Max-stack-size - default_stksize

Question

Hi,
First, sorry for my english ^^
I'm new on Solaris. I installed Solaris 10 on Sun V490 . I use Core Network.
Why with the default setting, daemons or process of the OS "don't work" like this:
Jun 17 14:50:10 unknown genunix: [ID 883052 kern.notice] basic rctl process.max-stack-size (value 8683520) exceeded by process 353In this example I generate the error with the format command. But I get this error with other command or deamon like nscd.
I had the same problem with the value max-file-descriptor. The values set by projmod for the system project did not seem take effect. Thus I used the "old" parameters rlim_fd_cur, rlim_fd_max. Now it's ok.
I find this parameter default_stksize in the Sun documentation. I put this in my /etc/system file:
set default_stksize=16384At the boot time I have no error message for the value, but the max-stack-size value is the same:
prctl -n process.max-stack-size 130
process: 130: /usr/sbin/nscd
NAME PRIVILEGE VALUE FLAG ACTION RECIPIENT
process.max-stack-size
basic 8,28MB - deny 130
privileged 127MB - deny -
system 2,00GB max deny -
nscd is in the system project:
ps -p 130 -o project
PROJECT
system
Thank in advance, any idea is welcome.
Guillaume

Answer 1

Hi Prasad,
Block Size is the number of parallel processes that are being executed in background during the application.  This is normally a configuration activity to be configured in line with basis.
In Block size, we enter the number of objects to be processed per block by the CIF comparison/reconciliation during data selection in SAP APO or in the partner system.
If you increase the block size, the memory required also increases. This has a positive effect on performance. If processes are cancelled due to lack of memory, you can decrease the block size to save memory.
If you do not enter anything here, the system determines the block size dynamically from the number of objects that actually exist and the maximum number of work processes available.
Normally, when you execute a job in background, it picks up the application server automatically or by manually defined server.  In parallel processing, at a time, one or more job of the same identify can be triggered under this scenario by defining application servers.  But too many parallel processing activity will affect the performance.
One needs to define the parallel processes also to control system behaviour.  The Parallel processing profile is defined for parallel processing of background jobs. You then assign these profiles to variants in the applications.
Regards
R. Senthil Mareeswaran.