[cgrs5722]

Description

GatewayService.assetIdGatewayIdMap is a plain HashMap that is written under synchronized in processGatewayChildAssetChange, but read without any synchronization from:

• onAttributeEventIntercepted (get, attribute event interceptor thread)
• deleteGateway (entrySet().stream() while writes can happen concurrently)
• getLocallyRegisteredGatewayId (get, called from the Camel route predicate)

This is a data race: unsynchronized reads against synchronized writes on a HashMap can throw ConcurrentModificationException while streaming the entry set, or observe inconsistent state.

The fix switches the map to ConcurrentHashMap, consistent with the adjacent gatewayConnectorMap and tunnelInfos fields.

Fixes #3449.

Verification

• ./gradlew :manager:compileJava passes (JDK 21).
• A standalone probe replicating the exact access pattern (one thread doing synchronized put/remove as in processGatewayChildAssetChange, another thread doing get + entrySet().stream().filter(...) as in deleteGateway) fails with ConcurrentModificationException on HashMap within ~1000 reads, while ConcurrentHashMap completes hundreds of thousands of operations without error.
• The integration tests in the test project (e.g. GatewayTest) require a running PostgreSQL + Keycloak stack via docker compose, which I don't have available locally.

🤖 Generated with Claude Code

[CLAassistant]

All committers have signed the CLA.

[richturner]

Hi @cgrs5722 thanks for picking this up, by switching to ConcurrentHashMap any existing uses of synchronized on the map can also be removed as described in the linked issue.

[richturner]

@cgrs5722 would be great to get this finished off, otherwise we'll have to close this off.

[sonarqubecloud]

Quality Gate passed

Issues
0 New issues
0 Accepted issues

Measures
0 Security Hotspots
0.0% Coverage on New Code
0.0% Duplication on New Code

See analysis details on SonarQube Cloud