|
|||||||||||
|
Can anyone help me with a "Lock wait timeout exceeded" problem?
From: Bond Masuda <bond.masuda(at)jlbond.com>
Date: Thu Jul 19 2007 - 21:15:49 EDT
I have a 2 data node, 3 api node, 2 mgm node MySQL cluster. We're using 5.0.26 in production and 5.0.45 in our lab (both community editions). The cluster has been working great for a good 8 months until about 2 months ago, our application started seeing "Lock wait timeout exceeded" errors and the SQL would fail. These were nightly data imports, so lots of INSERT and UPDATE queries. Strange thing is, none of the cluster logs, data node logs, api node logs record any problems. Also, if I repeat the "data import" process several times from the same state, sometimes it fails with above error, and sometimes it succeeds. I'm having a difficult time narrowing down the conditions where we can reproduce the problem. After searching the internet, I've seen several others post with this same error message, but mostly got no replies. Based on the information I found via google, I tried increasing the TransactionDeadlockDetectionTimeout from the default 1200 to 3000, and then to 5000. Those changes made no difference in the occurance of this problem. I know 5.0.26 is pretty old now, so in our lab, I'm using 5.0.45 and experiencing the same problem. Does anyone have any suggestions on how I can obtain more data to find the root cause? Or, has anyone experienced this type of problem and have some suggestions to fix it?
I would appreciate any help if possible.
-- This archive was generated by hypermail 2.1.8 : Thu Aug 09 2007 - 19:30:32 EDT |
||||||||||
|
|||||||||||