1berry/MariaDB-server
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

BUG#22530768 Innodb freeze running REPLACE statements

Browse Source 
we can see from the hang stacktrace, srv_monitor_thread is blocked
when getting log_sys::mutex, so that sync_arr_wake_threads_if_sema_free
cannot get a change to break the mutex deadlock.

The fix is simply removing any mutex wait in srv_monitor_thread.

Patch is reviewed by Sunny over IM.
This commit is contained in:
Shaohua Wang 2016-01-12 15:08:09 +08:00
parent c06138e46d
commit 79032a7ae1
2 changed files with 21 additions and 17 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

7
storage/innobase/buf/buf0flu.c
View File

@ -1,6 +1,6 @@
/*****************************************************************************
Copyright (c) 1995, 2011, Oracle and/or its affiliates. All Rights Reserved.
Copyright (c) 1995, 2016, Oracle and/or its affiliates. All Rights Reserved.
This program is free software; you can redistribute it and/or modify it under
the terms of the GNU General Public License as published by the Free Software
@ -2103,7 +2103,10 @@ buf_flush_stat_update(void)
ib_uint64_t lsn;
ulint n_flushed;
lsn = log_get_lsn();
if (!log_peek_lsn(&lsn)) {
return;
}
if (buf_flush_stat_cur.redo == 0) {
/* First time around. Just update the current LSN
and return. */

31
storage/innobase/srv/srv0srv.c
View File

@ -1,6 +1,6 @@
/*****************************************************************************
Copyright (c) 1995, 2013, Oracle and/or its affiliates. All Rights Reserved.
Copyright (c) 1995, 2016, Oracle and/or its affiliates. All Rights Reserved.
Copyright (c) 2008, 2009 Google Inc.
Copyright (c) 2009, Percona Inc.
@ -2413,6 +2413,8 @@ exit_func:
/*********************************************************************//**
A thread which prints warnings about semaphore waits which have lasted
too long. These can be used to track bugs which cause hangs.
Note: In order to make sync_arr_wake_threads_if_sema_free work as expected,
we should avoid waiting any mutexes in this function!
@return a dummy parameter */
UNIV_INTERN
os_thread_ret_t
@ -2450,23 +2452,22 @@ loop:
/* Try to track a strange bug reported by Harald Fuchs and others,
where the lsn seems to decrease at times */
if (log_peek_lsn(&new_lsn)) {
if (new_lsn < old_lsn) {
ut_print_timestamp(stderr);
fprintf(stderr,
" InnoDB: Error: old log sequence number %llu"
" was greater\n"
"InnoDB: than the new log sequence number %llu!\n"
"InnoDB: Please submit a bug report"
" to http://bugs.mysql.com\n",
old_lsn, new_lsn);
ut_ad(0);
}
new_lsn = log_get_lsn();
if (new_lsn < old_lsn) {
ut_print_timestamp(stderr);
fprintf(stderr,
" InnoDB: Error: old log sequence number %llu"
" was greater\n"
"InnoDB: than the new log sequence number %llu!\n"
"InnoDB: Please submit a bug report"
" to http://bugs.mysql.com\n",
old_lsn, new_lsn);
ut_ad(0);
old_lsn = new_lsn;
}
old_lsn = new_lsn;
if (difftime(time(NULL), srv_last_monitor_time) > 60) {
/* We referesh InnoDB Monitor values so that averages are
printed from at most 60 last seconds */