Description:
Start mysqld (call it "mysqld 1") on a cluster data node.
mysqld 1 is not actually connected to the cluster.
Create a logfile group, tablespace, and disk data table on the cluster (using a "mysqld 2" that *is* connected to the cluster). All works as expected, selecting from I_S.FILES on mysqld 2 shows that all expected files were created.
On mysqld 1, do a SELECT * FROM INFORMATION_SCHEMA.FILES;
mysqld 1 crashes. mysqld2 and the cluster continue to run without issue.
Backtrace (mysqld 1):
0x81f5b12
0xffffe420
0x86b4e5c
0x8549aee
0x835288d
0x8358de7
0x8358eb2
0x83644c6
0x82f39e7
0x833fdc5
0x82f3a20
0x82f3966
0x826802b
0x82694cf
0x826970f
0x820ede3
0x821641e
0x8216c88
0x8217db4
0x8218173
0x40055297
0x4019037e
Resolved trace (mysqld 1):
0x81f5b12 handle_segfault + 438
0xffffe420 _end + -142793464
0x86b4e5c neg_map + 124
0x8549aee _ZN3NdbC1EP22Ndb_cluster_connectionPKcS3_ + 66
0x835288d _ZN7Thd_ndbC1Ev + 65
0x8358de7 _ZN13ha_ndbcluster13seize_thd_ndbEv + 31
0x8358eb2 _Z16check_ndb_in_thdP3THD + 34
0x83644c6 _Z27ndbcluster_fill_files_tableP3THDP13st_table_listP4Item + 32
0x82f39e7 _Z26run_hton_fill_schema_filesP3THDP13st_plugin_intPv + 61
0x833fdc5 _Z14plugin_foreachP3THDPFcS0_P13st_plugin_intPvEiS3_ + 235
0x82f3a20 _Z17fill_schema_filesP3THDP13st_table_listP4Item + 46
0x82f3966 _Z24get_schema_tables_resultP4JOIN + 420
0x826802b _ZN4JOIN4execEv + 1809
0x82694cf _Z12mysql_selectP3THDPPP4ItemP13st_table_listjR4ListIS1_ES2_jP8st_orderSB_S2_SB_mP13select_resultP18st_select_lex_unitP13st_sel + 563
0x826970f _Z13handle_selectP3THDP6st_lexP13select_resultm + 309
0x820ede3 _Z21mysql_execute_commandP3THD + 1841
0x821641e _Z11mysql_parseP3THDPcj + 606
0x8216c88 _Z16dispatch_command19enum_server_commandP3THDPcj + 1818
0x8217db4 _Z10do_commandP3THD + 262
0x8218173 handle_one_connection + 729
0x40055297 _end + 931304319
0x4019037e _end + 932594790
No errors logged by mysqld 2.
How to repeat:
See description.
Suggested fix:
Unsure, but "mysqld 1" probably should not crash.
