Skip to main content

harmos_storage/
deletion.rs

1/* ***
2 *
3 * Harmos
4 *
5 * SPDX-License-Identifier: Apache-2.0
6 *
7 * ***
8 */
9
10/* === Modules === */
11
12/* === Re-Exports === */
13
14/* === Imports === */
15
16// Standard Library Imports
17use std::sync::Arc;
18
19// Third Party Imports
20use sha2::{Digest, Sha256};
21
22// Local Imports
23use super::config::MAX_MUTATIONS;
24use super::manager::{Inner, State, Target};
25use super::receipts::ReceiptState;
26use super::{MutationId, MutationReceipt, MutationValue, StagedFile, StorageManager};
27use crate::{Error, Identifier, MemorySize, Result};
28
29/* === Constants === */
30
31/* === Macros === */
32
33/* === Type Aliases === */
34
35/* === Structs === */
36
37pub(super) struct Deletion {
38    pub(super) work: Arc<DeleteWork>,
39    pub(super) bytes: usize,
40    failure: Option<Arc<Error>>,
41}
42
43pub(super) struct DeleteWork {
44    pub(super) context: crate::execution::Context,
45    pub(super) queued: std::sync::Mutex<Option<tracing::Span>>,
46    target: Target,
47    pub(super) file: StagedFile,
48    receipt: crate::StorageBatch<MutationReceipt>,
49}
50
51/* === Enums === */
52
53/* === Traits === */
54
55/* === Implementations === */
56
57impl Drop for DeleteWork {
58    fn drop(&mut self) {
59        self.context.sync(|| drop(self.queued.get_mut().unwrap().take()));
60    }
61}
62
63impl<C> StorageManager<C> {
64    #[tracing::instrument(name = "storage.deletion.delete_file_once", level = "debug", skip_all, fields(outcome = "cancelled"))]
65    pub async fn delete_file_once(&self, mutation: &MutationId, file: &StagedFile) -> Result<StagedFile> {
66        let result = async {
67            self.inner.primary.history().check(&file.history)?;
68            if !self.inner.primary.supports_mutation_receipts() || !self.inner.primary.supports_file_deletion() {
69                return Err(invalid("backend must support durable deletion"));
70            }
71            let fingerprint = fingerprint(file);
72            let mut retry = true;
73            let mut completion = None;
74            let mut changed = self.inner.changed.subscribe();
75            loop {
76                let mut notify = false;
77                {
78                    let mut state = self.inner.state.lock().unwrap();
79                    if state.status.closing {
80                        return Err(Error::Closing { domain: "file deletion" });
81                    }
82                    if !state.ready || !state.status.running {
83                        return Err(Error::NotReady { domain: "file deletion" });
84                    }
85                    match state.receipts.lookup(mutation, &fingerprint, completion.as_ref())? {
86                        ReceiptState::Complete(value) => return value.deleted(),
87                        ReceiptState::Pending => {
88                            if completion.is_none() {
89                                let owner = state.receipts.completion(mutation);
90                                completion = Some(owner);
91                            }
92                            let deletion = state.deletions.get(mutation).ok_or_else(|| invalid("pending identity owns different work"))?;
93                            if let Some(source) = deletion.failure.clone() {
94                                if !retry {
95                                    return Err(Error::Persistence { source });
96                                }
97                                if let Some(other) = &state.status.failure
98                                    && !Arc::ptr_eq(other, &source)
99                                {
100                                    return Err(Error::Persistence { source: other.clone() });
101                                }
102                                state.deletions.get_mut(mutation).unwrap().failure = None;
103                                state.status.failure = None;
104                                state.retry = state.retry.wrapping_add(1);
105                                notify = true;
106                            } else if let Some(source) = &state.status.failure {
107                                return Err(Error::Persistence { source: source.clone() });
108                            }
109                            retry = false;
110                        }
111                        ReceiptState::Missing => {
112                            if let Some(source) = &state.status.failure {
113                                return Err(Error::Persistence { source: source.clone() });
114                            }
115                            if state.files.values().any(|active| active.identifies(file)) {
116                                tracing::error!(file = %file.id, reason = "active producer", "file deletion rejected");
117                                return Err(invalid("active production must finish or abandon before deletion"));
118                            }
119                            if state.deleting(&file.id) {
120                                return Err(invalid("another deletion owns this file"));
121                            }
122                            if state.deletions.len() >= MAX_MUTATIONS.get() {
123                                return Err(exhausted());
124                            }
125                            let bytes = crate::btree_map_memory::<MutationId, Deletion>(1)
126                                .saturating_add(std::mem::size_of::<DeleteWork>())
127                                .saturating_add(std::mem::size_of::<MutationReceipt>())
128                                .saturating_add(std::mem::size_of::<MutationValue>())
129                                .saturating_add(mutation.memory_size().saturating_mul(2))
130                                .saturating_add(file.memory_size().saturating_mul(2))
131                                .saturating_add(2 * std::mem::size_of::<usize>());
132                            state.collect();
133                            let capacity = self.inner.settings.budget.read_memory_bytes.get();
134                            if bytes > capacity || state.status.read_bytes > capacity - bytes {
135                                return Err(exhausted());
136                            }
137                            let outcome = MutationValue::Deleted(file.clone());
138                            state.receipts.insert(mutation, fingerprint, outcome.memory_size())?;
139                            let owner = state.receipts.completion(mutation);
140                            completion = Some(owner);
141                            let receipt = state.receipts.prepared(mutation, outcome);
142                            let target = state.data_accepted;
143                            state.deletions.insert(
144                                mutation.clone(),
145                                Deletion {
146                                    work: Arc::new(DeleteWork {
147                                        context: crate::execution::Context::current(),
148                                        queued: std::sync::Mutex::new(Some(tracing::debug_span!("storage.deletion.queue"))),
149                                        target,
150                                        file: file.clone(),
151                                        receipt: crate::StorageBatch::new(vec![receipt])
152                                            .retained(crate::batch::WriteCharge::read(&self.inner, bytes)),
153                                    }),
154                                    bytes,
155                                    failure: None,
156                                },
157                            );
158                            retry = false;
159                            notify = true;
160                        }
161                    }
162                    state.collect();
163                }
164                if notify {
165                    self.inner.notify();
166                }
167                changed.changed().await.map_err(|_| Error::Closing { domain: "file deletion" })?;
168            }
169        }
170        .await;
171        tracing::Span::current().record("outcome", if result.is_ok() { "completed" } else { "failed" });
172        result
173    }
174
175    #[tracing::instrument(name = "storage.deletion.settle_deletions", level = "debug", skip_all)]
176    pub(super) async fn settle_deletions(&self) -> Result<()> {
177        let mut changed = self.inner.changed.subscribe();
178        {
179            let mut state = self.inner.state.lock().unwrap();
180            if state.deletions.is_empty() {
181                return Ok(());
182            }
183            for deletion in state.deletions.values_mut() {
184                deletion.failure = None;
185            }
186            state.status.failure = None;
187            state.retry = state.retry.wrapping_add(1);
188        }
189        self.inner.notify();
190        loop {
191            {
192                let state = self.inner.state.lock().unwrap();
193                if state.deletions.is_empty() {
194                    return Ok(());
195                }
196                if let Some(source) = &state.status.failure {
197                    return Err(Error::Persistence { source: source.clone() });
198                }
199                if !state.status.running {
200                    return Err(Error::NotReady { domain: "file deletion worker" });
201                }
202            }
203            changed.changed().await.map_err(|_| Error::Closing { domain: "file deletion" })?;
204        }
205    }
206}
207
208impl State {
209    pub(super) fn deleting(&self, file: &Identifier) -> bool {
210        self.deletions.values().any(|deletion| deletion.work.file.id == *file)
211    }
212
213    pub(super) fn request_deletions(&mut self) {
214        for deletion in self.deletions.values().filter(|deletion| deletion.failure.is_none()) {
215            let target = deletion.work.target;
216            if target.index > self.data_flushed && self.data_request.is_none_or(|current| target.index < current.index) {
217                self.data_request = Some(target);
218            }
219        }
220    }
221
222    pub(super) fn deletion_action(&self) -> Option<Arc<DeleteWork>> {
223        self.deletions
224            .values()
225            .find(|deletion| deletion.failure.is_none() && deletion.work.target.index <= self.data_flushed)
226            .map(|deletion| deletion.work.clone())
227    }
228
229    pub(super) fn deleting_data(&self, index: u64) -> bool {
230        self.deletions.values().any(|deletion| deletion.failure.is_none() && deletion.work.target.index >= index)
231    }
232
233    pub(super) fn deletion_failed(&mut self, mutation: &MutationId, source: Arc<Error>) {
234        if let Some(deletion) = self.deletions.get_mut(mutation) {
235            tracing::error!(file = %deletion.work.file.id, window = %mutation.window, mutation = %mutation.key, reason = "file deletion failed", "file deletion retained for retry");
236            deletion.failure = Some(source);
237        }
238    }
239
240    pub(super) fn deletion_data_failed(&mut self, index: u64, source: Arc<Error>) {
241        for deletion in self.deletions.values_mut().filter(|deletion| deletion.work.target.index >= index) {
242            deletion.failure = Some(source.clone());
243        }
244    }
245}
246
247impl Drop for Deletion {
248    fn drop(&mut self) {
249        self.work.receipt.retire();
250    }
251}
252
253impl DeleteWork {
254    pub(super) fn receipt(&self) -> &MutationReceipt {
255        &self.receipt.as_slice()[0]
256    }
257
258    #[tracing::instrument(name = "storage.deletion.perform", level = "debug", skip_all)]
259    pub(super) async fn perform(&self, inner: &Arc<Inner>) -> Result<()> {
260        inner.bounded("primary.delete_file", inner.primary.delete_file(&self.file)).await?;
261        inner.bounded("primary.save_mutations", inner.primary.save_mutations(self.receipt.clone())).await?;
262        tracing::info!(file = %self.file.id, mutation = %self.receipt.as_slice()[0].mutation.key, reason = "explicit deletion", "file deleted");
263        Ok(())
264    }
265}
266
267/* === Public Functions === */
268
269/* === Private Functions === */
270
271fn fingerprint(file: &StagedFile) -> [u8; 32] {
272    let mut hash = Sha256::new();
273    hash.update(b"storage.delete_file:v1\0");
274    for bytes in [file.history.store.as_str().as_bytes(), file.history.epoch.as_str().as_bytes(), file.id.as_str().as_bytes()] {
275        hash.update((bytes.len() as u64).to_le_bytes());
276        hash.update(bytes);
277    }
278    hash.finalize().into()
279}
280fn invalid(reason: &'static str) -> Error {
281    Error::InvalidInput { field: "file deletion", reason }
282}
283fn exhausted() -> Error {
284    Error::ResourceExhausted { resource: "file deletion requests" }
285}
286
287/* === Tests === */