Bug #68437 | Result set return previous result after one time out exception | ||
---|---|---|---|
Submitted: | 20 Feb 2013 2:18 | Modified: | 27 Feb 2014 7:05 |
Reporter: | Myra Wei | Email Updates: | |
Status: | Can't repeat | Impact on me: | |
Category: | Connector / J | Severity: | S2 (Serious) |
Version: | 3.1.14 | OS: | Linux (2.6.18-128.el5) |
Assigned to: | Alexander Soklakov | CPU Architecture: | Any |
[20 Feb 2013 2:18]
Myra Wei
[20 Feb 2013 6:41]
Myra Wei
I redo the test on mysql 5.5.23, and will got the same result. And from my testing, if one read operation timed out, the connection will be closed automatically, there will be no such issue. But if one write operation timed out, the connection will not be closed automatically, we can find the issue happen. And if the lock is released before the same connection used for second query, there will be no issue. Only during the second query on same connection the lock is released, this issue happen.
[23 Feb 2013 10:05]
Sveta Smirnova
Thank you for the report. Do you free result after first timeout? If not, this is not a bug.
[25 Feb 2013 0:47]
Myra Wei
Thanks a lot for your reply. The first operation is one insert operation; we did close the statement. and we also close the resultset and statement on each subsequent read operation.
[25 Feb 2013 1:28]
Myra Wei
And please note that, the "flush tables with read lock" was executed from local mysql before the testing code run, and the "unlock tables" was executed after "showActiveTables------------"
[4 Mar 2013 0:29]
Myra Wei
Hi, do you have any further comments about this issue? Thanks a lot! Myra
[4 Mar 2013 18:28]
Sveta Smirnova
Thank you for the feedback. I can repeat error "ResultSet is from UPDATE. No Data." with version 3.1.14, but can not repeat wrong results of other queries. I also can not repeat this bug with current version 5.1.22. I still mark this report as "Verified", but I strongly recommend you to upgrade to version 5.1.22, because 3.1 series are obsolete.
[5 Mar 2013 0:49]
Myra Wei
Thanks a lot for your reply. I downloaded the 5.1.23 Connector, and there is no such issue. With the 3.1.12 version, about the wrong query result, it only occurs when the latter query is similar to the previous query. Like in my test case: all queries are "show table status like ", and we only retrieve the table name from result set. If the later query changed to "show variables", and we tried to retrieve the variable name, there will be no issue.
[27 Feb 2014 7:05]
Alexander Soklakov
Thank you Myra, Sveta! I close this report as "Can't repeat" because the error reported doesn't appear with c/J 5.1 we currently maintain, and we don't support c/J 3.1 version.