Details
-
Bug
-
Status: Closed (View Workflow)
-
Blocker
-
Resolution: Fixed
-
10.5, 10.6
Description
The failed one in http://buildbot.askmonty.org/buildbot/builders/winx64-debug/builds/22164
commit 657fcdf430f39a3103dff51a6a2b2bd3 with the following stack trace
server!inline_mysql_cond_wait(struct st_mysql_cond * that = 0x00007ffb`ec653b68, struct st_mysql_mutex * mutex = 0x00007ffb`ec653b38, char * src_file = 0x00007ffb`ea98b140 "D:\winx64-debug\build\src\sql\log.cc", unsigned int src_line = 0x10f8) [D:\winx64-debug\build\src\include\mysql\psi\mysql_thread.h @ 1222]
|
server!MYSQL_BIN_LOG::reset_logs(class THD * thd = 0x00000165`d43d15c8, bool create_new_log = true, struct rpl_gtid * init_state = 0x00000000`00000000, unsigned int init_state_len = 0, unsigned long next_log_number = 0) [D:\winx64-debug\build\src\sql\log.cc @ 4345]
|
server!reset_master(class THD * thd = 0x00000165`d43d15c8, struct rpl_gtid * init_state = 0x00000000`00000000, unsigned int init_state_len = 0, unsigned long next_log_number = 0) [D:\winx64-debug\build\src\sql\sql_repl.cc @ 3966]
|
server!reload_acl_and_cache(class THD * thd = 0x00000165`d43d15c8, unsigned int64 options = 0x80, struct TABLE_LIST * tables = 0x00000000`00000000, int * write_to_binlog = 0x0000002a`c42fdc44) [D:\winx64-debug\build\src\sql\sql_reload.cc @ 362]
|
server!mysql_execute_command(class THD * thd = 0x00000165`d43d15c8) [D:\winx64-debug\build\src\sql\sql_parse.cc @ 5480]
|
server!mysql_parse(class THD * thd = 0x00000165`d43d15c8, char * rawbuf = 0x00000165`d44da280 "--- memory read error at address 0x00000165`d44da280 ---", unsigned int length = 0xc, class Parser_state * parser_state =
|
As it was never observed before the extra to 10.5 commits
need examination to clear out
possibility of missed out commit_checkpoint_notify_ha() invocation from innobase_mysql_log_notify() which is one of possibilities.
657fcdf430f and 7b1252c03d7 are rated as potentially relevant to the failure showdown.
A similar failure
main.mysqldump-max 'innodb' w2 [ fail ] timeout after 900 seconds
|
exists in http://buildbot.askmonty.org/buildbot/builders/win32-debug/builds/18570/steps/test/logs/stdio
Upon more analysis the most probable suspect
--- a/storage/innobase/handler/ha_innodb.cc
|
+++ b/storage/innobase/handler/ha_innodb.cc
|
@@ -4444,12 +4444,6 @@ innobase_mysql_log_notify( |
struct pending_checkpoint * entry;
|
struct pending_checkpoint * last_ready;
|
|
- /* It is safe to do a quick check for NULL first without lock. |
- Even if we should race, we will at most skip one checkpoint and
|
- take the next one, which is harmless. */
|
- if (!pending_checkpoint_list) |
- return; |
-
|
is identified. , to eliminate (the pasted block is from being tested fixes). Removal won't hurt performance in normal cases when binlog rotation is not frequent event (say not few in a second).
Attachments
Issue Links
- blocks
-
MDEV-25611 RESET MASTER still causes the server to hang
- Closed
- causes
-
MDEV-25313 Assertion `pending == log_requests.start.load(std::memory_order_relaxed)' failed in log_flush_notify_and_unlock or crash in MYSQL_BIN_LOG::mark_xid_done
- Closed
- is caused by
-
MDEV-232 Remove one fsync() inside engine's commit() method
- Closed
-
MDEV-532 Implement async commit checkpoint in InnoDB and XtraDB
- Closed
- relates to
-
MDEV-24526 binlog rotate via FLUSH LOGS may obsolate binlog file for recovery too early
- Closed
-
MDEV-24886 main.sp_trans_log fails in buildbot with timeout
- Open