[mobile_config] track owner refactoring #1131

kurotych · 2026-02-05T12:32:40Z

The main change is creating a new hash calculation based on the Gateway fields instead of counterintuitive hashing in mobile_hotspot_info and then saving it into the gateways table.
The additional reason for creating a new hash mechanism is using the owner field for tracking changes that is not part of the mobile_hotspot_info table. Now if owner changed then hash is recalculated.

Additional improvements:

Optimize MobileHotspotInfo::stream (get rid of cbrs in stream)
Add gateway's HashParams which makes clear what parameters are participating in hash calculation
Drastically simplifying and optimize an execute function

P.S.
The migration was added that set all hash to NULL in gateways table,
during startup the backfill_hashes function will be executed the recalculates hashes (locally it takes less then 1 min)

…e_type

Copilot

Pull request overview

Refactors the mobile gateway tracking/data model to consolidate hash-related fields (including owner) under a new HashParams container, updates integration tests accordingly, and adds a temporary runtime backfill to repair missing hashes during a migration.

Changes:

Move gateway “hash inputs” (type/location/deployment/owner/etc.) into HashParams and update call sites to use accessor methods.
Add a temporary backfill_hashes path in the gateway tracker to populate missing hashes.
Add a migration that nulls out gateways.hash and update integration tests to reflect the refactor (e.g., removed inserted_at in test builders/usages).

Reviewed changes

Copilot reviewed 14 out of 14 changed files in this pull request and generated 6 comments.

Show a summary per file

File	Description
mobile_config/src/gateway/db.rs	Introduces `HashParams`, refactors `Gateway` fields/accessors, adds null-hash backfill helpers.
mobile_config/src/gateway/tracker.rs	Adds `backfill_hashes` and refactors `execute` to use new `Gateway` construction/change detection.
mobile_config/src/gateway/metadata_db.rs	Refactors `MobileHotspotInfo` to expose fields needed by tracker/db and filters CBRS in SQL.
mobile_config/src/gateway/service/info.rs	Updates gateway info conversion to use new accessors.
mobile_config/src/gateway/service/info_v3.rs	Updates v3 conversion to use new accessors.
mobile_config/src/gateway/service/info_v4.rs	Updates v4 conversion to use new accessors and owner sourcing.
mobile_config/migrations/20260209112231_null_hash.sql	Adds migration that sets `gateways.hash` to NULL (intended to trigger backfill).
mobile_config/tests/integrations/*.rs	Updates integration tests to use new accessors and `HashParams` construction; removes `inserted_at` usage in builders.

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

Copilot · 2026-02-09T11:56:51Z

mobile_config/src/gateway/tracker.rs

+        backfill_hashes(&self.pool, &self.metadata).await?;
        loop {


backfill_hashes introduces migration-critical behavior (repairing NULL hashes) but there are no integration/unit tests covering it (existing integration tests call tracker::execute directly and won’t exercise the new backfill path). Adding a focused test that sets gateways.hash to NULL and asserts backfill_hashes updates the latest row’s hash (and does not regress execute) would help prevent a production-only failure during/after the migration.

Copilot · 2026-02-09T11:56:51Z

mobile_config/migrations/20260209112231_null_hash.sql

@@ -0,0 +1 @@
+UPDATE gateways SET hash = NULL;


The new 20260209112231_null_hash.sql sets gateways.hash to NULL for every row. Since Gateway::from_row and several queries (e.g. get_by_address, get_by_addresses, streams) still select hash directly and deserialize it into a non-optional String, this migration can cause runtime decode errors anywhere gateways are read before the tracker backfill runs (or for addresses not covered by the metadata stream). Consider backfilling hashes in the migration (or at least setting to empty string), or making reads tolerate NULL (e.g., COALESCE in all selects / Option<String> during migration) until the column is repopulated.

Copilot · 2026-02-09T11:56:51Z

mobile_config/src/gateway/db.rs

            created_at: row.try_get("created_at")?,
-            inserted_at: row.try_get("inserted_at")?,
            last_changed_at: row.try_get("last_changed_at")?,
            hash: row.try_get("hash")?,
-            antenna: to_u32(row.try_get("antenna")?),
-            elevation: to_u32(row.try_get("elevation")?),
-            azimuth: to_u32(row.try_get("azimuth")?),
-            location: to_u64(row.try_get("location")?),
            location_changed_at: row.try_get("location_changed_at")?,
-            location_asserts: to_u32(row.try_get("location_asserts")?),
-            owner: row.try_get("owner")?,
            owner_changed_at: row.try_get("owner_changed_at")?,


Gateway::from_row deserializes hash as a non-optional String (row.try_get("hash")?). Combined with the new migration that sets hash = NULL, any query that selects hash without COALESCE(hash, '') will fail to decode until backfilled. To make the migration safe, either change hash to Option<String> temporarily, or update all gateway SELECTs to COALESCE hash, or avoid setting existing rows to NULL in the first place.

Copilot · 2026-02-09T11:56:52Z

mobile_config/src/gateway/tracker.rs

+    let count: i64 = sqlx::query_scalar(
+        r#"
+        SELECT COUNT(*) FROM (
+            SELECT DISTINCT ON (address)
+                address,
+                gateway_type,
+                created_at,
+                last_changed_at,
+                COALESCE(hash, '') as hash,
+                antenna,
+                elevation,
+                azimuth,
+                location,
+                location_changed_at,
+                location_asserts,
+                owner,
+                owner_changed_at
+            FROM gateways
+            ORDER BY address, inserted_at DESC
+        ) AS distinct_gateways
+        WHERE hash = ''
+        "#,
+    )


backfill_hashes counts “null hash” gateways via COALESCE(hash, '') as hash ... WHERE hash = '', but the actual fetch/update path uses Gateway::get_by_addresses_with_null_hash which filters strictly on hash IS NULL. This can lead to count > 0 while total updated is 0 (e.g., when hash is an empty string rather than NULL) and makes the log message misleading. Align both sides to the same definition (prefer hash IS NULL everywhere, or handle both NULL and empty).

Copilot · 2026-02-09T11:56:52Z

mobile_config/src/gateway/db.rs

+                .unwrap_or([0u8; 4])
+                .as_ref(),
+        );
+        hasher.update(self.owner.as_deref().unwrap_or_default().as_bytes());


HashParams::compute_hash now includes owner (hasher.update(self.owner... )). In Gateway::new_if_changed, owner_changed intentionally ignores transitions where the new owner is None, but because the hash includes owner, a Some(x) -> None change will still change new_hash and create a new Gateway row while leaving owner_changed_at unchanged. This makes owner_changed_at inconsistent with the stored owner and can break consumers that rely on owner_changed_at for change tracking. Consider either excluding owner from the hash, or normalizing None to “keep last owner” for hash purposes, or updating the owner_changed/owner_changed_at logic to treat Some -> None as a change if you want owner removals to be tracked.

Suggested change

hasher.update(self.owner.as_deref().unwrap_or_default().as_bytes());

Transferring from Some(Owner) to None is not expected. Ignoring such "change" in mobile_hotspot_info table is intentional

Copilot · 2026-02-09T11:56:52Z

mobile_config/src/gateway/db.rs

+                owner_changed_at
+            FROM gateways
+            WHERE address = ANY($1)
+              AND hash IS NULL


Gateway::get_by_addresses_with_null_hash filters with AND hash IS NULL but also selects COALESCE(hash, '') as hash. If the intent is to backfill both NULL and empty-string hashes (as backfill_hashes’s COUNT query suggests), this method won’t return rows where hash = ''. Either broaden the predicate to include empty strings too, or adjust the upstream COUNT/logging to match hash IS NULL only.

Suggested change

AND hash IS NULL

AND (hash IS NULL OR hash = '')

kurotych added 15 commits February 5, 2026 14:31

Optimize MobileHotspotInfo::stream. Use gateway_type instead of devic…

49a73ba

…e_type

Delete unused update_bulk_location_changed_at

13c0bf2

Add gateway's HashParams

e13fc86

Add geters to Gateway struct

e5b8a6c

Rework pub async fn execute in tracker

68d1361

Remove unused compute_hash

408a9b0

Fix some todos

cda7ab7

Delete inserted_at as Gateway (struct) field

e3bec3a

Add HashParams::from_hotspot_info

1a0a20f

Implement new_if_changed func in Gateway

16c9471

Implement Gateway::from_mobile_hotspot_info

af654ff

Make new_if_changed based only mobile hotspot info

4b89c2d

Avoid useless cloning

a3e456d

About useless clooning (2)

a923050

Add migartion and backfill

9be74cb

kurotych requested a review from Copilot February 9, 2026 11:50

Copilot started reviewing on behalf of kurotych February 9, 2026 11:50 View session

Copilot AI reviewed Feb 9, 2026

View reviewed changes

kurotych added 2 commits February 9, 2026 14:04

Simplify query

d11a13c

Add tests

7cc8391

kurotych changed the title ~~[DRAFT] [mobile_config] track owner refactoring~~ [mobile_config] track owner refactoring Feb 9, 2026

kurotych requested review from bbalser, jeffgrunewald, macpie and michaeldjeffrey February 9, 2026 15:48

kurotych marked this pull request as ready for review February 9, 2026 15:49

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

[mobile_config] track owner refactoring #1131

[mobile_config] track owner refactoring #1131

Uh oh!

kurotych commented Feb 5, 2026 •

edited

Loading

Uh oh!

Copilot AI left a comment

Uh oh!

Copilot AI Feb 9, 2026

Uh oh!

Copilot AI Feb 9, 2026

Uh oh!

Copilot AI Feb 9, 2026

Uh oh!

Copilot AI Feb 9, 2026

Uh oh!

Copilot AI Feb 9, 2026

Uh oh!

kurotych Feb 9, 2026 •

edited

Loading

Uh oh!

Copilot AI Feb 9, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

[mobile_config] track owner refactoring #1131

Are you sure you want to change the base?

[mobile_config] track owner refactoring #1131

Uh oh!

Conversation

kurotych commented Feb 5, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Reviewed changes

Uh oh!

Copilot AI Feb 9, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Feb 9, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Feb 9, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Feb 9, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Feb 9, 2026

Choose a reason for hiding this comment

Uh oh!

kurotych Feb 9, 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!

Copilot AI Feb 9, 2026

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

kurotych commented Feb 5, 2026 •

edited

Loading

kurotych Feb 9, 2026 •

edited

Loading