Bug #58750 | 'out of connection objects' due to missing close() of prev. full_table_scan() | ||
---|---|---|---|
Submitted: | 6 Dec 2010 9:59 | Modified: | 25 Jan 2011 10:42 |
Reporter: | Ole John Aske | Email Updates: | |
Status: | Closed | Impact on me: | |
Category: | MySQL Cluster: Cluster (NDB) storage engine | Severity: | S2 (Serious) |
Version: | mysql-5.1-telco-7.1 | OS: | Any |
Assigned to: | Ole John Aske | CPU Architecture: | Any |
Tags: | 5.1.51-ndb-7.1.9 |
[6 Dec 2010 9:59]
Ole John Aske
[6 Dec 2010 11:36]
Bugs System
A patch for this bug has been committed. After review, it may be pushed to the relevant source trees for release in the next version. You can access the patch from: http://lists.mysql.com/commits/126120 4015 Ole John Aske 2010-12-06 Fix for bug#58750, for review. If ha_ndbcluster::read_range_first_to_buf() had to do a ::full_table_scan() on a UNIQUE_INDEX, the resulting ScanOperation was not properly closed at end of its lifetime. If lots of such operations is executed within the same transaction, we will eventually run out of operation or transaction objects. (Transaction objects is due to a dumnmy Hupp'ed transaction object is allocated as part of a ScanOperation) This fix ensures that ::close_scan() is called for any open cursors at start of ha_ndbcluster::read_range_first_to_buf().
[6 Dec 2010 11:38]
Bugs System
A patch for this bug has been committed. After review, it may be pushed to the relevant source trees for release in the next version. You can access the patch from: http://lists.mysql.com/commits/126122 3388 Ole John Aske 2010-12-06 SPJ-scan-scan: Cherry picked fix for bug#58750. If ha_ndbcluster::read_range_first_to_buf() had to do a ::full_table_scan() on a UNIQUE_INDEX, the resulting ScanOperation was not properly closed at end of its lifetime. If lots of such operations is executed within the same transaction, we will eventually run out of operation or transaction objects. (Transaction objects is due to a dumnmy Hupp'ed transaction object is allocated as part of a ScanOperation) This fix ensures that ::close_scan() is called for any open cursors at start of ha_ndbcluster::read_range_first_to_buf().
[6 Dec 2010 11:39]
Ole John Aske
Reset status to 'Open' as bug still has to be verified
[17 Jan 2011 9:57]
Ole John Aske
Set Status to 'Patch pending' as there is a patch awaiting review for this bug.
[17 Jan 2011 12:33]
Ole John Aske
Patch pushed to mysql-5.1-telco-7.0 ++
[25 Jan 2011 8:30]
Jonas Oreland
pushed to 7.0.21 and 7.1.10
[25 Jan 2011 10:42]
Jon Stephens
Documented as follows in the NDB 7-0-21 and 7.1.10 changelogs: When executing a full table scan caused by a WHERE condition using unique_key IS NULL in combination with a join, NDB failed to close the scan. Closed.