Do not wait for ?rlog_sync table

[ieQu1]

Speed up cluster restart by not waiting for the master of ?rlog_sync table. It's a null_copies table used solely for the events, its contents are irrelevant.

[ieQu1]

Note: we no longer use gen_rpc as a transport in production.

[Copilot]

Pull request overview

This PR adjusts Mria startup/schema bootstrap behavior to avoid getting stuck waiting on the ?rlog_sync table, and adds/updates CT coverage to exercise full cluster restart scenarios.

Changes:

• Remove blocking waits for ?rlog_sync during schema bootstrap and introduce a local force-load step instead.
• Reduce mria_mnesia:wait_for_tables/1 per-iteration wait timeout (30s → 5s) before retry/diagnostics.
• Update/add CT suites to better simulate realistic cluster conditions and validate full cluster restart behavior.

Reviewed changes

Copilot reviewed 5 out of 5 changed files in this pull request and generated 4 comments.

Show a summary per file

File	Description
test/mria_mnesia_test_util.erl	Removes an unused/ineffective env setting from common test env.
test/mria_autoclean_SUITE.erl	Makes autoclean test run with a replicant and adds more “realistic” setup steps.
test/mria_SUITE.erl	Adds a new full-cluster-restart testcase and updates copyright year.
src/mria_schema.erl	Stops waiting on ?rlog_sync and adds a force_load/1 helper during bootstrap.
src/mria_mnesia.erl	Lowers the wait timeout used by wait_for_tables/1 retry loop.

Comments suppressed due to low confidence (1)

src/mria_mnesia.erl:280

• Reducing the mnesia:wait_for_tables/2 timeout from 30s to 5s increases the frequency of the timeout path, which logs a warning, runs diagnostics, and calls mnesia_controller:connect_nodes/1 before retrying. This can significantly increase log volume and connection churn in slow/large clusters. Consider keeping a larger default, adding a backoff, or making the timeout configurable.

    case mnesia:wait_for_tables(Tables, 5_000) of
        ok ->
            ?tp(mria_wait_for_tables_done, #{result => ok}),
            ok;
        {error, Reason} ->
            ?tp(mria_wait_for_tables_done, #{result => {error, Reason}}),
            {error, Reason};
        {timeout, BadTables} ->
            logger:warning("~p: still waiting for table(s): ~p", [?MODULE, BadTables]),
            catch diagnosis(BadTables),
            %% lets try to force reconnect all the db_nodes to get schema merged,
            %% mnesia_controller is smart enough to not force reconnect the node that is already connected.
            mnesia_controller:connect_nodes(mnesia:system_info(db_nodes)),
            wait_for_tables(BadTables)

---

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