Bug #87299 mysql crash sometimes
Submitted: 3 Aug 2017 9:21 Modified: 26 Nov 2017 14:51
Reporter: lou shuai (OCA) Email Updates:
Status: No Feedback Impact on me:
None 
Category:MySQL Server: InnoDB storage engine Severity:S1 (Critical)
Version:5.7.18 OS:Any
Assigned to: CPU Architecture:Any

[3 Aug 2017 9:21] lou shuai
Description:
MySQL CRASH when creating tables with the following error log

Thread pointer: 0x7f30d9c16000
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 = 7f30da7bdde8 thread_stack 0x40000
/app/dbcluster/mydb/mysql/bin/mysqld(my_print_stacktrace+0x35)[0xf37585]
/app/dbcluster/mydb/mysql/bin/mysqld(handle_fatal_signal+0x4b4)[0x7c1924]
/lib64/libpthread.so.0[0x3e74a0f7e0]
/app/dbcluster/mydb/mysql/bin/mysqld(_Z32innobase_parse_hint_from_commentP3THDP12dict_table_tPK11TABLE_SHARE+0x118)[0xfd4c08]
/app/dbcluster/mydb/mysql/bin/mysqld(_ZN19create_table_info_t24create_table_update_dictEv+0x23f)[0xfda98f]
/app/dbcluster/mydb/mysql/bin/mysqld(_ZN11ha_innopart6createEPKcP5TABLEP24st_ha_create_information+0x727)[0xff5f17]
/app/dbcluster/mydb/mysql/bin/mysqld(_Z15ha_create_tableP3THDPKcS2_S2_P24st_ha_create_informationbb+0x2a3)[0x81b5e3]
/app/dbcluster/mydb/mysql/bin/mysqld(_Z16rea_create_tableP3THDPKcS2_S2_P24st_ha_create_informationR4ListI12Create_fieldEjP6st_keyP7handlerb+0x11a)[0xdd819a]
/app/dbcluster/mydb/mysql/bin/mysqld[0xd76d37]
/app/dbcluster/mydb/mysql/bin/mysqld(_Z26mysql_create_table_no_lockP3THDPKcS2_P24st_ha_create_informationP10Alter_infojPb+0xe7)[0xd77607]
/app/dbcluster/mydb/mysql/bin/mysqld(_Z18mysql_create_tableP3THDP10TABLE_LISTP24st_ha_create_informationP10Alter_info+0xcb)[0xd77e0b]
/app/dbcluster/mydb/mysql/bin/mysqld(_Z21mysql_execute_commandP3THDb+0x4a0f)[0xd1177f]
/app/dbcluster/mydb/mysql/bin/mysqld(_Z11mysql_parseP3THDP12Parser_state+0x635)[0xd127b5]
/app/dbcluster/mydb/mysql/bin/mysqld(_Z16dispatch_commandP3THDPK8COM_DATA19enum_server_command+0x11a5)[0xd139c5]
/app/dbcluster/mydb/mysql/bin/mysqld(_Z10do_commandP3THD+0x1d4)[0xd144f4]
/app/dbcluster/mydb/mysql/bin/mysqld(handle_connection+0x2cc)[0xdeba2c]
/app/dbcluster/mydb/mysql/bin/mysqld(pfs_spawn_thread+0x171)[0xf50ca1]
/lib64/libpthread.so.0[0x3e74a07aa1]
/lib64/libc.so.6(clone+0x6d)[0x3e746e8aad]

How to repeat:
Create many tables in a script but hard to repeat cause the crash is not always crash at specific SQL.
[3 Aug 2017 9:27] lou shuai
Plz check the code according to the stack, cause hard to repeat!
[26 Oct 2017 14:51] MySQL Verification Team
Hi!

Thank you for your bug report. Unfortunately, we can not deduce the problem from the stack trace, as it is completely new to us.

What we need is a complete, fully repeatable test case. That means a CREATE TABLE statement that crashed. That would be rather easy as you do not contents of the table, since the crash occurs on the CREATE TABLE. We do not have pending bugs with similar stack traces.

If you do not know what CREATE TABLE is that produces the crash, then there is another way.

If you are on Unix, you should get a core file that is produced in the crash and send it to us, together with the binary that produced the crash.

However .....

You indicate random crashes. Random crashes happen mostly when hardware is not of the high standard. For example, you can get crashes due to memory glitches if you are not using ECC RAM modules.

It is also possible that you have faulty hardware, like disk, controller, caches on both, motherboard, CPU L1 or L2 caches etc .... On Unix, there are sometimes reports on these problems in the administrative logs.
[26 Oct 2017 15:37] MySQL Verification Team
Hi!

One additional comment. A bug like in this happens with Group Replication.

If you do not use GR Plugin, then just ignore this comment.
[27 Nov 2017 1:00] Bugs System
No feedback was provided for this bug for over a month, so it is
being suspended automatically. If you are able to provide the
information that was originally requested, please do so and change
the status of the bug back to "Open".