We have a mysql-5.1.34 ndb-7.0.6 5 node cluster installed in a data center
running mission critical applications. The cluster is installed on windows
2003 64 bit servers.
It has been noticed that the handle count for the mysqld process nodes in
the two SQL nodes increases gradually (around 4 to 5 handles every 10 secs)
and within 10 days reaches very high numbers like 2,00000 and more. Only
when we restart the mysqld service does the handle count come down.
Is this a bug with this version of mysql cluster? How do we keep the handle
count in check? Please help.