galera.galera_wsrep_schema_detached fails: Server did not transition to READY state

galera.galera_wsrep_schema_detached fails with Server did not transition to READY state on the 10.11, 11.4, 11.8, 12.3 and main (13.2) branches.

Build Run Info

Failure text

Latest run on main (worker number, timestamps and paths differ between runs; the 12.3 run reads the same):

galera.galera_wsrep_schema_detached      w6 [ fail ]

        Test ended at 2026-10-16 17:34:55

CURRENT_TEST: galera.galera_wsrep_schema_detached

mysqltest: In included file "./include/galera_wait_ready.inc": 

included from /home/buildbot/aarch64-debian-12/build/mysql-test/suite/galera/include/start_mysqld.inc at line 16:

included from /home/buildbot/aarch64-debian-12/build/mysql-test/suite/galera/t/galera_wsrep_schema_detached.test at line 49:

At line 28: "Server did not transition to READY state"

The result from queries just before the failure was:

< snip >

2

INSERT INTO mysql.wsrep_allowlist (ip) VALUES (0);

SET GLOBAL wsrep_cluster_address = @wsrep_cluster_address_orig;

SELECT 1;

1

1

DELETE FROM mysql.wsrep_allowlist;

connection node_2;

Killing server ...

connection node_1;

SELECT COUNT(*) AS EXPECT_0 FROM mysql.wsrep_streaming_log;

EXPECT_0

0

SELECT COUNT(*) AS EXPECT_1 FROM mysql.wsrep_cluster;

EXPECT_1

1

SELECT COUNT(*) AS EXPECT_1 FROM mysql.wsrep_cluster_members;

EXPECT_1

1

connection node_2;

More results from queries before failure can be found in /home/buildbot/aarch64-debian-12/build/mysql-test/var/6/log/galera_wsrep_schema_detached.log

 - saving '/home/buildbot/aarch64-debian-12/build/mysql-test/var/6/log/galera.galera_wsrep_schema_detached/' to '/home/buildbot/aarch64-debian-12/build/mysql-test/var/log/galera.galera_wsrep_schema_detached/'

Retrying test galera.galera_wsrep_schema_detached, attempt(2/3)...

On 10.11, 11.4 and 11.8 the message reads At line 43: "Server did not transition to READY state in seconds" instead (all 8 release-branch rows there, 17 of the 23 rows overall). The value is empty in all of them: since MDEV-40281 the die in galera_wait_ready.inc prints $_galera_wait_ready_timeout, while the new parameter is called $galera_wait_ready_timeout, without the underscore.

Analysis

Two builders are affected: aarch64-debian-12 (10.11, 11.8, 12.3, main) and ppc64le-ubuntu-2204 (10.11, 11.4, 11.8). Between 2026-10-12 and 2026-10-18 the failure was recorded 23 times (12 on aarch64-debian-12, 11 on ppc64le-ubuntu-2204): 11 rows on release branches and 12 on PR and bb-/st- branches. All 23 are raised by galera_wait_ready.inc, called through start_mysqld.inc from galera_wsrep_schema_detached.test at line 49, when node 2 is restarted after kill_galera.inc and does not report wsrep_ready ON within the wait loop (600 polls, 0.1 s apart, so 60 s). MDEV-40281 added an optional $galera_wait_ready_timeout for restarts that wait for a full SST and gave galera.galera_wsrep_new_cluster 300 s; it is not in 12.3 or main yet, and this test does not set it. MDEV-38401 covers failures on aarch64-debian-12 that end with the same message; the comments there refer to SST timeouts at the donor under parallel MTR load and to MDEV-40727. The same message ends galera.galera_wan_restart_sst (MDEV-41387) and galera.galera_gra_log (MDEV-41487).

Created via the mtr-failure-create-bug skill.