We have a four node cluster NW 6.5 SP6 with iPrint/NDPS installed as a cluster ressource. We get many
APACHE_Worker_Process ABEND in APACHE2.NLM on the node running iPrint. Without ippsrvr.nlm ndps works without abends. When we reload ippsrvr.nlm the server abends after a few days.

Rebuilding the Managerdatabase (using PaGen) didn't help.

Since the abends are happening at mid day (12:00, 13:00) and in the evening around 5 pm we assume it could be a problem with heavy usage (the students are busy when the lessons end). Or perhaps a client workstation which behaves badly.
There are 197 PAs managed by one NDPSManager.

We already tried to tune the apache.nlm by adding possible threads:
ThreadStackSize 65536
StartThreads 250
MinSpareThreads 25
MaxSpareThreads 250
MaxThreads 1024
MaxRequestsPerChild 0
MaxMemFree 100

Last ABEND LOG:

Server 1 halted Wednesday, 20 February 2008 13.13.29,837
Abend 1 on P00: Server-5.70.07-1937: CPU Hog Detected by Timer

Registers:
CS = 0008 DS = 0023 ES = 0023 FS = 0023 GS = 0023 SS = 0010
EAX = FBF17BA3 EBX = 4ABAFA30 ECX = 00000007 EDX = 426A4800
ESI = 4ABB1811 EDI = 426A4875 EBP = 4ABB1810 ESP = 44F43D30
EIP = 00000000 FLAGS = 00000097


Running process: Apache_Worker 175 Process
Thread Owned by NLM: APACHE2.NLM
Stack pointer: 44F43DA0
OS Stack limit: 44F34F40
Scheduling priority: 67371008
Wait state: 5050030 Blocked on Semaphore
Stack: --FBF17BA3 ?
42684B1E (APACHE2.NLM|ssl_var_lookup+A5E)
42669BC9 (APACHE2.NLM|APACHE2@ap_get_status_line+179)
--61502D58 ?
--4ABB1815 ?
--64696F76 ?
--4ABAFA30 ?
--4594A958 ?
--4ABB1748 ?
--44505FE0 ?
42695D18 (APACHE2.NLM|APACHE2@ap_pass_brigade+58)
--4ABB1588 ?
--4ABB1748 ?
--4ABB06E0 ?
4266A7D0 (APACHE2.NLM|APACHE2@ap_http_header_filter+390)
--4ABB1588 ?
--4ABB1748 ?
--434F4E54 ?
--00000000 (LOADER.NLM|KernelAddressSpace+0)
--459475B8 ?
-4269CB18 (APACHE2.NLM|http_module+870)
--4ABAF9F8 ?
--4ABB1748 ?
-4214639C (APRLIB.NLM|apr_bucket_type_eos+0)
---snip----

Thanks in advance

Martina Schacht
University of Hamburg, Germany