oes2sp1 - Ndsd using up all threads
We have a root server that holds a replica of every partition in our tree, which is needed to run dirxml syncs to a flat tree we have that is an authoritative source for identity management in our environment.
The root server cannot keep up with thread need / usage, when it cannot keep up anymore it locks up and effects services from enterprise logins to repkication across the entire tree. Users cannot login around the enterprise when it happens, replica partition syncs stop, very, very, very ugly.
The thread count eventually gets to over 320 and the server ndsd needs to be restarted.
To get around this we are restarting ndsd with a script, using a trigger of if used threads get close to 320.
It is a OES2SP1SLES10SP2 64 bit, updated patches being managed by zlm.
Latest version of edir also: 8.8.5
Anyone else seeing this? Any ideas?
Thanks
|
Threaded Mode
