feat: Milestone 1 - added replication plumbing, tests and shared helpers by jimezesinachi · Pull Request #332 · deven96/ahnlich

jimezesinachi · 2026-05-08T15:22:25Z

No description provided.

Signed-off-by: Jim Ezesinachi <ezesinachijim@gmail.com>

github-actions · 2026-05-08T15:38:12Z

Test Results

362 tests 362 ✅ 19m 20s ⏱️
42 suites 0 💤
4 files 0 ❌

Results for commit 3029cb0.

♻️ This comment has been updated with latest results.

Signed-off-by: Jim Ezesinachi <ezesinachijim@gmail.com>

github-actions · 2026-05-08T17:19:16Z

Benchmark Results

group                                                        main                                   pr
-----                                                        ----                                   --
predicate_query_with_index/size_100                          1.02      3.1±0.00µs        ? ?/sec    1.00      3.0±0.00µs        ? ?/sec
predicate_query_with_index/size_1000                         1.00     31.6±0.05µs        ? ?/sec    1.05     33.1±0.01µs        ? ?/sec
predicate_query_with_index/size_10000                        1.00    400.2±2.28µs        ? ?/sec    1.00    398.6±0.25µs        ? ?/sec
predicate_query_with_index/size_100000                       1.02      5.7±0.70ms        ? ?/sec    1.00      5.6±0.38ms        ? ?/sec
predicate_query_without_index/size_100                       1.02      7.2±0.00µs        ? ?/sec    1.00      7.0±0.01µs        ? ?/sec
predicate_query_without_index/size_1000                      1.05    106.3±0.15µs        ? ?/sec    1.00    101.7±0.06µs        ? ?/sec
predicate_query_without_index/size_10000                     1.02    772.0±1.36µs        ? ?/sec    1.00    755.8±2.77µs        ? ?/sec
predicate_query_without_index/size_100000                    1.00     12.7±0.86ms        ? ?/sec    1.07     13.6±1.30ms        ? ?/sec
store_batch_insertion_without_predicates/size_100            1.01    187.4±1.50µs        ? ?/sec    1.00    186.4±1.75µs        ? ?/sec
store_batch_insertion_without_predicates/size_1000           1.02   1256.7±8.17µs        ? ?/sec    1.00  1229.6±14.57µs        ? ?/sec
store_batch_insertion_without_predicates/size_10000          1.03     14.0±0.09ms        ? ?/sec    1.00     13.6±0.10ms        ? ?/sec
store_batch_insertion_without_predicates/size_100000         1.03    137.5±0.79ms        ? ?/sec    1.00    134.1±0.77ms        ? ?/sec
store_retrieval_linear_cosine_similarity/size_100            1.00     85.4±1.66µs        ? ?/sec    1.04     88.7±2.71µs        ? ?/sec
store_retrieval_linear_cosine_similarity/size_1000           1.00    719.1±2.75µs        ? ?/sec    1.00    717.8±3.01µs        ? ?/sec
store_retrieval_linear_cosine_similarity/size_10000          1.00      7.2±0.02ms        ? ?/sec    1.00      7.2±0.03ms        ? ?/sec
store_retrieval_linear_cosine_similarity/size_100000         1.00     75.6±0.35ms        ? ?/sec    1.00     75.4±0.42ms        ? ?/sec
store_retrieval_linear_dot_product/size_100                  1.00     84.9±1.87µs        ? ?/sec    1.01     85.6±3.50µs        ? ?/sec
store_retrieval_linear_dot_product/size_1000                 1.00    707.9±3.40µs        ? ?/sec    1.00    706.6±3.10µs        ? ?/sec
store_retrieval_linear_dot_product/size_10000                1.01      7.1±0.03ms        ? ?/sec    1.00      7.0±0.01ms        ? ?/sec
store_retrieval_linear_dot_product/size_100000               1.00     73.6±0.29ms        ? ?/sec    1.00     73.3±0.23ms        ? ?/sec
store_retrieval_linear_euclidean_distance/size_100           1.00     83.7±1.20µs        ? ?/sec    1.02     85.0±1.43µs        ? ?/sec
store_retrieval_linear_euclidean_distance/size_1000          1.00    700.4±2.93µs        ? ?/sec    1.01    708.0±4.13µs        ? ?/sec
store_retrieval_linear_euclidean_distance/size_10000         1.00      7.1±0.07ms        ? ?/sec    1.00      7.1±0.04ms        ? ?/sec
store_retrieval_linear_euclidean_distance/size_100000        1.00     72.9±0.43ms        ? ?/sec    1.02     74.2±0.24ms        ? ?/sec
store_retrieval_no_condition/size_100                        1.02     86.7±2.58µs        ? ?/sec    1.00     85.0±1.29µs        ? ?/sec
store_retrieval_no_condition/size_1000                       1.00    724.0±3.21µs        ? ?/sec    1.00    722.7±3.41µs        ? ?/sec
store_retrieval_no_condition/size_10000                      1.00      7.3±0.03ms        ? ?/sec    1.01      7.3±0.09ms        ? ?/sec
store_retrieval_no_condition/size_100000                     1.02     77.3±0.56ms        ? ?/sec    1.00     75.7±0.49ms        ? ?/sec
store_retrieval_non_linear_hnsw/size_100                     1.00    169.4±0.39µs        ? ?/sec    1.00    169.5±0.30µs        ? ?/sec
store_retrieval_non_linear_hnsw/size_1000                    1.00    485.2±2.18µs        ? ?/sec    1.02    493.4±0.84µs        ? ?/sec
store_retrieval_non_linear_hnsw/size_10000                   1.00      2.1±0.04ms        ? ?/sec    1.00      2.1±0.03ms        ? ?/sec
store_retrieval_non_linear_hnsw/size_100000                  1.07     12.4±0.16ms        ? ?/sec    1.00     11.6±0.10ms        ? ?/sec
store_retrieval_non_linear_kdtree/size_100                   1.00    179.8±0.54µs        ? ?/sec    1.00    180.2±0.16µs        ? ?/sec
store_retrieval_non_linear_kdtree/size_1000                  1.00   1155.5±2.17µs        ? ?/sec    1.00   1153.7±1.93µs        ? ?/sec
store_retrieval_non_linear_kdtree/size_10000                 1.01     12.4±0.07ms        ? ?/sec    1.00     12.3±0.10ms        ? ?/sec
store_retrieval_non_linear_kdtree/size_100000                1.01    149.5±0.90ms        ? ?/sec    1.00    148.6±1.02ms        ? ?/sec
store_sequential_insertion_without_predicates/size_100       1.00    260.8±0.24µs        ? ?/sec    1.01    262.5±0.17µs        ? ?/sec
store_sequential_insertion_without_predicates/size_1000      1.00      2.6±0.00ms        ? ?/sec    1.00      2.6±0.01ms        ? ?/sec
store_sequential_insertion_without_predicates/size_10000     1.00     25.6±0.05ms        ? ?/sec    1.00     25.7±0.02ms        ? ?/sec
store_sequential_insertion_without_predicates/size_100000    1.00    256.2±0.10ms        ? ?/sec    1.00    256.5±0.17ms        ? ?/sec

deven96 · 2026-05-12T09:35:18Z

+    DelPred(Vec<u8>),
+    DropPredIndex(Vec<u8>),
+    DropNonLinearAlgorithmIndex(Vec<u8>),
+    DropStore(Vec<u8>),


I thought we added PurgeStores recently to DB?

A'firm, good catch

Turns out, there appears to be no purge_stores handler for DB. There is one for AI, but none for DB. See here: https://github.com/deven96/ahnlich/blob/main/ahnlich/db/src/server/handler.rs

Oh schade... we definitely don't have then
Skip for now

deven96 · 2026-05-12T10:01:01Z

@@ -0,0 +1,425 @@
+// `clippy::result_large_err` and `clippy::type_complexity` fire on signatures


Prefix mod with a definition of what the statemachine is and what it does

Signed-off-by: Jim Ezesinachi <ezesinachijim@gmail.com>

…even96/ahnlich into feat/replication-m1-foundation

deven96 · 2026-05-13T17:37:12Z

+        let mut inner = self.inner.lock().expect("state machine lock poisoned");
+        let mut responses = Vec::new();
+
+        for entry in entries {


A bit skeptical about running the entries in a loop to inner.handler.apply. If there's an error with applying entry in the middle, we return error but have already applied some previous entries?

A bit skeptical about running the entries in a loop to inner.handler.apply. If there's an error with applying entry in the middle, we return error but have already applied some previous entries?

Shey, I see what you're saying. I'm thinking of potential solutions: maybe rollback, or just try and check and fail early somehow?

Yeah maybe fail early for a start as I'm not sure we have some sure fire way to rollback yet

deven96 · 2026-05-13T17:39:17Z

+            source: StorageIOError::read_state_machine(&e),
+        })?;
+
+        let mut inner = self.inner.lock().expect("state machine lock poisoned");


Curious about panicking as a whole with respect to lock poisoning of the state machine

Shouldn't it be captured with StorageError instead?

Curious about panicking as a whole with respect to lock poisoning of the state machine

Shouldn't it be captured with StorageError instead?

I see what you're saying. I think my thinking here was that if lock is poisoned, the state machine is corrupt and in a way "unrecoverable until node restart". But user API wise, it does make sense to just bubble up as a StorageError, possibly with a message saying "the state machine lock is poisoned"

deven96 · 2026-05-13T17:42:29Z

+        opts.create_if_missing(true);
+        opts.create_missing_column_families(true);
+        let cfs = vec![
+            ColumnFamilyDescriptor::new("logs", Options::default()),


Would rather we have static strings referencing the column names and reuse across the rocksdb mod

Makes sense. I'll use constants

deven96 · 2026-05-13T17:53:21Z

+        })
+    }
+
+    fn save_vote_sync(&mut self, vote: &Vote<C::NodeId>) -> Result<(), StorageError<C::NodeId>> {


Perhaps a stronger harness around setting&getting meta.... I'm largely wary about having to repeat strings as there's a huge chance that with an update things can be missed

Makes sense. I'll use helper functions here

deven96 · 2026-05-13T18:10:14Z

+        meta: &SnapshotMeta<C::NodeId, C::Node>,
+        snapshot: Box<C::SnapshotData>,
+    ) -> Result<(), StorageError<C::NodeId>> {
+        let cursor: Cursor<Vec<u8>> = (*snapshot).clone().into();


Do we have to clone the entire snapshot in order to deserialize it?

Looking again, I'm not sure. Will take a closer look and fix

Signed-off-by: Jim Ezesinachi <ezesinachijim@gmail.com>

feat: Milestone 1 - added replication plumbing, tests and shared helpers

ddb112e

Signed-off-by: Jim Ezesinachi <ezesinachijim@gmail.com>

jimezesinachi added 2 commits May 8, 2026 17:47

updated sdk proto generated types

257ac2b

Signed-off-by: Jim Ezesinachi <ezesinachijim@gmail.com>

format go files

36faff4

Signed-off-by: Jim Ezesinachi <ezesinachijim@gmail.com>

jimezesinachi requested review from Iamdavidonuh and deven96 and removed request for Iamdavidonuh May 9, 2026 07:50

deven96 reviewed May 12, 2026

View reviewed changes

jimezesinachi added 3 commits May 13, 2026 08:08

Merge branch 'main' into feat/replication-m1-foundation

8c73788

added module comments to state_machine

42d6d5d

Signed-off-by: Jim Ezesinachi <ezesinachijim@gmail.com>

Merge branch 'feat/replication-m1-foundation' of https://github.com/d…

1e2fff5

…even96/ahnlich into feat/replication-m1-foundation

deven96 reviewed May 13, 2026

View reviewed changes

jimezesinachi added 2 commits May 15, 2026 20:31

Merge branch 'main' into feat/replication-m1-foundation

7da7f76

updates based on comments

3029cb0

Signed-off-by: Jim Ezesinachi <ezesinachijim@gmail.com>

		@@ -0,0 +1,425 @@
		// `clippy::result_large_err` and `clippy::type_complexity` fire on signatures

Conversation

jimezesinachi commented May 8, 2026

Uh oh!

github-actions Bot commented May 8, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Test Results

Uh oh!

github-actions Bot commented May 8, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Benchmark Results

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

deven96 May 12, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

jimezesinachi May 15, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

jimezesinachi May 15, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

github-actions Bot commented May 8, 2026 •

edited

Loading

github-actions Bot commented May 8, 2026 •

edited

Loading

deven96 May 12, 2026 •

edited

Loading

jimezesinachi May 15, 2026 •

edited

Loading

jimezesinachi May 15, 2026 •

edited

Loading