|
On a server that has a quite old datadir (probably dates back at least to 5.5), upgrading from 10.3.38 to 10.4.28 resulted in mysqld doing a segfault / crash with signal 11 with this backtrace on the error log :
Thread pointer: 0x562278421b08
|
Attempting backtrace. You can use the following information to find out
|
where mysqld died. If you see no messages after this, something went
|
terribly wrong...
|
stack_bottom = 0x7ffe5c9c1248 thread_stack 0x49000
|
/usr/sbin/mysqld(my_print_stacktrace+0x2e)[0x5622692faf8e]
|
/usr/sbin/mysqld(handle_fatal_signal+0x54d)[0x562268dca99d]
|
/lib/x86_64-linux-gnu/libpthread.so.0(+0x12730)[0x7f0c70c52730]
|
/usr/sbin/mysqld(+0x63307a)[0x562268b5207a]
|
/usr/sbin/mysqld(_Z10acl_reloadP3THD+0x678)[0x562268b5cba8]
|
/usr/sbin/mysqld(_Z8acl_initb+0x133)[0x562268b5df13]
|
/usr/sbin/mysqld(_Z11mysqld_mainiPPc+0x8f0)[0x562268b20e80]
|
csu/libc-start.c:342(__libc_start_main)[0x7f0c7079609b]
|
/usr/sbin/mysqld(_start+0x2a)[0x562268b14a6a]
|
After running the process on gdb (i haven't been able to produce a core file, not sure why), the backtrace revealed that it was crashing on the function get_magic_sort being called from acl_load which was reading the mysql.host table (i removed the optimized out variables) :
Thread 1 "mysqld" received signal SIGSEGV, Segmentation fault.
|
0x0000555555b9222a in get_magic_sort (templ=0x555556410e5c "d", templ@entry=0x555556410e5b "hd") at ./sql/sql_acl_getsort.ic:160
|
160 ./sql/sql_acl_getsort.ic: No such file or directory.
|
(gdb) bt full
|
#0 0x0000555555b9222a in get_magic_sort (templ=0x555556410e5c "d", templ@entry=0x555556410e5b "hd") at ./sql/sql_acl_getsort.ic:160
|
pat = 0x0
|
sort = 0
|
args = {{gp_offset = 24, fp_offset = 0, overflow_arg_area = 0x7fffffffdd40, reg_save_area = 0x7fffffffdcd8}}
|
#1 0x0000555555b9eae0 in acl_load (tables=..., thd=0x555564623758) at ./sql/sql_acl.cc:2442
|
host = {<ACL_ACCESS> = {sort = 0, access = 31931455}, host = {hostname = 0x555564665760 "10.1.0.0/255.255.0.0", ip = 167837696, ip_mask = 4294901760}, db = 0x0}
|
read_record_info = {table = 0x555564637bc8, unlock_row = 0x555555c44f60 <rr_unlock_row(st_join_table*)>, read_record_func = 0x555555f3d850 <rr_sequential(READ_RECORD*)>, thd = 0x555564623758, select = 0x0, ref_length = 7, reclength = 0, rec_cache_size = 0, error_offset = 0, ref_pos = 0x0, rec_buf = 0x0, cache = 0x0, cache_pos = 0x0, cache_end = 0x0, read_positions = 0x0, addon_field = 0x0, io_cache = 0x0, print_error = true, unpack = 0x0, copy_field = 0x0,
|
copy_field_end = 0x0}
|
tmp_name = "`ZbdUU\000\000\000\207\033\221 AF\373\000\004\000\000\000\000\000\000X\nZdUU\000\000\020\000\001\000\000\000\000\000\b\004\000\000\000\000\000\000 \342\377\377\377\177\000\000d\334\063VUU\000\000X7bdUU\000\000\030\352\266VUU\000\000h\225bdUU\000\000H\224bdUU\000\000\320\342\377\377\377\177\000\000\327\212\275UUU\000\000\230\223bdUU\000\000 sbdUU\000\000\370pbdUU\000\000\330RbdUU\000\000\240\070bdUU\000\000\b\232bdUU\000\000(\222bdUU\000\000\000\000\000\000\000\000\000\000\370<bdUU\000\000\320<bdUU\000\000\350SbdUU\000\000("...
|
host_table = @0x7fffffffe160: {<Grant_table_base> = {min_columns = 8, start_priv_columns = 2, end_priv_columns = 20, m_table = 0x555564637bc8}, <No data fields>}
|
So i did copy the myisam files for the table mysql.host to another server and the content did look like this :
*************************** 1. row ***************************
|
Host: 10.1.0.0/255.255.0.0
|
Db:
|
Select_priv: Y
|
Insert_priv: Y
|
Update_priv: Y
|
Delete_priv: Y
|
Create_priv: Y
|
Drop_priv: Y
|
Grant_priv: Y
|
References_priv: Y
|
Index_priv: Y
|
Alter_priv: Y
|
Create_tmp_table_priv: Y
|
Lock_tables_priv: Y
|
Create_view_priv: Y
|
Show_view_priv: Y
|
Create_routine_priv: Y
|
Alter_routine_priv: Y
|
Execute_priv: Y
|
Trigger_priv: N
|
*************************** 2. row ***************************
|
Host: 10.5.0.0/255.255.0.0
|
Db:
|
Select_priv: Y
|
Insert_priv: Y
|
Update_priv: Y
|
Delete_priv: Y
|
Create_priv: Y
|
Drop_priv: Y
|
Grant_priv: Y
|
References_priv: Y
|
Index_priv: Y
|
Alter_priv: Y
|
Create_tmp_table_priv: Y
|
Lock_tables_priv: Y
|
Create_view_priv: Y
|
Show_view_priv: Y
|
Create_routine_priv: Y
|
Alter_routine_priv: Y
|
Execute_priv: Y
|
Trigger_priv: N
|
*************************** 3. row ***************************
|
Host: 10.8.1.0/255.255.255.0
|
Db:
|
Select_priv: Y
|
Insert_priv: Y
|
Update_priv: Y
|
Delete_priv: Y
|
Create_priv: Y
|
Drop_priv: Y
|
Grant_priv: Y
|
References_priv: Y
|
Index_priv: Y
|
Alter_priv: Y
|
Create_tmp_table_priv: Y
|
Lock_tables_priv: Y
|
Create_view_priv: Y
|
Show_view_priv: Y
|
Create_routine_priv: Y
|
Alter_routine_priv: Y
|
Execute_priv: Y
|
Trigger_priv: N
|
Those entries are at least 4 or 5 years old (probably more) as those IP ranges are not in use anymore and have survived several MariaDB major version upgrade.
I assume it's the empty "Db" column that is not valid so i did copy the empty host table from the server i was on to the crashing one and it did fix it.
|