Hi Heikki,
Heikki Tuuri wrote:
Don,
it is the assertion below which fails.
Do you use FOREIGN KEY .... ON UPDATE CASCADE?
Nope.
Do you have any idea which query causes the crash?
I didn't, but after Googling for similar problems, I found a thread where you had talked about a bug in a column prefix index. (here's the message: http://archives.neohapsis.com/archives/mysql/2003-q4/0697.html )
I was fairly sure I wasn't using any column prefix indexes, but sure enough, there was one. I removed the index, and MySQL has stopped crashing completely.
Now, it's entirely possible that whichever user was submitting some query just stopped at the same time I removed that index. But prior to my doing that, MySQL was crashing every few minutes. Probably a dozen times within an hour or two.
I'm keeping a close eye on it, and if it happens again, I'll turn the query log on to see if I can catch it.
Please run CHECK TABLE on suspicious tables.
Can you test on an x86 computer?
I would be happy to, but since I don't know which query is doing it, I'm not sure it'd be much use. I can't easily put an x86 box into production to get the same load on it.
Have you changed the default character set of the server in my.cnf?
Nope.
Can you please send me your my.cnf.
Here it is. I actually had the innodb buffer set to 4G before, and lowered it to see if we were still hitting some old 32bit limit or something. Of course, there was no change, the crashes still occurred.
FWIW, this on Red Hat Enterprise 3 for AMD64.
---
[mysqld] datadir = /xserve1/mysql port = 3306 socket = /tmp/mysql.sock pid-file = /xserve1/mysql/zeus.pid skip-external-locking set-variable = key_buffer=256M set-variable = max_allowed_packet=10M set-variable = table_cache=256 set-variable = sort_buffer=1M set-variable = read_buffer_size=1M set-variable = thread_cache=8 set-variable = thread_concurrency=4 set-variable = myisam_sort_buffer_size=512M set-variable = query_cache_size=512M set-variable = query_cache_type=1 set-variable = max_connections=500 set-variable = long_query_time=1
log-bin server-id=1
innodb_data_home_dir = /xserve1/mysql/ innodb_data_file_path = ibdata1:10M:autoextend innodb_log_group_home_dir = /xserve1/mysql/ innodb_log_arch_dir = /xserve1/mysql/ set-variable = innodb_buffer_pool_size=2G set-variable = innodb_additional_mem_pool_size=20M set-variable = innodb_log_file_size=512M set-variable = innodb_log_buffer_size=8M innodb_flush_log_at_trx_commit=1 set-variable = innodb_lock_wait_timeout=50
log-slow-queries log-error open-files-limit=8192 ft_min_word_len=3
max_connect_errors = 1000000000
---
Thanks for your reply! Please let me know if there's anything else I can do. I'm happy to help test and debug.
Don
Heikki Tuuri Innobase Oy http://www.innodb.com Foreign keys, transactions, and row level locking for MySQL InnoDB Hot Backup - hot backup tool for InnoDB which also backs up MyISAM tables
Order MySQL technical support from https://order.mysql.com/
.................. Builds an update vector from those fields which in a secondary index entry differ from a record that has the equal ordering fields. NOTE: we compare the fields as binary strings! */
upd_t* row_upd_build_sec_rec_difference_binary( /*====================================*/ /* out, own: update vector of differing fields */ dict_index_t* index, /* in: index */ dtuple_t* entry, /* in: entry to insert */ rec_t* rec, /* in: secondary index record */ mem_heap_t* heap) /* in: memory heap from which allocated */ { upd_field_t* upd_field; dfield_t* dfield; byte* data; ulint len; upd_t* update; ulint n_diff; ulint i;
/* This function is used only for a secondary index */ ut_ad(0 == (index->type & DICT_CLUSTERED));
update = upd_create(dtuple_get_n_fields(entry), heap);
n_diff = 0;
for (i = 0; i < dtuple_get_n_fields(entry); i++) {
data = rec_get_nth_field(rec, i, &len);
dfield = dtuple_get_nth_field(entry, i);
ut_a(len == dfield_get_len(dfield));
----- Original Message ----- From: "Don MacAskill" <[EMAIL PROTECTED]>
Newsgroups: mailing.database.myodbc
Sent: Wednesday, December 10, 2003 7:49 PM
Subject: MySQL 4.0.16 64bit crash report
I'm running 4.0.16 on a dual-Opteron using the 64bit mysql-max binary distribution (tar.gz, not rpm) from mysql.com. I have 8GB of RAM, and it was using about 4.5GB at the time of the crash.
031210 9:24:34 InnoDB: Assertion failure in thread 1171265904 in file row0upd.c line 713 InnoDB: Failing assertion: len == dfield_get_len(dfield) InnoDB: We intentionally generate a memory trap. InnoDB: Send a detailed bug report to [EMAIL PROTECTED] InnoDB: Thread 1183455600 stopped in file ha_innodb.cc line 396 InnoDB: Thread 1176770928 stopped in file ha_innodb.cc line 396 mysqld got signal 11; This could be because you hit a bug. It is also possible that this binary or one of the libraries it was linked against is corrupt, improperly
built,
or misconfigured. This error can also be caused by malfunctioning
hardware.
We will try our best to scrape up some info that will hopefully help diagnose the problem, but since we have already crashed, something is definitely wrong and this may fail.
key_buffer_size=268435456 read_buffer_size=1044480 max_used_connections=336 max_connections=500 threads_connected=143 It is possible that mysqld could use up to key_buffer_size + (read_buffer_size + sort_buffer_size)*max_connections = 1284140 K bytes of memory Hope that's ok; if not, decrease some variables in the equation.
InnoDB: Thread 1174018416 stopped in file ha_innodb.cc line 396 InnoDB: Thread 1214912880 stopped in file ha_innodb.cc line 396
Number of processes running now: 0
Any ideas?
Thanks,
Don
-- MySQL General Mailing List For list archives: http://lists.mysql.com/mysql To unsubscribe:
http://lists.mysql.com/[EMAIL PROTECTED]
-- MySQL General Mailing List For list archives: http://lists.mysql.com/mysql To unsubscribe: http://lists.mysql.com/[EMAIL PROTECTED]
