database/mariadb
1
0
Fork 0
mirror of https://github.com/MariaDB/server.git synced 2025-08-08 11:22:35 +03:00
Code Activity

MDEV-21910 Deadlock between BF abort and manual KILL command

Browse Source 
When high priority replication slave applier encounters lock conflict in innodb,
it will force the conflicting lock holder transaction (victim) to rollback.
This is a must in multi-master sychronous replication model to avoid cluster lock-up.
This high priority victim abort (aka "brute force" (BF) abort), is started
from innodb lock manager while holding the victim's transaction's (trx) mutex.
Depending on the execution state of the victim transaction, it may happen that the
BF abort will call for THD::awake() to wake up the victim transaction for the rollback.
Now, if BF abort requires THD::awake() to be called, then the applier thread executed
locking protocol of: victim trx mutex -> victim THD::LOCK_thd_data

If, at the same time another DBMS super user issues KILL command to abort the same victim,
it will execute locking protocol of: victim THD::LOCK_thd_data  -> victim trx mutex.
These two locking protocol acquire mutexes in opposite order, hence unresolvable mutex locking
deadlock may occur.

The fix in this commit adds THD::wsrep_aborter flag to synchronize who can kill the victim
This flag is set both when BF is called for from innodb and by KILL command.
Either path of victim killing will bail out if victim's wsrep_killed is already
set to avoid mutex conflicts with the other aborter execution. THD::wsrep_aborter
records the aborter THD's ID. This is needed to preserve the right to kill
the victim from different locations for the same aborter thread.
It is also good error logging, to see who is reponsible for the abort.

A new test case was added in galera.galera_bf_kill_debug.test for scenario where
wsrep applier thread and manual KILL command try to kill same idle victim
This commit is contained in:
sjaakola
2020-05-19 11:12:26 +03:00
committed by Jan Lindström
parent 141b390d82
commit 5a7794d3a8
15 changed files with 290 additions and 73 deletions
Download Patch File Download Diff File Expand all files Collapse all files

5
include/mysql/service_wsrep.h
View File

@@ -84,6 +84,7 @@ extern struct wsrep_service_st {
my_bool (*wsrep_get_debug_func)();
void (*wsrep_commit_ordered_func)(MYSQL_THD thd);
my_bool (*wsrep_thd_is_applying_func)(const MYSQL_THD thd);
bool (*wsrep_thd_set_wsrep_aborter_func)(MYSQL_THD bf_thd, MYSQL_THD thd);
} *wsrep_service;
#define MYSQL_SERVICE_WSREP_INCLUDED
@@ -124,6 +125,7 @@ extern struct wsrep_service_st {
#define wsrep_get_debug() wsrep_service->wsrep_get_debug_func()
#define wsrep_commit_ordered(T) wsrep_service->wsrep_commit_ordered_func(T)
#define wsrep_thd_is_applying(T) wsrep_service->wsrep_thd_is_applying_func(T)
#define wsrep_thd_set_wsrep_aborter(T) wsrep_service->wsrep_thd_set_wsrep_aborter_func(T1, T2)
#else
@@ -176,6 +178,8 @@ extern "C" my_bool wsrep_thd_is_local(const MYSQL_THD thd);
/* Return true if thd is in high priority mode */
/* todo: rename to is_high_priority() */
extern "C" my_bool wsrep_thd_is_applying(const MYSQL_THD thd);
/* set wsrep_aborter for the target THD */
extern "C" bool wsrep_thd_set_wsrep_aborter(MYSQL_THD bf_thd, MYSQL_THD victim_thd);
/* Return true if thd is in TOI mode */
extern "C" my_bool wsrep_thd_is_toi(const MYSQL_THD thd);
/* Return true if thd is in replicating TOI mode */
@@ -216,6 +220,7 @@ extern "C" my_bool wsrep_get_debug();
extern "C" void wsrep_commit_ordered(MYSQL_THD thd);
extern "C" my_bool wsrep_thd_is_applying(const MYSQL_THD thd);
extern "C" bool wsrep_thd_set_wsrep_aborter(MYSQL_THD bf_thd, MYSQL_THD victim_thd);
#endif
#endif /* MYSQL_SERVICE_WSREP_INCLUDED */