LDAP fails to respond to requests
ndstrace is unresponsive or slow
imonitor is slow
ndsrepair -E or -T is slow to respond
If ndsd is restarted, eDirectory functions normally again for a while.
Problem occurs more frequent under higher load
gcore of ndsd process shows many threads in the following state:
A look at where for the threads will show:
#0 0xb746a15b in sem_wait@GLIBC_2.0 () from /lib/tls/libpthread.so.0
#1 0xb5a76d00 in NMLogIDs () from /usr/lib//nds-modules/libauditds.so
#2 0xb597fee6 in LogEventExt () from /usr/sbin/../lib/liblogevent.so
#3 0xb5a60f79 in EventHandler () from /usr/lib//nds-modules/libauditds.so
"ndstrace -c threads" typically shows threads are available in the ndsd pool for processing
top does not indicate ndsd is consuming a large amount memory or CPU time.
If Nsure Audit is not loaded, the ndsd does not become unresponsive under heavy load.
The documentation indicates this option can cause a slowness in eDirectory.
Please do the following steps to see if you have the eDirectory instrumentation configured in inline mode or not.
4.) In the "Global" section near the top, there is a setting"Register For Events Inline". By default this is unchecked. If there is a check next to it, uncheck it and save your settings.
When the eDirectory instrumentation is set to inline mode, events are processed serially. eDirectory will not be able to process the next eDirectory event/request until Novell Audit has committed the data to the SLS. If you have a server that is running under heavy eDirectory load, it is not recommended that you register for events inline.