: Option<u32>) -> Result<Generation, ApiError> {
1291 0 : if state.conf.control_plane_api.is_some() {
1292 0 : req_gen
1293 0 : .map(Generation::new)
1294 0 : .ok_or(ApiError::BadRequest(anyhow!(
1295 0 : "generation attribute missing"
1296 0 : )))
1297 : } else {
1298 : // Legacy mode: all tenants operate with no generation
1299 0 : Ok(Generation::none())
1300 : }
1301 0 : }
1302 :
1303 0 : async fn tenant_create_handler(
1304 0 : mut request: Request<Body>,
1305 0 : _cancel: CancellationToken,
1306 0 : ) -> Result<Response<Body>, ApiError> {
1307 0 : let request_data: TenantCreateRequest = json_request(&mut request).await?;
1308 0 : let target_tenant_id = request_data.new_tenant_id;
1309 0 : check_permission(&request, None)?;
1310 :
1311 0 : let _timer = STORAGE_TIME_GLOBAL
1312 0 : .get_metric_with_label_values(&[StorageTimeOperation::CreateTenant.into()])
1313 0 : .expect("bug")
1314 0 : .start_timer();
1315 :
1316 0 : let tenant_conf =
1317 0 : TenantConfOpt::try_from(&request_data.config).map_err(ApiError::BadRequest)?;
1318 :
1319 0 : let state = get_state(&request);
1320 :
1321 0 : let generation = get_request_generation(state, request_data.generation)?;
1322 :
1323 0 : let ctx = RequestContext::new(TaskKind::MgmtRequest, DownloadBehavior::Warn);
1324 0 :
1325 0 : let location_conf =
1326 0 : LocationConf::attached_single(tenant_conf, generation, &request_data.shard_parameters);
1327 :
1328 0 : let new_tenant = state
1329 0 : .tenant_manager
1330 0 : .upsert_location(
1331 0 : target_tenant_id,
1332 0 : location_conf,
1333 0 : None,
1334 0 : SpawnMode::Create,
1335 0 : &ctx,
1336 0 : )
1337 0 : .await?;
1338 :
1339 0 : let Some(new_tenant) = new_tenant else {
1340 : // This should never happen: indicates a bug in upsert_location
1341 0 : return Err(ApiError::InternalServerError(anyhow::anyhow!(
1342 0 : "Upsert succeeded but didn't return tenant!"
1343 0 : )));
1344 : };
1345 : // We created the tenant. Existing API semantics are that the tenant
1346 : // is Active when this function returns.
1347 0 : new_tenant
1348 0 : .wait_to_become_active(ACTIVE_TENANT_TIMEOUT)
1349 0 : .await?;
1350 :
1351 0 : json_response(
1352 0 : StatusCode::CREATED,
1353 0 : TenantCreateResponse(new_tenant.tenant_shard_id().tenant_id),
1354 0 : )
1355 0 : }
1356 :
1357 0 : async fn get_tenant_config_handler(
1358 0 : request: Request<Body>,
1359 0 : _cancel: CancellationToken,
1360 0 : ) -> Result<Response<Body>, ApiError> {
1361 0 : let tenant_shard_id: TenantShardId = parse_request_param(&request, "tenant_shard_id")?;
1362 0 : check_permission(&request, Some(tenant_shard_id.tenant_id))?;
1363 0 : let state = get_state(&request);
1364 :
1365 0 : let tenant = state
1366 0 : .tenant_manager
1367 0 : .get_attached_tenant_shard(tenant_shard_id)?;
1368 :
1369 0 : let response = HashMap::from([
1370 : (
1371 : "tenant_specific_overrides",
1372 0 : serde_json::to_value(tenant.tenant_specific_overrides())
1373 0 : .context("serializing tenant specific overrides")
1374 0 : .map_err(ApiError::InternalServerError)?,
1375 : ),
1376 : (
1377 0 : "effective_config",
1378 0 : serde_json::to_value(tenant.effective_config())
1379 0 : .context("serializing effective config")
1380 0 : .map_err(ApiError::InternalServerError)?,
1381 : ),
1382 : ]);
1383 :
1384 0 : json_response(StatusCode::OK, response)
1385 0 : }
1386 :
1387 0 : async fn update_tenant_config_handler(
1388 0 : mut request: Request<Body>,
1389 0 : _cancel: CancellationToken,
1390 0 : ) -> Result<Response<Body>, ApiError> {
1391 0 : let request_data: TenantConfigRequest = json_request(&mut request).await?;
1392 0 : let tenant_id = request_data.tenant_id;
1393 0 : check_permission(&request, Some(tenant_id))?;
1394 :
1395 0 : let new_tenant_conf =
1396 0 : TenantConfOpt::try_from(&request_data.config).map_err(ApiError::BadRequest)?;
1397 :
1398 0 : let state = get_state(&request);
1399 0 :
1400 0 : let tenant_shard_id = TenantShardId::unsharded(tenant_id);
1401 :
1402 0 : let tenant = state
1403 0 : .tenant_manager
1404 0 : .get_attached_tenant_shard(tenant_shard_id)?;
1405 0 : tenant.wait_to_become_active(ACTIVE_TENANT_TIMEOUT).await?;
1406 :
1407 : // This is a legacy API that only operates on attached tenants: the preferred
1408 : // API to use is the location_config/ endpoint, which lets the caller provide
1409 : // the full LocationConf.
1410 0 : let location_conf = LocationConf::attached_single(
1411 0 : new_tenant_conf.clone(),
1412 0 : tenant.get_generation(),
1413 0 : &ShardParameters::default(),
1414 0 : );
1415 0 :
1416 0 : crate::tenant::Tenant::persist_tenant_config(state.conf, &tenant_shard_id, &location_conf)
1417 0 : .await
1418 0 : .map_err(ApiError::InternalServerError)?;
1419 0 : tenant.set_new_tenant_config(new_tenant_conf);
1420 0 :
1421 0 : json_response(StatusCode::OK, ())
1422 0 : }
1423 :
1424 0 : async fn put_tenant_location_config_handler(
1425 0 : mut request: Request<Body>,
1426 0 : _cancel: CancellationToken,
1427 0 : ) -> Result<Response<Body>, ApiError> {
1428 0 : let tenant_shard_id: TenantShardId = parse_request_param(&request, "tenant_shard_id")?;
1429 :
1430 0 : let request_data: TenantLocationConfigRequest = json_request(&mut request).await?;
1431 0 : let flush = parse_query_param(&request, "flush_ms")?.map(Duration::from_millis);
1432 0 : let lazy = parse_query_param(&request, "lazy")?.unwrap_or(false);
1433 0 : check_permission(&request, Some(tenant_shard_id.tenant_id))?;
1434 :
1435 0 : let ctx = RequestContext::new(TaskKind::MgmtRequest, DownloadBehavior::Warn);
1436 0 : let state = get_state(&request);
1437 0 : let conf = state.conf;
1438 0 :
1439 0 : // The `Detached` state is special, it doesn't upsert a tenant, it removes
1440 0 : // its local disk content and drops it from memory.
1441 0 : if let LocationConfigMode::Detached = request_data.config.mode {
1442 0 : if let Err(e) = state
1443 0 : .tenant_manager
1444 0 : .detach_tenant(conf, tenant_shard_id, &state.deletion_queue_client)
1445 0 : .instrument(info_span!("tenant_detach",
1446 : tenant_id = %tenant_shard_id.tenant_id,
1447 0 : shard_id = %tenant_shard_id.shard_slug()
1448 : ))
1449 0 : .await
1450 : {
1451 0 : match e {
1452 0 : TenantStateError::SlotError(TenantSlotError::NotFound(_)) => {
1453 0 : // This API is idempotent: a NotFound on a detach is fine.
1454 0 : }
1455 0 : _ => return Err(e.into()),
1456 : }
1457 0 : }
1458 0 : return json_response(StatusCode::OK, ());
1459 0 : }
1460 :
1461 0 : let location_conf =
1462 0 : LocationConf::try_from(&request_data.config).map_err(ApiError::BadRequest)?;
1463 :
1464 : // lazy==true queues up for activation or jumps the queue like normal when a compute connects,
1465 : // similar to at startup ordering.
1466 0 : let spawn_mode = if lazy {
1467 0 : tenant::SpawnMode::Lazy
1468 : } else {
1469 0 : tenant::SpawnMode::Eager
1470 : };
1471 :
1472 0 : let tenant = state
1473 0 : .tenant_manager
1474 0 : .upsert_location(tenant_shard_id, location_conf, flush, spawn_mode, &ctx)
1475 0 : .await?;
1476 0 : let stripe_size = tenant.as_ref().map(|t| t.get_shard_stripe_size());
1477 0 : let attached = tenant.is_some();
1478 :
1479 0 : if let Some(_flush_ms) = flush {
1480 0 : match state
1481 0 : .secondary_controller
1482 0 : .upload_tenant(tenant_shard_id)
1483 0 : .await
1484 : {
1485 : Ok(()) => {
1486 0 : tracing::info!("Uploaded heatmap during flush");
1487 : }
1488 0 : Err(e) => {
1489 0 : tracing::warn!("Failed to flush heatmap: {e}");
1490 : }
1491 : }
1492 : } else {
1493 0 : tracing::info!("No flush requested when configuring");
1494 : }
1495 :
1496 : // This API returns a vector of pageservers where the tenant is attached: this is
1497 : // primarily for use in the sharding service. For compatibilty, we also return this
1498 : // when called directly on a pageserver, but the payload is always zero or one shards.
1499 0 : let mut response = TenantLocationConfigResponse {
1500 0 : shards: Vec::new(),
1501 0 : stripe_size: None,
1502 0 : };
1503 0 : if attached {
1504 0 : response.shards.push(TenantShardLocation {
1505 0 : shard_id: tenant_shard_id,
1506 0 : node_id: state.conf.id,
1507 0 : });
1508 0 : if tenant_shard_id.shard_count.count() > 1 {
1509 : // Stripe size should be set if we are attached
1510 0 : debug_assert!(stripe_size.is_some());
1511 0 : response.stripe_size = stripe_size;
1512 0 : }
1513 0 : }
1514 :
1515 0 : json_response(StatusCode::OK, response)
1516 0 : }
1517 :
1518 0 : async fn list_location_config_handler(
1519 0 : request: Request<Body>,
1520 0 : _cancel: CancellationToken,
1521 0 : ) -> Result<Response<Body>, ApiError> {
1522 0 : let state = get_state(&request);
1523 0 : let slots = state.tenant_manager.list();
1524 0 : let result = LocationConfigListResponse {
1525 0 : tenant_shards: slots
1526 0 : .into_iter()
1527 0 : .map(|(tenant_shard_id, slot)| {
1528 0 : let v = match slot {
1529 0 : TenantSlot::Attached(t) => Some(t.get_location_conf()),
1530 0 : TenantSlot::Secondary(s) => Some(s.get_location_conf()),
1531 0 : TenantSlot::InProgress(_) => None,
1532 : };
1533 0 : (tenant_shard_id, v)
1534 0 : })
1535 0 : .collect(),
1536 0 : };
1537 0 : json_response(StatusCode::OK, result)
1538 0 : }
1539 :
1540 0 : async fn get_location_config_handler(
1541 0 : request: Request<Body>,
1542 0 : _cancel: CancellationToken,
1543 0 : ) -> Result<Response<Body>, ApiError> {
1544 0 : let state = get_state(&request);
1545 0 : let tenant_shard_id: TenantShardId = parse_request_param(&request, "tenant_shard_id")?;
1546 0 : let slot = state.tenant_manager.get(tenant_shard_id);
1547 :
1548 0 : let Some(slot) = slot else {
1549 0 : return Err(ApiError::NotFound(
1550 0 : anyhow::anyhow!("Tenant shard not found").into(),
1551 0 : ));
1552 : };
1553 :
1554 0 : let result: Option<LocationConfig> = match slot {
1555 0 : TenantSlot::Attached(t) => Some(t.get_location_conf()),
1556 0 : TenantSlot::Secondary(s) => Some(s.get_location_conf()),
1557 0 : TenantSlot::InProgress(_) => None,
1558 : };
1559 :
1560 0 : json_response(StatusCode::OK, result)
1561 0 : }
1562 :
1563 : // Do a time travel recovery on the given tenant/tenant shard. Tenant needs to be detached
1564 : // (from all pageservers) as it invalidates consistency assumptions.
1565 0 : async fn tenant_time_travel_remote_storage_handler(
1566 0 : request: Request<Body>,
1567 0 : cancel: CancellationToken,
1568 0 : ) -> Result<Response<Body>, ApiError> {
1569 0 : let tenant_shard_id: TenantShardId = parse_request_param(&request, "tenant_shard_id")?;
1570 :
1571 0 : check_permission(&request, Some(tenant_shard_id.tenant_id))?;
1572 :
1573 0 : let timestamp_raw = must_get_query_param(&request, "travel_to")?;
1574 0 : let timestamp = humantime::parse_rfc3339(×tamp_raw)
1575 0 : .with_context(|| format!("Invalid time for travel_to: {timestamp_raw:?}"))
1576 0 : .map_err(ApiError::BadRequest)?;
1577 :
1578 0 : let done_if_after_raw = must_get_query_param(&request, "done_if_after")?;
1579 0 : let done_if_after = humantime::parse_rfc3339(&done_if_after_raw)
1580 0 : .with_context(|| format!("Invalid time for done_if_after: {done_if_after_raw:?}"))
1581 0 : .map_err(ApiError::BadRequest)?;
1582 :
1583 : // This is just a sanity check to fend off naive wrong usages of the API:
1584 : // the tenant needs to be detached *everywhere*
1585 0 : let state = get_state(&request);
1586 0 : let we_manage_tenant = state.tenant_manager.manages_tenant_shard(tenant_shard_id);
1587 0 : if we_manage_tenant {
1588 0 : return Err(ApiError::BadRequest(anyhow!(
1589 0 : "Tenant {tenant_shard_id} is already attached at this pageserver"
1590 0 : )));
1591 0 : }
1592 0 :
1593 0 : if timestamp > done_if_after {
1594 0 : return Err(ApiError::BadRequest(anyhow!(
1595 0 : "The done_if_after timestamp comes before the timestamp to recover to"
1596 0 : )));
1597 0 : }
1598 0 :
1599 0 : tracing::info!("Issuing time travel request internally. timestamp={timestamp_raw}, done_if_after={done_if_after_raw}");
1600 :
1601 0 : remote_timeline_client::upload::time_travel_recover_tenant(
1602 0 : &state.remote_storage,
1603 0 : &tenant_shard_id,
1604 0 : timestamp,
1605 0 : done_if_after,
1606 0 : &cancel,
1607 0 : )
1608 0 : .await
1609 0 : .map_err(|e| match e {
1610 0 : TimeTravelError::BadInput(e) => {
1611 0 : warn!("bad input error: {e}");
1612 0 : ApiError::BadRequest(anyhow!("bad input error"))
1613 : }
1614 : TimeTravelError::Unimplemented => {
1615 0 : ApiError::BadRequest(anyhow!("unimplemented for the configured remote storage"))
1616 : }
1617 0 : TimeTravelError::Cancelled => ApiError::InternalServerError(anyhow!("cancelled")),
1618 : TimeTravelError::TooManyVersions => {
1619 0 : ApiError::InternalServerError(anyhow!("too many versions in remote storage"))
1620 : }
1621 0 : TimeTravelError::Other(e) => {
1622 0 : warn!("internal error: {e}");
1623 0 : ApiError::InternalServerError(anyhow!("internal error"))
1624 : }
1625 0 : })?;
1626 :
1627 0 : json_response(StatusCode::OK, ())
1628 0 : }
1629 :
1630 : /// Testing helper to transition a tenant to [`crate::tenant::TenantState::Broken`].
1631 0 : async fn handle_tenant_break(
1632 0 : r: Request<Body>,
1633 0 : _cancel: CancellationToken,
1634 0 : ) -> Result<Response<Body>, ApiError> {
1635 0 : let tenant_shard_id: TenantShardId = parse_request_param(&r, "tenant_shard_id")?;
1636 :
1637 0 : let state = get_state(&r);
1638 0 : state
1639 0 : .tenant_manager
1640 0 : .get_attached_tenant_shard(tenant_shard_id)?
1641 0 : .set_broken("broken from test".to_owned())
1642 0 : .await;
1643 :
1644 0 : json_response(StatusCode::OK, ())
1645 0 : }
1646 :
1647 : // Obtains an lsn lease on the given timeline.
1648 0 : async fn lsn_lease_handler(
1649 0 : request: Request<Body>,
1650 0 : _cancel: CancellationToken,
1651 0 : ) -> Result<Response<Body>, ApiError> {
1652 0 : let tenant_shard_id: TenantShardId = parse_request_param(&request, "tenant_shard_id")?;
1653 0 : let timeline_id: TimelineId = parse_request_param(&request, "timeline_id")?;
1654 0 : check_permission(&request, Some(tenant_shard_id.tenant_id))?;
1655 :
1656 0 : let lsn: Lsn = parse_query_param(&request, "lsn")?
1657 0 : .ok_or_else(|| ApiError::BadRequest(anyhow!("missing 'lsn' query parameter")))?;
1658 :
1659 0 : let ctx = RequestContext::new(TaskKind::MgmtRequest, DownloadBehavior::Download);
1660 0 :
1661 0 : let state = get_state(&request);
1662 :
1663 0 : let timeline =
1664 0 : active_timeline_of_active_tenant(&state.tenant_manager, tenant_shard_id, timeline_id)
1665 0 : .await?;
1666 0 : let result = timeline
1667 0 : .make_lsn_lease(lsn, timeline.get_lsn_lease_length(), &ctx)
1668 0 : .map_err(|e| ApiError::InternalServerError(e.context("lsn lease http handler")))?;
1669 :
1670 0 : json_response(StatusCode::OK, result)
1671 0 : }
1672 :
1673 : // Run GC immediately on given timeline.
1674 0 : async fn timeline_gc_handler(
1675 0 : mut request: Request<Body>,
1676 0 : cancel: CancellationToken,
1677 0 : ) -> Result<Response<Body>, ApiError> {
1678 0 : let tenant_shard_id: TenantShardId = parse_request_param(&request, "tenant_shard_id")?;
1679 0 : let timeline_id: TimelineId = parse_request_param(&request, "timeline_id")?;
1680 0 : check_permission(&request, Some(tenant_shard_id.tenant_id))?;
1681 :
1682 0 : let gc_req: TimelineGcRequest = json_request(&mut request).await?;
1683 :
1684 0 : let ctx = RequestContext::new(TaskKind::MgmtRequest, DownloadBehavior::Download);
1685 0 : let gc_result = mgr::immediate_gc(tenant_shard_id, timeline_id, gc_req, cancel, &ctx).await?;
1686 :
1687 0 : json_response(StatusCode::OK, gc_result)
1688 0 : }
1689 :
1690 : // Run compaction immediately on given timeline.
1691 0 : async fn timeline_compact_handler(
1692 0 : request: Request<Body>,
1693 0 : cancel: CancellationToken,
1694 0 : ) -> Result<Response<Body>, ApiError> {
1695 0 : let tenant_shard_id: TenantShardId = parse_request_param(&request, "tenant_shard_id")?;
1696 0 : let timeline_id: TimelineId = parse_request_param(&request, "timeline_id")?;
1697 0 : check_permission(&request, Some(tenant_shard_id.tenant_id))?;
1698 :
1699 0 : let state = get_state(&request);
1700 0 :
1701 0 : let mut flags = EnumSet::empty();
1702 0 : if Some(true) == parse_query_param::<_, bool>(&request, "force_repartition")? {
1703 0 : flags |= CompactFlags::ForceRepartition;
1704 0 : }
1705 0 : if Some(true) == parse_query_param::<_, bool>(&request, "force_image_layer_creation")? {
1706 0 : flags |= CompactFlags::ForceImageLayerCreation;
1707 0 : }
1708 0 : let wait_until_uploaded =
1709 0 : parse_query_param::<_, bool>(&request, "wait_until_uploaded")?.unwrap_or(false);
1710 :
1711 0 : async {
1712 0 : let ctx = RequestContext::new(TaskKind::MgmtRequest, DownloadBehavior::Download);
1713 0 : let timeline = active_timeline_of_active_tenant(&state.tenant_manager, tenant_shard_id, timeline_id).await?;
1714 0 : timeline
1715 0 : .compact(&cancel, flags, &ctx)
1716 0 : .await
1717 0 : .map_err(|e| ApiError::InternalServerError(e.into()))?;
1718 0 : if wait_until_uploaded {
1719 0 : timeline.remote_client.wait_completion().await.map_err(ApiError::InternalServerError)?;
1720 0 : }
1721 0 : json_response(StatusCode::OK, ())
1722 0 : }
1723 0 : .instrument(info_span!("manual_compaction", tenant_id = %tenant_shard_id.tenant_id, shard_id = %tenant_shard_id.shard_slug(), %timeline_id))
1724 0 : .await
1725 0 : }
1726 :
1727 : // Run checkpoint immediately on given timeline.
1728 0 : async fn timeline_checkpoint_handler(
1729 0 : request: Request<Body>,
1730 0 : cancel: CancellationToken,
1731 0 : ) -> Result<Response<Body>, ApiError> {
1732 0 : let tenant_shard_id: TenantShardId = parse_request_param(&request, "tenant_shard_id")?;
1733 0 : let timeline_id: TimelineId = parse_request_param(&request, "timeline_id")?;
1734 0 : check_permission(&request, Some(tenant_shard_id.tenant_id))?;
1735 :
1736 0 : let state = get_state(&request);
1737 0 :
1738 0 : let mut flags = EnumSet::empty();
1739 0 : if Some(true) == parse_query_param::<_, bool>(&request, "force_repartition")? {
1740 0 : flags |= CompactFlags::ForceRepartition;
1741 0 : }
1742 0 : if Some(true) == parse_query_param::<_, bool>(&request, "force_image_layer_creation")? {
1743 0 : flags |= CompactFlags::ForceImageLayerCreation;
1744 0 : }
1745 0 : let wait_until_uploaded =
1746 0 : parse_query_param::<_, bool>(&request, "wait_until_uploaded")?.unwrap_or(false);
1747 :
1748 0 : async {
1749 0 : let ctx = RequestContext::new(TaskKind::MgmtRequest, DownloadBehavior::Download);
1750 0 : let timeline = active_timeline_of_active_tenant(&state.tenant_manager, tenant_shard_id, timeline_id).await?;
1751 0 : timeline
1752 0 : .freeze_and_flush()
1753 0 : .await
1754 0 : .map_err(|e| {
1755 0 : match e {
1756 0 : tenant::timeline::FlushLayerError::Cancelled => ApiError::ShuttingDown,
1757 0 : other => ApiError::InternalServerError(other.into()),
1758 :
1759 : }
1760 0 : })?;
1761 0 : timeline
1762 0 : .compact(&cancel, flags, &ctx)
1763 0 : .await
1764 0 : .map_err(|e|
1765 0 : match e {
1766 0 : CompactionError::ShuttingDown => ApiError::ShuttingDown,
1767 0 : CompactionError::Other(e) => ApiError::InternalServerError(e)
1768 0 : }
1769 0 : )?;
1770 :
1771 0 : if wait_until_uploaded {
1772 0 : timeline.remote_client.wait_completion().await.map_err(ApiError::InternalServerError)?;
1773 0 : }
1774 :
1775 0 : json_response(StatusCode::OK, ())
1776 0 : }
1777 0 : .instrument(info_span!("manual_checkpoint", tenant_id = %tenant_shard_id.tenant_id, shard_id = %tenant_shard_id.shard_slug(), %timeline_id))
1778 0 : .await
1779 0 : }
1780 :
1781 0 : async fn timeline_download_remote_layers_handler_post(
1782 0 : mut request: Request<Body>,
1783 0 : _cancel: CancellationToken,
1784 0 : ) -> Result<Response<Body>, ApiError> {
1785 0 : let tenant_shard_id: TenantShardId = parse_request_param(&request, "tenant_shard_id")?;
1786 0 : let timeline_id: TimelineId = parse_request_param(&request, "timeline_id")?;
1787 0 : let body: DownloadRemoteLayersTaskSpawnRequest = json_request(&mut request).await?;
1788 0 : check_permission(&request, Some(tenant_shard_id.tenant_id))?;
1789 :
1790 0 : let state = get_state(&request);
1791 :
1792 0 : let timeline =
1793 0 : active_timeline_of_active_tenant(&state.tenant_manager, tenant_shard_id, timeline_id)
1794 0 : .await?;
1795 0 : match timeline.spawn_download_all_remote_layers(body).await {
1796 0 : Ok(st) => json_response(StatusCode::ACCEPTED, st),
1797 0 : Err(st) => json_response(StatusCode::CONFLICT, st),
1798 : }
1799 0 : }
1800 :
1801 0 : async fn timeline_download_remote_layers_handler_get(
1802 0 : request: Request<Body>,
1803 0 : _cancel: CancellationToken,
1804 0 : ) -> Result<Response<Body>, ApiError> {
1805 0 : let tenant_shard_id: TenantShardId = parse_request_param(&request, "tenant_shard_id")?;
1806 0 : check_permission(&request, Some(tenant_shard_id.tenant_id))?;
1807 0 : let timeline_id: TimelineId = parse_request_param(&request, "timeline_id")?;
1808 0 : let state = get_state(&request);
1809 :
1810 0 : let timeline =
1811 0 : active_timeline_of_active_tenant(&state.tenant_manager, tenant_shard_id, timeline_id)
1812 0 : .await?;
1813 0 : let info = timeline
1814 0 : .get_download_all_remote_layers_task_info()
1815 0 : .context("task never started since last pageserver process start")
1816 0 : .map_err(|e| ApiError::NotFound(e.into()))?;
1817 0 : json_response(StatusCode::OK, info)
1818 0 : }
1819 :
1820 0 : async fn timeline_detach_ancestor_handler(
1821 0 : request: Request<Body>,
1822 0 : _cancel: CancellationToken,
1823 0 : ) -> Result<Response<Body>, ApiError> {
1824 : use crate::tenant::timeline::detach_ancestor::Options;
1825 0 : let tenant_shard_id: TenantShardId = parse_request_param(&request, "tenant_shard_id")?;
1826 0 : check_permission(&request, Some(tenant_shard_id.tenant_id))?;
1827 0 : let timeline_id: TimelineId = parse_request_param(&request, "timeline_id")?;
1828 :
1829 0 : let span = tracing::info_span!("detach_ancestor", tenant_id=%tenant_shard_id.tenant_id, shard_id=%tenant_shard_id.shard_slug(), %timeline_id);
1830 :
1831 0 : async move {
1832 0 : let mut options = Options::default();
1833 :
1834 0 : let rewrite_concurrency =
1835 0 : parse_query_param::<_, std::num::NonZeroUsize>(&request, "rewrite_concurrency")?;
1836 0 : let copy_concurrency =
1837 0 : parse_query_param::<_, std::num::NonZeroUsize>(&request, "copy_concurrency")?;
1838 :
1839 0 : [
1840 0 : (&mut options.rewrite_concurrency, rewrite_concurrency),
1841 0 : (&mut options.copy_concurrency, copy_concurrency),
1842 0 : ]
1843 0 : .into_iter()
1844 0 : .filter_map(|(target, val)| val.map(|val| (target, val)))
1845 0 : .for_each(|(target, val)| *target = val);
1846 0 :
1847 0 : let state = get_state(&request);
1848 :
1849 0 : let tenant = state
1850 0 : .tenant_manager
1851 0 : .get_attached_tenant_shard(tenant_shard_id)?;
1852 :
1853 0 : tenant.wait_to_become_active(ACTIVE_TENANT_TIMEOUT).await?;
1854 :
1855 0 : let ctx = RequestContext::new(TaskKind::DetachAncestor, DownloadBehavior::Download);
1856 0 : let ctx = &ctx;
1857 :
1858 0 : let timeline = tenant.get_timeline(timeline_id, true)?;
1859 :
1860 0 : let (_guard, prepared) = timeline
1861 0 : .prepare_to_detach_from_ancestor(&tenant, options, ctx)
1862 0 : .await?;
1863 :
1864 0 : let res = state
1865 0 : .tenant_manager
1866 0 : .complete_detaching_timeline_ancestor(tenant_shard_id, timeline_id, prepared, ctx)
1867 0 : .await;
1868 :
1869 0 : match res {
1870 0 : Ok(reparented_timelines) => {
1871 0 : let resp = pageserver_api::models::detach_ancestor::AncestorDetached {
1872 0 : reparented_timelines,
1873 0 : };
1874 0 :
1875 0 : json_response(StatusCode::OK, resp)
1876 : }
1877 0 : Err(e) => Err(ApiError::InternalServerError(
1878 0 : e.context("timeline detach completion"),
1879 0 : )),
1880 : }
1881 0 : }
1882 0 : .instrument(span)
1883 0 : .await
1884 0 : }
1885 :
1886 0 : async fn deletion_queue_flush(
1887 0 : r: Request<Body>,
1888 0 : cancel: CancellationToken,
1889 0 : ) -> Result<Response<Body>, ApiError> {
1890 0 : let state = get_state(&r);
1891 :
1892 0 : let execute = parse_query_param(&r, "execute")?.unwrap_or(false);
1893 0 :
1894 0 : let flush = async {
1895 0 : if execute {
1896 0 : state.deletion_queue_client.flush_execute().await
1897 : } else {
1898 0 : state.deletion_queue_client.flush().await
1899 : }
1900 0 : }
1901 : // DeletionQueueError's only case is shutting down.
1902 0 : .map_err(|_| ApiError::ShuttingDown);
1903 :
1904 : tokio::select! {
1905 : res = flush => {
1906 0 : res.map(|()| json_response(StatusCode::OK, ()))?
1907 : }
1908 : _ = cancel.cancelled() => {
1909 : Err(ApiError::ShuttingDown)
1910 : }
1911 : }
1912 0 : }
1913 :
1914 : /// Try if `GetPage@Lsn` is successful, useful for manual debugging.
1915 0 : async fn getpage_at_lsn_handler(
1916 0 : request: Request<Body>,
1917 0 : _cancel: CancellationToken,
1918 0 : ) -> Result<Response<Body>, ApiError> {
1919 0 : let tenant_shard_id: TenantShardId = parse_request_param(&request, "tenant_shard_id")?;
1920 0 : let timeline_id: TimelineId = parse_request_param(&request, "timeline_id")?;
1921 0 : check_permission(&request, Some(tenant_shard_id.tenant_id))?;
1922 0 : let state = get_state(&request);
1923 :
1924 : struct Key(crate::repository::Key);
1925 :
1926 : impl std::str::FromStr for Key {
1927 : type Err = anyhow::Error;
1928 :
1929 0 : fn from_str(s: &str) -> std::result::Result<Self, Self::Err> {
1930 0 : crate::repository::Key::from_hex(s).map(Key)
1931 0 : }
1932 : }
1933 :
1934 0 : let key: Key = parse_query_param(&request, "key")?
1935 0 : .ok_or_else(|| ApiError::BadRequest(anyhow!("missing 'key' query parameter")))?;
1936 0 : let lsn: Lsn = parse_query_param(&request, "lsn")?
1937 0 : .ok_or_else(|| ApiError::BadRequest(anyhow!("missing 'lsn' query parameter")))?;
1938 :
1939 0 : async {
1940 0 : let ctx = RequestContext::new(TaskKind::MgmtRequest, DownloadBehavior::Download);
1941 0 : let timeline = active_timeline_of_active_tenant(&state.tenant_manager, tenant_shard_id, timeline_id).await?;
1942 :
1943 0 : let page = timeline.get(key.0, lsn, &ctx).await?;
1944 :
1945 0 : Result::<_, ApiError>::Ok(
1946 0 : Response::builder()
1947 0 : .status(StatusCode::OK)
1948 0 : .header(header::CONTENT_TYPE, "application/octet-stream")
1949 0 : .body(hyper::Body::from(page))
1950 0 : .unwrap(),
1951 0 : )
1952 0 : }
1953 0 : .instrument(info_span!("timeline_get", tenant_id = %tenant_shard_id.tenant_id, shard_id = %tenant_shard_id.shard_slug(), %timeline_id))
1954 0 : .await
1955 0 : }
1956 :
1957 0 : async fn timeline_collect_keyspace(
1958 0 : request: Request<Body>,
1959 0 : _cancel: CancellationToken,
1960 0 : ) -> Result<Response<Body>, ApiError> {
1961 0 : let tenant_shard_id: TenantShardId = parse_request_param(&request, "tenant_shard_id")?;
1962 0 : let timeline_id: TimelineId = parse_request_param(&request, "timeline_id")?;
1963 0 : check_permission(&request, Some(tenant_shard_id.tenant_id))?;
1964 0 : let state = get_state(&request);
1965 :
1966 0 : let at_lsn: Option<Lsn> = parse_query_param(&request, "at_lsn")?;
1967 :
1968 0 : async {
1969 0 : let ctx = RequestContext::new(TaskKind::MgmtRequest, DownloadBehavior::Download);
1970 0 : let timeline = active_timeline_of_active_tenant(&state.tenant_manager, tenant_shard_id, timeline_id).await?;
1971 0 : let at_lsn = at_lsn.unwrap_or_else(|| timeline.get_last_record_lsn());
1972 0 : let (dense_ks, sparse_ks) = timeline
1973 0 : .collect_keyspace(at_lsn, &ctx)
1974 0 : .await
1975 0 : .map_err(|e| ApiError::InternalServerError(e.into()))?;
1976 :
1977 : // This API is currently used by pagebench. Pagebench will iterate all keys within the keyspace.
1978 : // Therefore, we split dense/sparse keys in this API.
1979 0 : let res = pageserver_api::models::partitioning::Partitioning { keys: dense_ks, sparse_keys: sparse_ks, at_lsn };
1980 0 :
1981 0 : json_response(StatusCode::OK, res)
1982 0 : }
1983 0 : .instrument(info_span!("timeline_collect_keyspace", tenant_id = %tenant_shard_id.tenant_id, shard_id = %tenant_shard_id.shard_slug(), %timeline_id))
1984 0 : .await
1985 0 : }
1986 :
1987 0 : async fn active_timeline_of_active_tenant(
1988 0 : tenant_manager: &TenantManager,
1989 0 : tenant_shard_id: TenantShardId,
1990 0 : timeline_id: TimelineId,
1991 0 : ) -> Result<Arc<Timeline>, ApiError> {
1992 0 : let tenant = tenant_manager.get_attached_tenant_shard(tenant_shard_id)?;
1993 :
1994 0 : tenant.wait_to_become_active(ACTIVE_TENANT_TIMEOUT).await?;
1995 :
1996 0 : Ok(tenant.get_timeline(timeline_id, true)?)
1997 0 : }
1998 :
1999 0 : async fn always_panic_handler(
2000 0 : req: Request<Body>,
2001 0 : _cancel: CancellationToken,
2002 0 : ) -> Result<Response<Body>, ApiError> {
2003 0 : // Deliberately cause a panic to exercise the panic hook registered via std::panic::set_hook().
2004 0 : // For pageserver, the relevant panic hook is `tracing_panic_hook` , and the `sentry` crate's wrapper around it.
2005 0 : // Use catch_unwind to ensure that tokio nor hyper are distracted by our panic.
2006 0 : let query = req.uri().query();
2007 0 : let _ = std::panic::catch_unwind(|| {
2008 0 : panic!("unconditional panic for testing panic hook integration; request query: {query:?}")
2009 0 : });
2010 0 : json_response(StatusCode::NO_CONTENT, ())
2011 0 : }
2012 :
2013 0 : async fn disk_usage_eviction_run(
2014 0 : mut r: Request<Body>,
2015 0 : cancel: CancellationToken,
2016 0 : ) -> Result<Response<Body>, ApiError> {
2017 0 : check_permission(&r, None)?;
2018 :
2019 0 : #[derive(Debug, Clone, Copy, serde::Serialize, serde::Deserialize)]
2020 : struct Config {
2021 : /// How many bytes to evict before reporting that pressure is relieved.
2022 : evict_bytes: u64,
2023 :
2024 : #[serde(default)]
2025 : eviction_order: crate::disk_usage_eviction_task::EvictionOrder,
2026 : }
2027 :
2028 : #[derive(Debug, Clone, Copy, serde::Serialize)]
2029 : struct Usage {
2030 : // remains unchanged after instantiation of the struct
2031 : evict_bytes: u64,
2032 : // updated by `add_available_bytes`
2033 : freed_bytes: u64,
2034 : }
2035 :
2036 : impl crate::disk_usage_eviction_task::Usage for Usage {
2037 0 : fn has_pressure(&self) -> bool {
2038 0 : self.evict_bytes > self.freed_bytes
2039 0 : }
2040 :
2041 0 : fn add_available_bytes(&mut self, bytes: u64) {
2042 0 : self.freed_bytes += bytes;
2043 0 : }
2044 : }
2045 :
2046 0 : let config = json_request::<Config>(&mut r).await?;
2047 :
2048 0 : let usage = Usage {
2049 0 : evict_bytes: config.evict_bytes,
2050 0 : freed_bytes: 0,
2051 0 : };
2052 0 :
2053 0 : let state = get_state(&r);
2054 0 : let eviction_state = state.disk_usage_eviction_state.clone();
2055 :
2056 0 : let res = crate::disk_usage_eviction_task::disk_usage_eviction_task_iteration_impl(
2057 0 : &eviction_state,
2058 0 : &state.remote_storage,
2059 0 : usage,
2060 0 : &state.tenant_manager,
2061 0 : config.eviction_order,
2062 0 : &cancel,
2063 0 : )
2064 0 : .await;
2065 :
2066 0 : info!(?res, "disk_usage_eviction_task_iteration_impl finished");
2067 :
2068 0 : let res = res.map_err(ApiError::InternalServerError)?;
2069 :
2070 0 : json_response(StatusCode::OK, res)
2071 0 : }
2072 :
2073 0 : async fn secondary_upload_handler(
2074 0 : request: Request<Body>,
2075 0 : _cancel: CancellationToken,
2076 0 : ) -> Result<Response<Body>, ApiError> {
2077 0 : let state = get_state(&request);
2078 0 : let tenant_shard_id: TenantShardId = parse_request_param(&request, "tenant_shard_id")?;
2079 0 : state
2080 0 : .secondary_controller
2081 0 : .upload_tenant(tenant_shard_id)
2082 0 : .await
2083 0 : .map_err(ApiError::InternalServerError)?;
2084 :
2085 0 : json_response(StatusCode::OK, ())
2086 0 : }
2087 :
2088 0 : async fn tenant_scan_remote_handler(
2089 0 : request: Request<Body>,
2090 0 : cancel: CancellationToken,
2091 0 : ) -> Result<Response<Body>, ApiError> {
2092 0 : let state = get_state(&request);
2093 0 : let tenant_id: TenantId = parse_request_param(&request, "tenant_id")?;
2094 :
2095 0 : let mut response = TenantScanRemoteStorageResponse::default();
2096 :
2097 0 : let (shards, _other_keys) =
2098 0 : list_remote_tenant_shards(&state.remote_storage, tenant_id, cancel.clone())
2099 0 : .await
2100 0 : .map_err(|e| ApiError::InternalServerError(anyhow::anyhow!(e)))?;
2101 :
2102 0 : for tenant_shard_id in shards {
2103 0 : let (timeline_ids, _other_keys) =
2104 0 : list_remote_timelines(&state.remote_storage, tenant_shard_id, cancel.clone())
2105 0 : .await
2106 0 : .map_err(|e| ApiError::InternalServerError(anyhow::anyhow!(e)))?;
2107 :
2108 0 : let mut generation = Generation::none();
2109 0 : for timeline_id in timeline_ids {
2110 0 : match download_index_part(
2111 0 : &state.remote_storage,
2112 0 : &tenant_shard_id,
2113 0 : &timeline_id,
2114 0 : Generation::MAX,
2115 0 : &cancel,
2116 0 : )
2117 0 : .instrument(info_span!("download_index_part",
2118 : tenant_id=%tenant_shard_id.tenant_id,
2119 0 : shard_id=%tenant_shard_id.shard_slug(),
2120 : %timeline_id))
2121 0 : .await
2122 : {
2123 0 : Ok((index_part, index_generation)) => {
2124 0 : tracing::info!("Found timeline {tenant_shard_id}/{timeline_id} metadata (gen {index_generation:?}, {} layers, {} consistent LSN)",
2125 0 : index_part.layer_metadata.len(), index_part.metadata.disk_consistent_lsn());
2126 0 : generation = std::cmp::max(generation, index_generation);
2127 : }
2128 : Err(DownloadError::NotFound) => {
2129 : // This is normal for tenants that were created with multiple shards: they have an unsharded path
2130 : // containing the timeline's initdb tarball but no index. Otherwise it is a bit strange.
2131 0 : tracing::info!("Timeline path {tenant_shard_id}/{timeline_id} exists in remote storage but has no index, skipping");
2132 0 : continue;
2133 : }
2134 0 : Err(e) => {
2135 0 : return Err(ApiError::InternalServerError(anyhow::anyhow!(e)));
2136 : }
2137 : };
2138 : }
2139 :
2140 0 : response.shards.push(TenantScanRemoteStorageShard {
2141 0 : tenant_shard_id,
2142 0 : generation: generation.into(),
2143 0 : });
2144 : }
2145 :
2146 0 : if response.shards.is_empty() {
2147 0 : return Err(ApiError::NotFound(
2148 0 : anyhow::anyhow!("No shards found for tenant ID {tenant_id}").into(),
2149 0 : ));
2150 0 : }
2151 0 :
2152 0 : json_response(StatusCode::OK, response)
2153 0 : }
2154 :
2155 0 : async fn secondary_download_handler(
2156 0 : request: Request<Body>,
2157 0 : _cancel: CancellationToken,
2158 0 : ) -> Result<Response<Body>, ApiError> {
2159 0 : let state = get_state(&request);
2160 0 : let tenant_shard_id: TenantShardId = parse_request_param(&request, "tenant_shard_id")?;
2161 0 : let wait = parse_query_param(&request, "wait_ms")?.map(Duration::from_millis);
2162 :
2163 : // We don't need this to issue the download request, but:
2164 : // - it enables us to cleanly return 404 if we get a request for an absent shard
2165 : // - we will use this to provide status feedback in the response
2166 0 : let Some(secondary_tenant) = state
2167 0 : .tenant_manager
2168 0 : .get_secondary_tenant_shard(tenant_shard_id)
2169 : else {
2170 0 : return Err(ApiError::NotFound(
2171 0 : anyhow::anyhow!("Shard {} not found", tenant_shard_id).into(),
2172 0 : ));
2173 : };
2174 :
2175 0 : let timeout = wait.unwrap_or(Duration::MAX);
2176 :
2177 0 : let status = match tokio::time::timeout(
2178 0 : timeout,
2179 0 : state.secondary_controller.download_tenant(tenant_shard_id),
2180 0 : )
2181 0 : .await
2182 : {
2183 : // Download job ran to completion.
2184 0 : Ok(Ok(())) => StatusCode::OK,
2185 : // Edge case: downloads aren't usually fallible: things like a missing heatmap are considered
2186 : // okay. We could get an error here in the unlikely edge case that the tenant
2187 : // was detached between our check above and executing the download job.
2188 0 : Ok(Err(e)) => return Err(ApiError::InternalServerError(e)),
2189 : // A timeout is not an error: we have started the download, we're just not done
2190 : // yet. The caller will get a response body indicating status.
2191 0 : Err(_) => StatusCode::ACCEPTED,
2192 : };
2193 :
2194 0 : let progress = secondary_tenant.progress.lock().unwrap().clone();
2195 0 :
2196 0 : json_response(status, progress)
2197 0 : }
2198 :
2199 0 : async fn secondary_status_handler(
2200 0 : request: Request<Body>,
2201 0 : _cancel: CancellationToken,
2202 0 : ) -> Result<Response<Body>, ApiError> {
2203 0 : let state = get_state(&request);
2204 0 : let tenant_shard_id: TenantShardId = parse_request_param(&request, "tenant_shard_id")?;
2205 :
2206 0 : let Some(secondary_tenant) = state
2207 0 : .tenant_manager
2208 0 : .get_secondary_tenant_shard(tenant_shard_id)
2209 : else {
2210 0 : return Err(ApiError::NotFound(
2211 0 : anyhow::anyhow!("Shard {} not found", tenant_shard_id).into(),
2212 0 : ));
2213 : };
2214 :
2215 0 : let progress = secondary_tenant.progress.lock().unwrap().clone();
2216 0 :
2217 0 : json_response(StatusCode::OK, progress)
2218 0 : }
2219 :
2220 0 : async fn handler_404(_: Request<Body>) -> Result<Response<Body>, ApiError> {
2221 0 : json_response(
2222 0 : StatusCode::NOT_FOUND,
2223 0 : HttpErrorBody::from_msg("page not found".to_owned()),
2224 0 : )
2225 0 : }
2226 :
2227 0 : async fn post_tracing_event_handler(
2228 0 : mut r: Request<Body>,
2229 0 : _cancel: CancellationToken,
2230 0 : ) -> Result<Response<Body>, ApiError> {
2231 0 : #[derive(Debug, serde::Deserialize)]
2232 : #[serde(rename_all = "lowercase")]
2233 : enum Level {
2234 : Error,
2235 : Warn,
2236 : Info,
2237 : Debug,
2238 : Trace,
2239 : }
2240 0 : #[derive(Debug, serde::Deserialize)]
2241 : struct Request {
2242 : level: Level,
2243 : message: String,
2244 : }
2245 0 : let body: Request = json_request(&mut r)
2246 0 : .await
2247 0 : .map_err(|_| ApiError::BadRequest(anyhow::anyhow!("invalid JSON body")))?;
2248 :
2249 0 : match body.level {
2250 0 : Level::Error => tracing::error!(?body.message),
2251 0 : Level::Warn => tracing::warn!(?body.message),
2252 0 : Level::Info => tracing::info!(?body.message),
2253 0 : Level::Debug => tracing::debug!(?body.message),
2254 0 : Level::Trace => tracing::trace!(?body.message),
2255 : }
2256 :
2257 0 : json_response(StatusCode::OK, ())
2258 0 : }
2259 :
2260 0 : async fn force_aux_policy_switch_handler(
2261 0 : mut r: Request<Body>,
2262 0 : _cancel: CancellationToken,
2263 0 : ) -> Result<Response<Body>, ApiError> {
2264 0 : check_permission(&r, None)?;
2265 0 : let tenant_shard_id: TenantShardId = parse_request_param(&r, "tenant_shard_id")?;
2266 0 : let timeline_id: TimelineId = parse_request_param(&r, "timeline_id")?;
2267 0 : let policy: AuxFilePolicy = json_request(&mut r).await?;
2268 :
2269 0 : let state = get_state(&r);
2270 :
2271 0 : let tenant = state
2272 0 : .tenant_manager
2273 0 : .get_attached_tenant_shard(tenant_shard_id)?;
2274 0 : tenant.wait_to_become_active(ACTIVE_TENANT_TIMEOUT).await?;
2275 0 : let timeline =
2276 0 : active_timeline_of_active_tenant(&state.tenant_manager, tenant_shard_id, timeline_id)
2277 0 : .await?;
2278 0 : timeline
2279 0 : .do_switch_aux_policy(policy)
2280 0 : .map_err(ApiError::InternalServerError)?;
2281 :
2282 0 : json_response(StatusCode::OK, ())
2283 0 : }
2284 :
2285 0 : async fn put_io_engine_handler(
2286 0 : mut r: Request<Body>,
2287 0 : _cancel: CancellationToken,
2288 0 : ) -> Result<Response<Body>, ApiError> {
2289 0 : check_permission(&r, None)?;
2290 0 : let kind: crate::virtual_file::IoEngineKind = json_request(&mut r).await?;
2291 0 : crate::virtual_file::io_engine::set(kind);
2292 0 : json_response(StatusCode::OK, ())
2293 0 : }
2294 :
2295 : /// Polled by control plane.
2296 : ///
2297 : /// See [`crate::utilization`].
2298 0 : async fn get_utilization(
2299 0 : r: Request<Body>,
2300 0 : _cancel: CancellationToken,
2301 0 : ) -> Result<Response<Body>, ApiError> {
2302 0 : fail::fail_point!("get-utilization-http-handler", |_| {
2303 0 : Err(ApiError::ResourceUnavailable("failpoint".into()))
2304 0 : });
2305 :
2306 : // this probably could be completely public, but lets make that change later.
2307 0 : check_permission(&r, None)?;
2308 :
2309 0 : let state = get_state(&r);
2310 0 : let mut g = state.latest_utilization.lock().await;
2311 :
2312 0 : let regenerate_every = Duration::from_secs(1);
2313 0 : let still_valid = g
2314 0 : .as_ref()
2315 0 : .is_some_and(|(captured_at, _)| captured_at.elapsed() < regenerate_every);
2316 0 :
2317 0 : // avoid needless statvfs calls even though those should be non-blocking fast.
2318 0 : // regenerate at most 1Hz to allow polling at any rate.
2319 0 : if !still_valid {
2320 0 : let path = state.conf.tenants_path();
2321 0 : let doc = crate::utilization::regenerate(path.as_std_path())
2322 0 : .map_err(ApiError::InternalServerError)?;
2323 :
2324 0 : let mut buf = Vec::new();
2325 0 : serde_json::to_writer(&mut buf, &doc)
2326 0 : .context("serialize")
2327 0 : .map_err(ApiError::InternalServerError)?;
2328 :
2329 0 : let body = bytes::Bytes::from(buf);
2330 0 :
2331 0 : *g = Some((std::time::Instant::now(), body));
2332 0 : }
2333 :
2334 : // hyper 0.14 doesn't yet have Response::clone so this is a bit of extra legwork
2335 0 : let cached = g.as_ref().expect("just set").1.clone();
2336 0 :
2337 0 : Response::builder()
2338 0 : .header(hyper::http::header::CONTENT_TYPE, "application/json")
2339 0 : // thought of using http date header, but that is second precision which does not give any
2340 0 : // debugging aid
2341 0 : .status(StatusCode::OK)
2342 0 : .body(hyper::Body::from(cached))
2343 0 : .context("build response")
2344 0 : .map_err(ApiError::InternalServerError)
2345 0 : }
2346 :
2347 0 : async fn list_aux_files(
2348 0 : mut request: Request<Body>,
2349 0 : _cancel: CancellationToken,
2350 0 : ) -> Result<Response<Body>, ApiError> {
2351 0 : let tenant_shard_id: TenantShardId = parse_request_param(&request, "tenant_shard_id")?;
2352 0 : let timeline_id: TimelineId = parse_request_param(&request, "timeline_id")?;
2353 0 : let body: ListAuxFilesRequest = json_request(&mut request).await?;
2354 0 : check_permission(&request, Some(tenant_shard_id.tenant_id))?;
2355 :
2356 0 : let state = get_state(&request);
2357 :
2358 0 : let timeline =
2359 0 : active_timeline_of_active_tenant(&state.tenant_manager, tenant_shard_id, timeline_id)
2360 0 : .await?;
2361 :
2362 0 : let ctx = RequestContext::new(TaskKind::MgmtRequest, DownloadBehavior::Download);
2363 0 : let files = timeline.list_aux_files(body.lsn, &ctx).await?;
2364 0 : json_response(StatusCode::OK, files)
2365 0 : }
2366 :
2367 0 : async fn perf_info(
2368 0 : request: Request<Body>,
2369 0 : _cancel: CancellationToken,
2370 0 : ) -> Result<Response<Body>, ApiError> {
2371 0 : let tenant_shard_id: TenantShardId = parse_request_param(&request, "tenant_shard_id")?;
2372 0 : let timeline_id: TimelineId = parse_request_param(&request, "timeline_id")?;
2373 0 : check_permission(&request, Some(tenant_shard_id.tenant_id))?;
2374 :
2375 0 : let state = get_state(&request);
2376 :
2377 0 : let timeline =
2378 0 : active_timeline_of_active_tenant(&state.tenant_manager, tenant_shard_id, timeline_id)
2379 0 : .await?;
2380 :
2381 0 : let result = timeline.perf_info().await;
2382 :
2383 0 : json_response(StatusCode::OK, result)
2384 0 : }
2385 :
2386 0 : async fn ingest_aux_files(
2387 0 : mut request: Request<Body>,
2388 0 : _cancel: CancellationToken,
2389 0 : ) -> Result<Response<Body>, ApiError> {
2390 0 : let tenant_shard_id: TenantShardId = parse_request_param(&request, "tenant_shard_id")?;
2391 0 : let timeline_id: TimelineId = parse_request_param(&request, "timeline_id")?;
2392 0 : let body: IngestAuxFilesRequest = json_request(&mut request).await?;
2393 0 : check_permission(&request, Some(tenant_shard_id.tenant_id))?;
2394 :
2395 0 : let state = get_state(&request);
2396 :
2397 0 : let timeline =
2398 0 : active_timeline_of_active_tenant(&state.tenant_manager, tenant_shard_id, timeline_id)
2399 0 : .await?;
2400 :
2401 0 : let mut modification = timeline.begin_modification(
2402 0 : Lsn(timeline.get_last_record_lsn().0 + 8), /* advance LSN by 8 */
2403 0 : );
2404 0 : let ctx = RequestContext::new(TaskKind::MgmtRequest, DownloadBehavior::Download);
2405 0 : for (fname, content) in body.aux_files {
2406 0 : modification
2407 0 : .put_file(&fname, content.as_bytes(), &ctx)
2408 0 : .await
2409 0 : .map_err(ApiError::InternalServerError)?;
2410 : }
2411 0 : modification
2412 0 : .commit(&ctx)
2413 0 : .await
2414 0 : .map_err(ApiError::InternalServerError)?;
2415 :
2416 0 : json_response(StatusCode::OK, ())
2417 0 : }
2418 :
2419 : /// Report on the largest tenants on this pageserver, for the storage controller to identify
2420 : /// candidates for splitting
2421 0 : async fn post_top_tenants(
2422 0 : mut r: Request<Body>,
2423 0 : _cancel: CancellationToken,
2424 0 : ) -> Result<Response<Body>, ApiError> {
2425 0 : check_permission(&r, None)?;
2426 0 : let request: TopTenantShardsRequest = json_request(&mut r).await?;
2427 0 : let state = get_state(&r);
2428 0 :
2429 0 : fn get_size_metric(sizes: &TopTenantShardItem, order_by: &TenantSorting) -> u64 {
2430 0 : match order_by {
2431 0 : TenantSorting::ResidentSize => sizes.resident_size,
2432 0 : TenantSorting::MaxLogicalSize => sizes.max_logical_size,
2433 0 : }
2434 0 : }
2435 0 :
2436 0 : #[derive(Eq, PartialEq)]
2437 0 : struct HeapItem {
2438 0 : metric: u64,
2439 0 : sizes: TopTenantShardItem,
2440 0 : }
2441 0 :
2442 0 : impl PartialOrd for HeapItem {
2443 0 : fn partial_cmp(&self, other: &Self) -> Option<std::cmp::Ordering> {
2444 0 : Some(self.cmp(other))
2445 0 : }
2446 0 : }
2447 0 :
2448 0 : /// Heap items have reverse ordering on their metric: this enables using BinaryHeap, which
2449 0 : /// supports popping the greatest item but not the smallest.
2450 0 : impl Ord for HeapItem {
2451 0 : fn cmp(&self, other: &Self) -> std::cmp::Ordering {
2452 0 : Reverse(self.metric).cmp(&Reverse(other.metric))
2453 0 : }
2454 0 : }
2455 0 :
2456 0 : let mut top_n: BinaryHeap<HeapItem> = BinaryHeap::with_capacity(request.limit);
2457 :
2458 : // FIXME: this is a lot of clones to take this tenant list
2459 0 : for (tenant_shard_id, tenant_slot) in state.tenant_manager.list() {
2460 0 : if let Some(shards_lt) = request.where_shards_lt {
2461 : // Ignore tenants which already have >= this many shards
2462 0 : if tenant_shard_id.shard_count >= shards_lt {
2463 0 : continue;
2464 0 : }
2465 0 : }
2466 :
2467 0 : let sizes = match tenant_slot {
2468 0 : TenantSlot::Attached(tenant) => tenant.get_sizes(),
2469 : TenantSlot::Secondary(_) | TenantSlot::InProgress(_) => {
2470 0 : continue;
2471 : }
2472 : };
2473 0 : let metric = get_size_metric(&sizes, &request.order_by);
2474 :
2475 0 : if let Some(gt) = request.where_gt {
2476 : // Ignore tenants whose metric is <= the lower size threshold, to do less sorting work
2477 0 : if metric <= gt {
2478 0 : continue;
2479 0 : }
2480 0 : };
2481 :
2482 0 : match top_n.peek() {
2483 0 : None => {
2484 0 : // Top N list is empty: candidate becomes first member
2485 0 : top_n.push(HeapItem { metric, sizes });
2486 0 : }
2487 0 : Some(i) if i.metric > metric && top_n.len() < request.limit => {
2488 0 : // Lowest item in list is greater than our candidate, but we aren't at limit yet: push to end
2489 0 : top_n.push(HeapItem { metric, sizes });
2490 0 : }
2491 0 : Some(i) if i.metric > metric => {
2492 0 : // List is at limit and lowest value is greater than our candidate, drop it.
2493 0 : }
2494 0 : Some(_) => top_n.push(HeapItem { metric, sizes }),
2495 : }
2496 :
2497 0 : while top_n.len() > request.limit {
2498 0 : top_n.pop();
2499 0 : }
2500 : }
2501 :
2502 0 : json_response(
2503 0 : StatusCode::OK,
2504 0 : TopTenantShardsResponse {
2505 0 : shards: top_n.into_iter().map(|i| i.sizes).collect(),
2506 0 : },
2507 0 : )
2508 0 : }
2509 :
2510 : /// Common functionality of all the HTTP API handlers.
2511 : ///
2512 : /// - Adds a tracing span to each request (by `request_span`)
2513 : /// - Logs the request depending on the request method (by `request_span`)
2514 : /// - Logs the response if it was not successful (by `request_span`
2515 : /// - Shields the handler function from async cancellations. Hyper can drop the handler
2516 : /// Future if the connection to the client is lost, but most of the pageserver code is
2517 : /// not async cancellation safe. This converts the dropped future into a graceful cancellation
2518 : /// request with a CancellationToken.
2519 0 : async fn api_handler<R, H>(request: Request<Body>, handler: H) -> Result<Response<Body>, ApiError>
2520 0 : where
2521 0 : R: std::future::Future<Output = Result<Response<Body>, ApiError>> + Send + 'static,
2522 0 : H: FnOnce(Request<Body>, CancellationToken) -> R + Send + Sync + 'static,
2523 0 : {
2524 0 : if request.uri() != &"/v1/failpoints".parse::<Uri>().unwrap() {
2525 0 : fail::fail_point!("api-503", |_| Err(ApiError::ResourceUnavailable(
2526 0 : "failpoint".into()
2527 0 : )));
2528 :
2529 0 : fail::fail_point!("api-500", |_| Err(ApiError::InternalServerError(
2530 0 : anyhow::anyhow!("failpoint")
2531 0 : )));
2532 0 : }
2533 :
2534 : // Spawn a new task to handle the request, to protect the handler from unexpected
2535 : // async cancellations. Most pageserver functions are not async cancellation safe.
2536 : // We arm a drop-guard, so that if Hyper drops the Future, we signal the task
2537 : // with the cancellation token.
2538 0 : let token = CancellationToken::new();
2539 0 : let cancel_guard = token.clone().drop_guard();
2540 0 : let result = request_span(request, move |r| async {
2541 0 : let handle = tokio::spawn(
2542 0 : async {
2543 0 : let token_cloned = token.clone();
2544 0 : let result = handler(r, token).await;
2545 0 : if token_cloned.is_cancelled() {
2546 0 : // dropguard has executed: we will never turn this result into response.
2547 0 : //
2548 0 : // at least temporarily do {:?} logging; these failures are rare enough but
2549 0 : // could hide difficult errors.
2550 0 : match &result {
2551 0 : Ok(response) => {
2552 0 : let status = response.status();
2553 0 : info!(%status, "Cancelled request finished successfully")
2554 0 : }
2555 0 : Err(e) => error!("Cancelled request finished with an error: {e:?}"),
2556 0 : }
2557 0 : }
2558 0 : // only logging for cancelled panicked request handlers is the tracing_panic_hook,
2559 0 : // which should suffice.
2560 0 : //
2561 0 : // there is still a chance to lose the result due to race between
2562 0 : // returning from here and the actual connection closing happening
2563 0 : // before outer task gets to execute. leaving that up for #5815.
2564 0 : result
2565 0 : }
2566 0 : .in_current_span(),
2567 0 : );
2568 0 :
2569 0 : match handle.await {
2570 0 : // TODO: never actually return Err from here, always Ok(...) so that we can log
2571 0 : // spanned errors. Call api_error_handler instead and return appropriate Body.
2572 0 : Ok(result) => result,
2573 0 : Err(e) => {
2574 0 : // The handler task panicked. We have a global panic handler that logs the
2575 0 : // panic with its backtrace, so no need to log that here. Only log a brief
2576 0 : // message to make it clear that we returned the error to the client.
2577 0 : error!("HTTP request handler task panicked: {e:#}");
2578 0 :
2579 0 : // Don't return an Error here, because then fallback error handler that was
2580 0 : // installed in make_router() will print the error. Instead, construct the
2581 0 : // HTTP error response and return that.
2582 0 : Ok(
2583 0 : ApiError::InternalServerError(anyhow!("HTTP request handler task panicked"))
2584 0 : .into_response(),
2585 0 : )
2586 0 : }
2587 0 : }
2588 0 : })
2589 0 : .await;
2590 :
2591 0 : cancel_guard.disarm();
2592 0 :
2593 0 : result
2594 0 : }
2595 :
2596 : /// Like api_handler, but returns an error response if the server is built without
2597 : /// the 'testing' feature.
2598 0 : async fn testing_api_handler<R, H>(
2599 0 : desc: &str,
2600 0 : request: Request<Body>,
2601 0 : handler: H,
2602 0 : ) -> Result<Response<Body>, ApiError>
2603 0 : where
2604 0 : R: std::future::Future<Output = Result<Response<Body>, ApiError>> + Send + 'static,
2605 0 : H: FnOnce(Request<Body>, CancellationToken) -> R + Send + Sync + 'static,
2606 0 : {
2607 0 : if cfg!(feature = "testing") {
2608 0 : api_handler(request, handler).await
2609 : } else {
2610 0 : std::future::ready(Err(ApiError::BadRequest(anyhow!(
2611 0 : "Cannot {desc} because pageserver was compiled without testing APIs",
2612 0 : ))))
2613 0 : .await
2614 : }
2615 0 : }
2616 :
2617 0 : pub fn make_router(
2618 0 : state: Arc<State>,
2619 0 : launch_ts: &'static LaunchTimestamp,
2620 0 : auth: Option<Arc<SwappableJwtAuth>>,
2621 0 : ) -> anyhow::Result<RouterBuilder<hyper::Body, ApiError>> {
2622 0 : let spec = include_bytes!("openapi_spec.yml");
2623 0 : let mut router = attach_openapi_ui(endpoint::make_router(), spec, "/swagger.yml", "/v1/doc");
2624 0 : if auth.is_some() {
2625 0 : router = router.middleware(auth_middleware(|request| {
2626 0 : let state = get_state(request);
2627 0 : if state.allowlist_routes.contains(request.uri()) {
2628 0 : None
2629 : } else {
2630 0 : state.auth.as_deref()
2631 : }
2632 0 : }))
2633 0 : }
2634 :
2635 0 : router = router.middleware(
2636 0 : endpoint::add_response_header_middleware(
2637 0 : "PAGESERVER_LAUNCH_TIMESTAMP",
2638 0 : &launch_ts.to_string(),
2639 0 : )
2640 0 : .expect("construct launch timestamp header middleware"),
2641 0 : );
2642 0 :
2643 0 : Ok(router
2644 0 : .data(state)
2645 0 : .get("/metrics", |r| request_span(r, prometheus_metrics_handler))
2646 0 : .get("/v1/status", |r| api_handler(r, status_handler))
2647 0 : .put("/v1/failpoints", |r| {
2648 0 : testing_api_handler("manage failpoints", r, failpoints_handler)
2649 0 : })
2650 0 : .post("/v1/reload_auth_validation_keys", |r| {
2651 0 : api_handler(r, reload_auth_validation_keys_handler)
2652 0 : })
2653 0 : .get("/v1/tenant", |r| api_handler(r, tenant_list_handler))
2654 0 : .post("/v1/tenant", |r| api_handler(r, tenant_create_handler))
2655 0 : .get("/v1/tenant/:tenant_shard_id", |r| {
2656 0 : api_handler(r, tenant_status)
2657 0 : })
2658 0 : .delete("/v1/tenant/:tenant_shard_id", |r| {
2659 0 : api_handler(r, tenant_delete_handler)
2660 0 : })
2661 0 : .get("/v1/tenant/:tenant_shard_id/synthetic_size", |r| {
2662 0 : api_handler(r, tenant_size_handler)
2663 0 : })
2664 0 : .put("/v1/tenant/config", |r| {
2665 0 : api_handler(r, update_tenant_config_handler)
2666 0 : })
2667 0 : .put("/v1/tenant/:tenant_shard_id/shard_split", |r| {
2668 0 : api_handler(r, tenant_shard_split_handler)
2669 0 : })
2670 0 : .get("/v1/tenant/:tenant_shard_id/config", |r| {
2671 0 : api_handler(r, get_tenant_config_handler)
2672 0 : })
2673 0 : .put("/v1/tenant/:tenant_shard_id/location_config", |r| {
2674 0 : api_handler(r, put_tenant_location_config_handler)
2675 0 : })
2676 0 : .get("/v1/location_config", |r| {
2677 0 : api_handler(r, list_location_config_handler)
2678 0 : })
2679 0 : .get("/v1/location_config/:tenant_shard_id", |r| {
2680 0 : api_handler(r, get_location_config_handler)
2681 0 : })
2682 0 : .put(
2683 0 : "/v1/tenant/:tenant_shard_id/time_travel_remote_storage",
2684 0 : |r| api_handler(r, tenant_time_travel_remote_storage_handler),
2685 0 : )
2686 0 : .get("/v1/tenant/:tenant_shard_id/timeline", |r| {
2687 0 : api_handler(r, timeline_list_handler)
2688 0 : })
2689 0 : .post("/v1/tenant/:tenant_shard_id/timeline", |r| {
2690 0 : api_handler(r, timeline_create_handler)
2691 0 : })
2692 0 : .post("/v1/tenant/:tenant_id/attach", |r| {
2693 0 : api_handler(r, tenant_attach_handler)
2694 0 : })
2695 0 : .post("/v1/tenant/:tenant_id/detach", |r| {
2696 0 : api_handler(r, tenant_detach_handler)
2697 0 : })
2698 0 : .post("/v1/tenant/:tenant_shard_id/reset", |r| {
2699 0 : api_handler(r, tenant_reset_handler)
2700 0 : })
2701 0 : .post(
2702 0 : "/v1/tenant/:tenant_shard_id/timeline/:timeline_id/preserve_initdb_archive",
2703 0 : |r| api_handler(r, timeline_preserve_initdb_handler),
2704 0 : )
2705 0 : .get("/v1/tenant/:tenant_shard_id/timeline/:timeline_id", |r| {
2706 0 : api_handler(r, timeline_detail_handler)
2707 0 : })
2708 0 : .get(
2709 0 : "/v1/tenant/:tenant_shard_id/timeline/:timeline_id/get_lsn_by_timestamp",
2710 0 : |r| api_handler(r, get_lsn_by_timestamp_handler),
2711 0 : )
2712 0 : .get(
2713 0 : "/v1/tenant/:tenant_shard_id/timeline/:timeline_id/get_timestamp_of_lsn",
2714 0 : |r| api_handler(r, get_timestamp_of_lsn_handler),
2715 0 : )
2716 0 : .post(
2717 0 : "/v1/tenant/:tenant_shard_id/timeline/:timeline_id/lsn_lease",
2718 0 : |r| api_handler(r, lsn_lease_handler),
2719 0 : )
2720 0 : .put(
2721 0 : "/v1/tenant/:tenant_shard_id/timeline/:timeline_id/do_gc",
2722 0 : |r| api_handler(r, timeline_gc_handler),
2723 0 : )
2724 0 : .put(
2725 0 : "/v1/tenant/:tenant_shard_id/timeline/:timeline_id/compact",
2726 0 : |r| testing_api_handler("run timeline compaction", r, timeline_compact_handler),
2727 0 : )
2728 0 : .put(
2729 0 : "/v1/tenant/:tenant_shard_id/timeline/:timeline_id/checkpoint",
2730 0 : |r| testing_api_handler("run timeline checkpoint", r, timeline_checkpoint_handler),
2731 0 : )
2732 0 : .post(
2733 0 : "/v1/tenant/:tenant_shard_id/timeline/:timeline_id/download_remote_layers",
2734 0 : |r| api_handler(r, timeline_download_remote_layers_handler_post),
2735 0 : )
2736 0 : .get(
2737 0 : "/v1/tenant/:tenant_shard_id/timeline/:timeline_id/download_remote_layers",
2738 0 : |r| api_handler(r, timeline_download_remote_layers_handler_get),
2739 0 : )
2740 0 : .put(
2741 0 : "/v1/tenant/:tenant_shard_id/timeline/:timeline_id/detach_ancestor",
2742 0 : |r| api_handler(r, timeline_detach_ancestor_handler),
2743 0 : )
2744 0 : .delete("/v1/tenant/:tenant_shard_id/timeline/:timeline_id", |r| {
2745 0 : api_handler(r, timeline_delete_handler)
2746 0 : })
2747 0 : .get(
2748 0 : "/v1/tenant/:tenant_shard_id/timeline/:timeline_id/layer",
2749 0 : |r| api_handler(r, layer_map_info_handler),
2750 0 : )
2751 0 : .get(
2752 0 : "/v1/tenant/:tenant_shard_id/timeline/:timeline_id/layer/:layer_file_name",
2753 0 : |r| api_handler(r, layer_download_handler),
2754 0 : )
2755 0 : .delete(
2756 0 : "/v1/tenant/:tenant_shard_id/timeline/:timeline_id/layer/:layer_file_name",
2757 0 : |r| api_handler(r, evict_timeline_layer_handler),
2758 0 : )
2759 0 : .post("/v1/tenant/:tenant_shard_id/heatmap_upload", |r| {
2760 0 : api_handler(r, secondary_upload_handler)
2761 0 : })
2762 0 : .get("/v1/tenant/:tenant_id/scan_remote_storage", |r| {
2763 0 : api_handler(r, tenant_scan_remote_handler)
2764 0 : })
2765 0 : .put("/v1/disk_usage_eviction/run", |r| {
2766 0 : api_handler(r, disk_usage_eviction_run)
2767 0 : })
2768 0 : .put("/v1/deletion_queue/flush", |r| {
2769 0 : api_handler(r, deletion_queue_flush)
2770 0 : })
2771 0 : .get("/v1/tenant/:tenant_shard_id/secondary/status", |r| {
2772 0 : api_handler(r, secondary_status_handler)
2773 0 : })
2774 0 : .post("/v1/tenant/:tenant_shard_id/secondary/download", |r| {
2775 0 : api_handler(r, secondary_download_handler)
2776 0 : })
2777 0 : .put("/v1/tenant/:tenant_shard_id/break", |r| {
2778 0 : testing_api_handler("set tenant state to broken", r, handle_tenant_break)
2779 0 : })
2780 0 : .get("/v1/panic", |r| api_handler(r, always_panic_handler))
2781 0 : .post("/v1/tracing/event", |r| {
2782 0 : testing_api_handler("emit a tracing event", r, post_tracing_event_handler)
2783 0 : })
2784 0 : .get(
2785 0 : "/v1/tenant/:tenant_shard_id/timeline/:timeline_id/getpage",
2786 0 : |r| testing_api_handler("getpage@lsn", r, getpage_at_lsn_handler),
2787 0 : )
2788 0 : .get(
2789 0 : "/v1/tenant/:tenant_shard_id/timeline/:timeline_id/keyspace",
2790 0 : |r| api_handler(r, timeline_collect_keyspace),
2791 0 : )
2792 0 : .put("/v1/io_engine", |r| api_handler(r, put_io_engine_handler))
2793 0 : .put(
2794 0 : "/v1/tenant/:tenant_shard_id/timeline/:timeline_id/force_aux_policy_switch",
2795 0 : |r| api_handler(r, force_aux_policy_switch_handler),
2796 0 : )
2797 0 : .get("/v1/utilization", |r| api_handler(r, get_utilization))
2798 0 : .post(
2799 0 : "/v1/tenant/:tenant_shard_id/timeline/:timeline_id/ingest_aux_files",
2800 0 : |r| testing_api_handler("ingest_aux_files", r, ingest_aux_files),
2801 0 : )
2802 0 : .post(
2803 0 : "/v1/tenant/:tenant_shard_id/timeline/:timeline_id/list_aux_files",
2804 0 : |r| testing_api_handler("list_aux_files", r, list_aux_files),
2805 0 : )
2806 0 : .post("/v1/top_tenants", |r| api_handler(r, post_top_tenants))
2807 0 : .post(
2808 0 : "/v1/tenant/:tenant_shard_id/timeline/:timeline_id/perf_info",
2809 0 : |r| testing_api_handler("perf_info", r, perf_info),
2810 0 : )
2811 0 : .any(handler_404))
2812 0 : }
|