ERROR 1205 (HY000): Lock wait timeout exceeded on update mysql

Question

I am running following update -

update table_x set name= 'xyz' where id = 121;

and getting - ERROR 1205 (HY000): Lock wait timeout exceeded; try restarting transaction ERROR 1205 (HY000): Lock wait timeout exceeded; try restarting transaction

I googled it number of times and adding extra time to innodb_lock_wait_timeout not helping me out.

Please let me know the root cause of this issue and how I can solve it. I am using mysql 5.6(master-master replication) on dedicated server. Also table_x(Innodb table) heavily used in database. Autocommit is on.

Answer 1

Looks like there is some lock on any of your other transaction. You can check the status of INNODB by using this:

SHOW ENGINE INNODB STATUS\G 

Check if there is any lock on the tables like this:

show open tables where in_use>0;

And then kill that processes which are locked.

Answer 2

Find out what other statement is running at the same time as this UPDATE . It sounds as if it is running a long time and hanging onto the rows that this UPDATE needs. Meanwhile this statement is waiting.

One way to see it is to do SHOW FULL PROCESSLIST; while the UPDATE is hung.

(In my opinion, the default of 50 seconds for innodb_lock_wait_timeout is much to high. Raising the value only aggravates the situation.)

If you give up on fixing the 'root cause' of the conflict, then you might tackle the issue a different way.

1. Lower innodb_lock_wait_timeout to, say, 5.
2. Programmatically catch the error when it times out and restart the UPDATE .
3. Do likewise for all other transactions. Other queries may also be piling up; restarting some may "uncork" the problem.

SHOW VARIABLES LIKE 'tx_isolation'; -- There may be a better setting for it, especially if a long-running SELECT is the villain.

Answer 3

I have solved the problem. I tried different values for innodb_lock_wait_timeout , also tried to change queries but got the same error. I did some research and asked my colleagues about hibernate.

They were doing numbers of transaction which include updating main table and committing in the end. So, I suggested them to use commit on each transaction. Finally I am not getting any lock wait time out errors.