Getting the Network errors

Question

Hello Experts,
One of my customer has the following issue:
Getting the Network errors when the following batch script is executed through a scheduler or through a command prompt.
login "$1" "$2" on "$3";
alter system kill request on application SEMPCNH;
alter system logout session on application SEMPCNH force;
alter system unload application SEMPCNH;
alter system load application SEMPCNH;
drop lock on database SEMPCNH.SEMPCNH;
alter database SEMPCNH.SEMPCNH unlock all objects;
alter application SEMPCNH disable connects;
logout;
exit;
Log file:
Thu 09/10/2009 10:38:16.69 BEGIN Disable_Connects.bat S1CN1HAVAS11
Essbase MaxL Shell - Release 9.3.1 (ESB9.3.1.3.0B017)
Copyright (c) 2000, 2007, Oracle and/or its affiliates.
All rights reserved.
MAXL> login "essadmin" "password" on "146.47.243.32";
OK/INFO - 1051034 - Logging in user [essadmin].
OK/INFO - 1051035 - Last login on Thursday, September 10, 2009 10:38:16 AM.
OK/INFO - 1241001 - Logged in to Essbase.
MAXL> alter system kill request on application SEMPCNH;
OK/INFO - 1056090 - System altered.
MAXL> alter system logout session on application SEMPCNH force;
OK/INFO - 1013220 - Supervisor [essadmin] has forced user [NL034] to logout.
OK/INFO - 1051037 - Logging out user [NL034], active for 50 minutes.
OK/INFO - 1013220 - Supervisor [essadmin] has forced user [NL034] to logout.
OK/INFO - 1051037 - Logging out user [NL034], active for 199 minutes.
OK/INFO - 1013220 - Supervisor [essadmin] has forced user [NL034] to logout.
OK/INFO - 1051037 - Logging out user [NL034], active for 214 minutes.
OK/INFO - 1056092 - Sessions logged out [3].
OK/INFO - 1056090 - System altered.
MAXL> alter system unload application SEMPCNH;
ERROR - 1042013 - Network error [10054]: Cannot Receive Data.
ERROR - 1042012 - Network error [10054]: Cannot Send Data.
MAXL> alter system load application SEMPCNH;
ERROR - 1042006 - Network error [10061]: Unable to connect to [146.47.243.32:1423]. The client timed out waiting to connect to Analytic Services Agent using TCP/IP. Check your network connections. Also make sure that server and port values are correct.
SESSION - 1241102 - Internal API error 1042006.
MAXL> drop lock on database SEMPCNH.SEMPCNH;
essmsh error: Invalid MaxL Session ID
MAXL> alter database SEMPCNH.SEMPCNH unlock all objects;
essmsh error: Invalid MaxL Session ID
MAXL> alter application SEMPCNH disable connects;
essmsh error: Invalid MaxL Session ID
MAXL> logout;
User essadmin is logged out
MaxL Shell completed
Suggestions / Action Plan:
a) Modified the script by adding sleep command as below:
login "$1" "$2" on "$3";
alter system kill request on application SEMPCNH;
alter system logout session on application SEMPCNH force;
alter system unload application SEMPCNH;
shell 'ping -n 30 localhost > nul';
alter system load application SEMPCNH;
shell 'ping -n 10 localhost > nul';
drop lock on database SEMPCNH.SEMPCNH;
alter database SEMPCNH.SEMPCNH unlock all objects;
alter application SEMPCNH disable connects;
logout;
exit;
b) Modified the essbase.cfg file by adding the following settings -
SERVERPORTBEGIN 42768
SERVERPORTEND 43768
NETDELAY 2000
NETRETRYCOUNT 2500
NETTCPCONNECTRETRYCOUNT 50
SERVERTHREADS 40
AGENTTHREADS 40
SAMPLINGNUMBLOCKS 20
c) Edited Registry by adding the following -
Add new DWORD Value named TcpTimedWaitDelay - Value = 30
Add new DWORD Value named MaxUserPort - Value = 65534
Add new DWORD Value named MaxFreeTcbs - Value = 6250
But found the issue still persists. Any suggestions would be of great help.
Thanks in advance.
Regards,
Sudhir
Edited by: 637223 on Dec 23, 2009 2:30 PM

Answer 1

So it's the unloading from memory of the database that causes the problem, yes?
Have you tried:
1) Unloading other databases -- just interactively, without the kill and logout commands? Do they error?
2) Getting on the system at a quiet time and trying to unload the app/db?
3) Trying to stop a different database after the kill and logout?
4) Running a validate (whatever it's called in MaxL) on the db in question?
My guess is that it's either a corrupt db or a corrupt essbase.sec (yeah, I know, not overly helpful) -- the above steps should help you determine which is the bad actor.
Keep the board posted.
Good luck,
Cameron Lackpour