Here is what I am seeing
I see three queries, all almost identical.
UPDATE people SET company_id = 1610, name = '<name>', password = '<hash>',
temp_password = NULL, reset_password_hash = NULL, email = '<redacted>@yahoo.com',
phone = NULL, mobile = '<phone>', iphone_device_id = 'android:<id>-<id>',
iphone_device_time = '2011-06-06 05:35:09', last_checkin = '2011-06-06 05:24:42',
location_lat = <lat>, location_long = -<lng>, gps_strength = 3296,
picture_blob_id = 1190,authority = 1, active = 1,
date_created = '2011-04-13 20:21:20',
last_login = '2011-06-06 05:35:09', panic_mode = 0, battery_level = NULL,
battery_state = NULL WHERE people_id = 3125;
The differences
TRANSACTION 1
iphone_device_time = '2011-06-06 05:24:42', last_checkin = '2011-06-06 05:35:07'
TRANSACTION 2
iphone_device_time = '2011-06-06 05:35:09', last_checkin = '2011-06-06 05:24:42'
Please notice that the column values are flipped. Normally, a deadlock occurs when two different transactions are accessing two locks from two tables with TX1 (Transaction 1) getting row A and then row B while TX2 is getting row B and then row A. In this case, it is TX1 and TX2 are accessing the same row but changing two different columns (iphone_device_time,last_checkin).
The values do not make any sense. At 5:24:42, your last checkin was 5:35:07. Ten minutes and 27 seconds later (5:35:07 - 05:24:42), the column values are reversed.
The big question is: Why is TX1 held up for almost 11 min ???
This is not really an answer. This is just bandwidth and throughout from me. I hope these observations help.
UPDATE 2011-06-06 09:57
Please check out this link concerning innodb_locks_unsafe_for_binlog : The reason I suggest reading this is something else I saw in your INNODB STATUS display. The phrase lock_mode X (exclusive lock) and lock_mode S (shared lock) indicates both locks being imposed (or attempting to impose) on the same row . There may be some internal serialization going on doing next row locking. The default is OFF. After reading this, you may need to consider enabling it.
UPDATE 2011-06-06 10:03
Another reason to examine this line of thought is the fact that all the transactions are traversing the PRIMARY key. Since the PRIMARY is a clustered index in InnoDB, the PRIMARY key and the row itself are together. Thus, traversing a row and and the PRIMARY KEY are one and the same. Therefore, any index lock on the PRIMARY KEY is a row level lock as well.
UPDATE 2011-06-06 19:21
Check what auocommit value you have. If autocommit is off, I can see two(2) possible problems
- updating the same row twice in the same transaction
- updating the same row in two different transactions
In fact, the SHOW ENGINE INNODB STATUS you show in the question has exactly both scenarios.
There are situations where InnoDB deadlock would come up when you least expect it. For example, SELECT queries can perform locks on the gen_clust_index, aka the Clustered Index.
Here are three past questions I agressively looked over with @RedBlueThing, the person who asked these questions. @RedBlueThing found work arounds for his questions.
Just to keep your question in perspective, when you look over these answers (don't look too deeply, even I get dizzy looking at my own convoluted answers) it should be quickly apparent that even SELECT queries can lock InnoDB data. Although autocommit could be enabled (rendering each query its own transaction), a single SQL statement can still be victimized by deadlocks.
You also have special cases of SELECT where you can lock specific rows on demand.
Based on InnoDB Deadlocking link, the sequences of event to cause this situation could theoretically be as follows:
- Your SQL UPDATEs a single row but generates an error
- The UPDATE causes a rollback of the one row
- The row has a lingering lock
Personally, that last statement scares me. It would have been nice for MySQL to inform everyone of this quirk instead of just documenting and hopong you google for it. Yet, that statement is from the MySQL Documentation. (Oh yeah, Oracle owns InnoDB)
Best Answer
What language is the client written in? You should test for errors after every statement (including the COMMIT) in every transaction. That is the way to locate the problem.
SHOW ENGINE InnoDB STATUSshould not, as you see, be depended upon. More significantly, it won't necessarily catch all deadlocks.