fix: prevent monitor crash when CLUSTER SLOTS times out

[hjianbo]

Background

This PR addresses a Redis Cluster refresh-timeout crash in eredis_cluster monitor process on the branch-0.7.x line.

Problems Fixed

1. Monitor crash during slot refresh timeout

   • eredis_cluster_monitor:get_cluster_slots/3 called eredis:q(Connection, ["CLUSTER", "SLOTS"]) directly in a gen_server call path.
   • When eredis:q/2 exits on timeout, the exception could bubble up and terminate the monitor process.

2. Potential unsafe connection stop in fallback branches

   • In some branches (unknown command / cluster disabled), stop used plain eredis:stop/1.
   • This PR unifies those branches to safe_eredis_stop/1.

3. Regression test cleanup instability

   • The new timeout regression test unloaded meck before monitor shutdown, producing noisy test-time noproc termination reports.
   • This PR stops monitor explicitly before unload.

Typical Error Logs Covered

These are the key log patterns related to this issue:

Failed to get cluster slots from redis node ...: {error,{cluster_slots_exit,timeout}}
Failed to get cluster slots: {<<"ERR all nodes are down">>, ...}

Before this fix, timeout could additionally lead to monitor termination logs (for example Generic server monitor_* terminating due to refresh path exceptions).

Changes

• src/eredis_cluster_monitor.erl
  • Add safe_get_cluster_slots/1 to catch exit from eredis:q/2 and convert it into normal {error, ...} flow.
  • Use safe_eredis_stop/1 consistently in all CLUSTER SLOTS branches.
• test/eredis_cluster_tests.erl
  • Add/keep timeout regression case and stop monitor cleanly in test teardown path.
• src/eredis_cluster.appup.src
  • Bump appup target to 0.7.9.
  • Add upgrade/downgrade instructions for 0.7.8.

Verification

• /home/moo/Worksapce/emqx-enterprise/rebar3 as test eunit -g eredis_cluster_tests:monitor_reload_slots_timeout_test_ -v

[Copilot]

Pull request overview

This PR hardens eredis_cluster_monitor against eredis:q(Connection, ["CLUSTER","SLOTS"]) exiting (e.g., timeout) during slot refresh, preventing the monitor gen_server from crashing and adding a regression test for the behavior.

Changes:

• Wrap CLUSTER SLOTS query in safe_get_cluster_slots/1 to convert exit into an error tuple.
• Introduce safe_eredis_stop/1 and use it on some slot-fetch paths to reduce crash risk during cleanup.
• Add an EUnit regression test covering the timeout/exit case during refresh_mapping/2.

Reviewed changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated 2 comments.

File	Description
src/eredis_cluster_monitor.erl	Adds safe wrapper for CLUSTER SLOTS and safe stop helper; updates slot refresh path to use them.
test/eredis_cluster_tests.erl	Adds regression test to ensure monitor survives CLUSTER SLOTS timeout/exit during refresh.

---

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

[Copilot]

Pull request overview

Copilot reviewed 3 out of 3 changed files in this pull request and generated no new comments.

---

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